corporate social capital and liability

Corporate Social Capital and Liability

Corporate Social Capital and Liability

Edited by

Roger Th.A.J. Leenders School of Management and OrganiZOlion University ojGroningen, The Netherlands

Shaul M. Gabbay Davidson Faculty of Industrial Engineering and Management Technicn - Israellflsli/utl! o!Tedllw{ogy

.... . , Springer Science+Business Media, LLC

... " Electronic Services <http://www.wkap.ol>

Library of Congress Cataloging-in-Publication Data

Corporate social capital and liability I edited by Roger Th. AJ. Leenders, Shaul M. Gabbay.

p. em. Includes index.

ISBN 978-1-4613-7284-4 ISBN 978-1-4615-5027-3 (eBook)DOI 10.1007/978-1-4615-5027-3

1. Business networks. 2. Social nelworks. 3. Industrial organizatian. 1. Leenders, Roger Th. A. J. II. Gabbay, Shaul M.

HD69.S8 C67 1999 302.3'5 - - dc21

99-28470 CIP

Copyright@I999bySpringerSciCIlCl'+RusinessMediaNewYork

Originali)' published by Kluwer Academic Publishers 1999 Softcover reprint of the hardcover 1 st edition 1999

AII rights reserved. No part of this publicatian may be reproduced, stored in a retrieval system or transmitted in any form Of by any means, mechanical, photocopying, recording, or otherwise, without the prior written permission of the publisher, Kluwer Academic Publishers, 101 Philip Drive, Assinippi Park, Norwell, Massachusens 02061

Printed an acid-free paper.

Contents

• INTRODUCTION

CSC: The Structure of Advantage and Disadvantage Shaul M. Gabbay & Roger Th.AJ. Leenders

Section I CONCEPTUAL ISSUES - Theory, Models, and Measurement

Organizational Networks and Corporate Social Capital 17 David Knoke

2 Social Capital of Organization: Conceptualization. Level of Analysis. and Performance Implications 43 Johannes M. Pennings & Kyungmook Lee

3 A Relational Resource Perspective on Social Capital 68 Luis Araujo & Geoff Easton

4 Social Capital by Design: Structures. Strategies. and Institutional Context 88 Wayne E. Baker & David Obstfeld

5 Corporate Social Capital and Liability: a Conditional Approach 106 to Three Consequenses of Corporate Social Structure lIan Talmud

6 Dimensions of Corporate Social Capital: Toward Models and Measures 118 Shin-Kap Han & Ronald L. Breiger

7 Organizational Standing as Corporate Social Capital 134 Patrick Doreian

8 Customer Service Dyads: Diagnosing Emperical Buyer - Seller Interactions along Gaming Profiles in a Dyadic Parametric Space 148 Dawn Iacobucci

Section II STRUCTURE AT THE INDIVIDUAL LEVEL - Social Capital in Jobs and Careers 9 The Sidekick Effect: Mentoring Relationships and the Development of

Social Capital 161 Monica Higgins & Nitin Nohria

10 Social Capital in Internal Staffing Practices 180 Peter V. Marsden & Elizabeth H. Gorman

11 Getting a Job as a Manager 197 Henk Flap & Ed Boxman

12 The Changing Value of Social Capital in an Expanding Social System: Lawyers in the Chicago Bar. 1975 and 1995 217 Rebecca L. Sandefur, Edward O. Laumann & John P. Heinz

vi - Corporate Social Capital and Liability

Section III STRUCTURE AT THE INDIVIDUAL LEVEL - Social Capital and Management 13 Generalized Exchange and Economic Performance: Social Embeddedness of

Labor Contracts in a Corporate Law Partnership 237 Emmanuel Lazega

14 CEO Demographics and Acquisitions: Network Effects of Educational and Functional Background 266 Pamela R. Haunscbild, Andrew D. Henderson & Alison Davis-Blake

15 Public Service Organizations - Social Networks and Social Capital 284 Ewan Ferlie

16 The Dark Side of Social Capital 298 Martin Gargiulo & Mario Benassi

17 Social Capital, Social Liabilities, and Social Resources Management 323 Daniel J. Brass & Giuseppe Labianca

Section IV STRUCTURE AT THE FIRM LEVEL - Social Capital in Collaboration and

Alliances 18 The Triangle: Roles of the Go-Between 341

Bart Nooteboom 19 The Management of Social Capital in R&D Collaboration 356

Onno Omta & Wouter van Rossum 20 Technological Prestige and the Accumulation of Alliance Capital 376

Toby E. Stuart 21 Networks and Knowledge Production: Collaboration and Patenting in

Biotechnology 390 Laurel Smitb-Doerr, Jason Owen-Smith, Kenneth W. Koput & Walter W. Powell

22 Supply Network Strategy and Social Capital 409 Christine Harland

Section V STRUCTURE AT THE FIRM LEVEL - Social Capital and Financial Capital 23 Choosing Ties from the Inside of a Prism: Egocentric Uncertainty and

Status in Venture Capital Markets 431 Joel M. Podolny & Fabrizio Castellucci

24 Corporate Social Capital and the Cost of Financial Capital: An Embeddedness Approach 446 Brian Uzzi & James J. Gillespie

25 Venture Capital as an Economy of Time 460 John Freeman

CONCLUSION CSC: An Agenda for the Future Roger Tb.A.J. Leenders & Sbaul M. Gabbay

REFERENCES INDEX CONTRIBUTORS EDITORS

483

495 545 559 563

• Introduction

•

CSC: The Structure of Advantage and Disadvantage

INTRODUCTION

Shaul M. Gabbay Roger Th.A.J. Leenders

Scholars of the firm have long concerned themselves with identifying the differential characteristics that make some firms-and some of their membersmore successful than others. A recent approach to the study of success and failure in the competitive marketplace is the theory of social capital. The theory of social capital suggests that players gain access to various kinds of resources that accrue to them by virtue of their engagement in various kinds of relationships. Social capital theory is fundamentally concerned with the resources inherent within structures and social exchange. Until now, social capital theory has mainly been applied to individual actors-human beings. In this volume the central question is how is social structure related to the attainment of goals of corporations and their members (denoted below by the terms 'corporate players' or 'corporate actors')? We suggest that Corporate Social Capital refers to the resources, inherent in the social structure, that accrue to corporate actors. Social structure refers to a network of actors who are in some way connected via a set of relationships.

In the current chapter, we will briefly discuss the concept of social capital in the context of organizations and introduce the concept corporate social capital. Subsequently, we will discuss the relationship and distinction between social structure and social capital. In so doing, we will introduce the notion of corporate social liability and focus our attention on the various levels of aggregation at which social structure, social capital, and social liability reside. We will coin the acronym CSC to denote the interplay of social structure and social capital/social liability , both at the firm, intermediate, and individual levels. In the concluding chapter of this volume, we will discuss the research and practical applications of further developments. We will highlight critical questions that, in our view, should be

2 - Corporate Social Capital and Liability

resolved. In our final discussion, we will open new questions for future discussions. At its initial stages, CSC is an emerging research agenda. This, of course, presents a wide window of opportunity for future and further contributions.

SOCIAL CAPITAL AND ORGANIZATIONS

The term 'social capital' first appeared in scientific literature around a century ago,' but it was not until the mid-1980's that a theory of social capital was developed and applied in the context of work and organizations. James Coleman was the first to develop a comprehensive theory of social capital and his work inspired the diffusion of the use of the theory in relation to the study of actors who are pursuing interest driven goals. Coleman's seminal work-Foundations of Social Theory (1990) sparked a host of studies applying social capital theory.2 The work of Burt (1992) was explicit in its focus on corporate actors. In his study of managerial mobility in a high technology firm, Burt was the first to introduce a quantitative measure for social capital. Studying White et al.'s (1976) suggestion that the absence of ties may provide an advantage, Burt (1992) showed that actors who are connected to disconnected others (structural holes) advance faster in the corporation under study.

These studies explicitly used the term social capital. Many social network studies, that did not explicitly employ the term 'social capital' have also proven important in the development of social capital theory. For example, the work of Granovetter (1973), The Strength of Weak Ties, greatly influenced the study of social capital as did the work of Lin (1982; Lin et al. 1981) and the work of Laumann (1973). The theory of social capital has gained a prominent place in a wide range of scientific fields, spanning social, economic, and political research agendas. For a discussion of the history and use of the theory of social capital we refer to Portes and Sensenbrenner (1993), Gabbay (1995, 1997), Nahapiet and Ghoshal (1998), and Woolcock (1998).

CORPORATE SOCIAL CAPITAL

Although a large body of research has emerged on social capital, and a growing group of researchers is now using social capital in their research, consensus on the definition of social capital has yet to be established. Some authors equate social capital with social structure, whereas others refer to the resources an actor can mobilize through the social structure. In the current volume, we specifically refer to the assets embedded within and available through networks of relationships. We define social capital as the set of resources, tangible or virtual, that accrue to an actor through the actor's social relationships, facilitating the attainment of goals.

Before moving on to the discussion to 'corporate' social capital, it is important to discuss briefly three elements of our view of social capital. First, we view social capital as goal-specific. A large number of social ties does not necessarily translate itself into social capital. It only does so if these ties assist the actor in the attainment of particular goals. An example of this is provided by the following event. In February 1996, Ben Van Schaik, the CEO of Dutch aircraft builder Fokker, gave a presentation to a potential alliance partner. 'We are the second largest aircraft builder in the world,' he said, supporting his claim with a colorful bar graph-the bars

CSC: The Structure of Advantage and Disadvantage - 3

represented various aircraft manufacturers; their length represented the number of the manufacturer's clients. With this graph, Van Schaik showed the audience that Boeing/McDonnell Douglas had 846 clients, Fokker had 225, and the rest trailed behind at a large distance.3 Unfortunately, the number of ties did not represent the number of aircraft sold, nor did it represent the credit rating of the customers. Fokker's clients were primarily small companies, leasing only one or two airplanes (rather than buying them), and many of them 'forgot' to pay their bills. Fokker declared bankruptcy only a few months later and was liquidated. Its 'much smaller' competition is still alive and kicking.4 Social network and social capital are different things. A social network only conveys social capital if its social ties are beneficial for the attainment of goals. In the case of Fokker, they clearly were not.

Second, an actor need not be conscious of the social capital he enjoys. The social structure in which an actor is embedded may confer advantages to the actor, without the actor even realizing it.

Third, the social structure that brings opportunities for the realization of particular goals need not have been built in the pursuit of these goals-social capital often is a by-product of other social activity.

Like other forms of capital, social capital is productive, making possible the achievement of goals that would not be attainable in its absence. This goal specificity has a number of implications.6 First of all, social structure may be beneficial for the attainment of multiple goals, since the multiplex character of many social relationships results in overlap of opportunity structures. Networks created for one purpose may be employed for another-which often was not foreseen when the actors initially engaged in a relationship with each other. In some situations, the same social structure can be beneficial for the attainment of one goal, while obstructing the attainment of others.7 Social networks can have a positive effect in providing network members with access to privileged resources, while lowering transaction costs, but can, simultaneously, place high demands on these network members, restricting their individual behavior and opportunities.

In this volume, we concern ourselves with corporations and their members.s Corporate social capital, then, refers to:

The set of resources, tangible or virtual, that accrue to a corporate player through the player's social relationships, facilitating the attainment of goals.

Although social structure and social capital are often equated in the social capital literature, they are different entities. In this volume we therefore make an explicit distinction between social structure per se and the outcomes of social structure. When these outcomes are positive, helpful in attaining specific goals, we say the social structure conveys social capital. But when social structure prohibits and obstructs action, it produces social liability. Although the absence of social structure precludes social capital from corning into existence, the two are distinct. They are generated by related, but distinct, processes.

SOCIAL CAPITAL AND SOCIAL LIABILITY

Just as bank accounts can run dry and can have a negative balance, social capital can vanish. Social structure that provided social capital in the past may not necessarily


do so in the present. The social structure required to sustain corporate social capital may shift as transactions, activities, and conditions change and become more or less complex. Relationships beneficial to the achievement of goals in the past may thwart goal attainment in the future. In their study of how managers adapt to changing environmental conditions, Gargiulo and Benassi (this volume) find that relational structures that in the past had provided ample social capital for managers, later increased the number of coordination failures for which these managers were responsible. The network no longer provided resources, but had become a constraint, impeding performance.

Brass (1984) and Blau and Alba (1982) found that relationships to the clique of top executives in an organization had a strong positive relationship with power and promotions. Top executives were likely to have more social capital in the form of more (relevant) information to share with those who were connected to them. Since men are more likely to maintain relationships to top executives than are women,9 women may be forced to forgo any preference for homophily in order to build connections with the dominant coalition and share the social capital. Brass and Labianca (this volume) therefore conclude that actor dissimilarity (in this case based on gender) may affect interaction patterns and may consequently exclude some people from sharing in the social capital. From the viewpoint of human resources management, Brass and Labianca provide ample examples of the positive and negative effects social relationships yield on social resource management outcomes such as recruitment, socialization, turnover, job satisfaction, power, and conflict.

Leenders (1995) shows that the relations among social service workers may help in preventing these workers from becoming burned out, but that feelings of burnout are also contagious through these exact same relationships. Increases in burnout experienced by social service workers also increased the level of burnout experienced by their co-workers. Conversely, decreases assisted in decreasing burnout among alters.

Gabbay (1995, 1997) shows that, for some actors, strong ties combined with structural holes ('structural ports') were beneficial at the inception stage of their business, but were detrimental for future expansion. Successful entrepreneurs strategically changed their networks over time in order to maintain or build social structures that would be rich sources of social capital.

Social structure translates into social liability in at least two different ways. First, ongoing, strong social relationships may constrain the behavior of actors, impeding their action and attainment of goals. IO For example, long-standing relationships with customers may stifle the firm by monopolizing a disproportionate share of its resources, inhibiting the firm from forming relationships with alternative customers. Second, actors may be unfavorably affected in their opportunities by negative ties in the social structure. In this volume, Brass and Labianca explore the effects of this type of social capital. As an example of the social liability stemming from negative ties, they argue that 'it is likely that an actor's negative ties within an organization will prevent promotion, particularly if those negative relationships are with influential others. Others may withhold critical information that worsens an actor's performance or they may provide bad references in order to prevent a promotion' (this volume: 324).


Organizations and their members continuously engage in social relationships. Effective organization requires a constant balancing of the potentially opposing forces inherent in social structures.

LEVELS OF ANALYSIS

Relational structures can be recognized at different levels of analysis and observation. Four levels of analysis normally characterize organizations. \I The individual human being is the basic building block of organizations. The human being is to an organization as a cell is to a biological system. One step higher, we find group or departments, where individuals work together on group tasks. Next is the organization itself: a collection of groups or departments. Above this level, organizations themselves can be grouped into an interorganizational network. 12

Social capital and social structure are relevant at all of these different levels. Classic methodological approaches require a researcher to choose one particular level of analysis as the primary focus of a study. However, as we will discuss below, the very nature of social capital runs through all these organizational levels. Social structures at the individual level translate into social capital and social liability for individual actors, but also translate into social capital effects at higher levels of analysis. Similarly, the structure/capital connection also works its way down the levels of analysis. A full study of social capital should thus incorporate structure and capitallliability at mUltiple levels of analysis.

Relations Among Individuals Intraorganizational networks have been the locus of exuberant research efforts. The effects of network structure on, among many other things, the distribution of power, job satisfaction, career opportunities, and productivity of individuals have been central on the research agenda. Undoubtedly the most attention has been given to the sources and consequences of the distribution of power in organizations, focusing on such concepts as prestige, status, and control. Next on the agenda has been the related topic of information flows in organizations. This type of research addresses questions related to 'who has what information when and how'? Considering the fact that we are only at the start of the information age, and are pushed along by rapid developments in information and communication technology, this latter question is likely to remain among the most popular objects of study for a long time.

The distribution of power and information is both an outcome and generator of network structure. As an outcome, it portrays (part of) the social capital (and social liability) that employees extract from their relationships. As a generator, social capital affects intraorganizational structure. For instance, by the very nature of his work, a top executive has many relational ties, spread throughout the organization, and is at the intersection of numerous flows of information. This network position, characterized by high centrality and the presence of structural holes, provides the manager with ample opportunity for control, power, and influence: the manager draws social capital from his network structure. In turn, his power gives him the opportunity to surround himself with a network of loyal subordinates and use his network ties to stay informed about events throughout the organization. By placing trusted friends in strategic positions, a manager gains power by being well-informed,


by having access to other (important) people in the organization, and by having mUltiple people depend on him. Capitalizing on his social capital, the manager effectively reengineers his position in the network. Ideally, this will then allow him to harvest more social capital from his (new) network, or secure the capital drawn from the old structure-social capital creates social capital.

Social structures among individual members of the firm relate to social capital at various levels of analysis. First, intraorganizational networks yield social capital for the individuals comprising these networks as individuals employ their direct and indirect ties to fulfill individual goals. Occupying a central position in various networks of relationships is often regarded as a source of power--power is social capital to those who can utilize it for getting the job done. Having a tie with a mentor can be both helpful and detrimental in securing jobs (see the chapter by Higgins and Nohria in this volume). Social ties may help individuals to find (more favorable) jobs both in the framework of internal market opportunities (Marsden and Gorman, this volume) and in the context of external labor markets (Flap and Boxman, this volume). In particular, having access to higher-status individuals is often argued to be beneficial in the job search process.

Second, social structures at the level of individuals also impact social capital at higher levels of analysis. For instance, Kratzer, Van Engelen, and Leenders (1998) discuss how the structure of various types of relationships among members of R&D teams affects their success. They contend that segmentation of the problem-solving network among the team members is detrimental to the performance of the team, an effect stronger for teams performing highly complex tasks than for teams performing tasks of low complexity. R&D teams that agree on their basic product development goals-but are characterized by a reasonable disagreement on how to achieve these goals-tend to develop products with a much higher probability of market success than teams whose members fully agree (or disagree) on these issues.13 Based on the studies focusing on social contagion,14 it is possible to conclude that the structure of social relationships between individuals affects their level of consensus and, consequently, their group performance. The structure of these relationships thus affects outcomes at the firm level.

Social structures at the individual level of observation can also have negative effects at these higher levels of analysis. For instance, consider the so-called grapevine-an informal, person-to-person communication network of employees that is not officially sanctioned by the organization. Grapevines are sources of rumors and gossip that spread quickly throughout an organization. Often, management decisions circulate through grapevines days ahead of their official announcement. 15 Because they feel threatened by it, managers often try to suppress the grapevine, but find themselves confronted with a nearly impossible exercise. Another example is related to the resilience of personal networks. Managers in charge of (re)designing business processes often experience difficulties in breaking through the power structures that exist between the firm's employees. 16 As Knoke (this volume: 40) notes, 'power structures are highly stable and resistant to change.'

In addition to relationships with others within a particular firm, individuals also maintain relationships with those outside of the firm. The relationship a lawyer maintains with her clients often directly translates into status, financial turnover, and


profitability at the law firm level. In effect, the firm draws social capital from its employee's contacts (who also draws social capital from the relationship herself). Interlocking directorates connect firms-they lead to information sharing and policy binding and can be a strong source of organizational power.17 In the chapters by Knoke and by Pennings and Lee, the boundaries differentiating roles of individuals are discussed extensively.

Relations Among Organizations No organization really stands on its own. Almost every organization needs suppliers and buyers, and deals with intermediaries. Ongoing exchange relations with other organizations translate into more or less stable interorganizational networks.

Drawing from theories as diverse as contingency theory, resource dependence theory, transaction cost theory, and population ecology, scholars are collecting data and devising explanations of interorganizational relationships. Unfortunately, however, these theories hardly devote any attention to how these interorganizational relationships lead to a network of relationships. 18 They also ignore how these affect opportunities and restrictions for the firms comprising these networks.

Recent trends influence the extent and intensity of the engagement in interorganizational relationships. The unification of Europe has led to a surge of mergers, alliances, cooperation, and licensing. More than ever before, European companies are looking at the structures around them, actively investing in new and reassessing old ties. Advantages are not necessarily in the creation of a large bundle of ties, but in the creation and maintenance of relations that provide access to new technologies, resources, and legitimacy.19 In other words, companies are actively molding their networks so as to draw as much corporate social capital from them as possible, with a long-term focus.

Europe is not the only stage where interorganizational relationships are crucial. With strong trends toward globalization in virtually every sector, and with rapid developments in information and communication technologies, it is possible and necessary for many firms to enter into intercorporate relationships.

Just as individuals can tap social capital from their social networks in order to facilitate some action or attainment of goals, so too can organizations extract resources from their networks. Harland (this volume) shows how corporate social capital is created through the buildup of strategic interorganizational supply networks. Using the cases of Benetton and Toyota she illustrates how firms, that are seeking to increase competitive advantage, actively pursue the role of a 'supply network hub' to facilitate effective and efficient flows of materials and information. Smith-Doerr et al. (this volume) examine the link between social structure and intellectual output in the context of the biotechnology industry, in which interorganizational collaboration is commonplace. Their interest is in the number of patents organizations obtain as a result of their collaborative relationships. For instance, they argue that firms that are more central in their collaboration network will attain more patents in a timely manner, due to their deeper insights into the knowledge-base of the field and their ability to combine and utilize a broad configuration of partners' expertise in choosing promising lines of research and framing patentable claims (this volume: 394). Not only does a favorable position in


the network (as measured by centrality) lead to the generation of more patents, but receiving more patents, in turn, also leads to an even more favorable network position. In other words, social capital is both an outcome and a generator of social structure.

Stuart (this volume) also studies the social capital that can be gained from alliances. In concert with Smith-Doerr et al. (this volume), patents playa central role in his study. Stuart's interest is in whether the status of an organization (as a function of the organization's position in the patent citation network) has an effect on the rate at which the organization acquires technologies from other firms. Technologically prestigious firms are found to be able to access the technological assets of other firms at a higher rate than are firms of lesser prestige. Interestingly, the rate at which a firm acquires technology from its competitors is more strongly positively related to the firm's technological prestige than is the rate at which the firm supplies technology. In this sense, high-status firms gain a positive balance of social capital from alliances with lower-status partners by taking more technology than they give. Being at the other end of the relationship, the lower-status firm sees some of the status of its partner reflected on its own technology, and draws a different type of social capital from the same relationship.

The question of how status relates to social capital is also the focus of the chapter by Freeman (this volume). Freeman shows that start-up businesses that receive support from centrally connected venture capital firms (VCs) have an increased probability of an initial public offering. In Freeman's study, the structural position of one organization (the VC) produces social capital for another (the startup business). But status also brings a risk. As the analyses show, VC centrality also increases the probability of a startup business being acquired. Here, the structural position of the VC creates (potential) social liability for the entrepreneur. The VCs themselves also experience risks. Freeman suggests that more central, high-status VCs behave like option traders by making high-risk investments and walking away from the losers. This could undermine the deal flow advantages that centrality affords alter the central position of the venture capital firm.

Uzzi and Gillespie (this volume) are concerned with the question of how social structure affects the probability of a small business securing capital. In partiCUlar, they examine the relationship between social structure and lending practices. They show that small businesses garner more loans at lower interest rates by increasing the duration and multiplexity of their relationships with financial institutions. Their findings clearly show that by strategically engineering social structure and by creating a proper mix and intensity of ties to financial institutions, small businesses can gain much social capital (in the sense of securing capital at decent prices).

So far, we have only concerned ourselves with the relationship between interorganizational networks and social capital at the organization level. But these structures also yield social capital at the individual level. For example, alliances or consortia can provide employees with the opportunity to be trained or temporarily be located at other organizations. Employees can also more easily change employers, moving between firms within one alliance. With employability becoming increasingly important in the workplace, these interorganizational relationships

Structure firm level


Structure individual level

Figure 1. esc framework: Interdependencies between social structure and social capital at various levels of analysis

provide the individual employee with the social capital of learning new skills and the opportunity to display his skills beyond the boundaries of the · firm he is (currently) affiliated with. Obviously, this also creates potential social liability for the organization, as it may lose valuable employees. In this volume, Higgins and Nohria study career opportunities of employees moving between companies.

Mixtures of Levels Employees of every firm are involved in social relationships that extend beyond their own firms. Consequently, firms themselves are part of wider social and economic networks that can be expected to influence their relationships with potential suppliers and customers.

Many relationships between organizations are mediated by individuals. Lawyers bring their clients to their law firms. Formal relationships between firms often start out as informal, personal relationships between representatives of these firms and then become institutionalized. But not all interorganizational relationships are mediated by individuals. As Pennings and Lee (this volume: 50) note 'as a legal entity, the firm is capable of contracting, of acting as a partner in any market relationship, including the setting up of joint ventures, the acquisition of another firm, or the shedding of a business unit to other firms, etc .. '

An important aspect of the institutionalization of ties maintained by the firm's members is that it shifts the agency of the social capital to the firm level. A consultant, whose social contacts bring in large revenues for his firm, may try to keep the relationship 'to himself.' By retaining the relationship in his personal portfolio, he has a powerful means to strengthen his position in the firm. A manager whose contacts lead to a formal, contractually regulated alliance transfers his claims to the relationship (and the social capital inherent in it) to his organization.


An organizational risk attached to the personal nature of relationships is that these relationships may be lost when the employee leaves the firm. Consultants may take their contacts with them to their new employer.20 Organizations search for ways to appropriate individual-level social capital, often by developing a reward system that favors employees who bring in the most social capital. Salesmen often are rewarded in proportion to their achievements in terms of sales and revenues; successful consultants are offered the opportunity to become a partner in the firm.

Relational structure may have both positive and negative consequences for the attainment of goals of corporate players at either level of analysis (see Figure 1).21

Strong inter-firm relationships (firm level structure) can provide the firm (firm level social capital) with corporate social capital-they provide the firm with resources while lowering risks and costs of opportunism. At the same time, these ties may provide firm members (individual level social capital) with social capital by giving them access to a number of other firms to which they can relocate to further their individual careers. As a result, the firm is confronted with the corporate social liability of seeing its valued employees leave (firm level social liability). Relationships of individuals of firm members with others outside the firm (individual level structure) can provide the firm with valuable clients (firm level social capital). The ties among individuals in teams and departments (individual level structure) affect their job satisfaction (indidividual level social capital) and affects the productivity of the group (firm level social capital). As argued above, a full study of social capital should deal with the association between structure and capitalIliability at mUltiple levels of analysis.

CSC

To summarize, we have the following elements: • Social structure and social capital are related but distinct entities; • Social capital represents the resources that accrue to an actor through the actor's

social relationships, facilitating the attainment of goals. When social structure hinders the attainment of goals, it yields social liability;

• 'Corporate social capital' refers to the social capital of corporate players: firms and their members;

• Social structure and social capital can be distinguished at different levels of analysis, at minimum at the levels of the firm and the individual;

• Social structure and social capital are not only associated within the same level of analysis, but are outcomes and generators of each other at all of these levels;

• Social structure can provide social capital to one player, but social liability to another. It can provide social capital for the attainment of one goal, but social liability for another goal.

We have found it useful to use the acronym 'CSC' to denote the constellation of {structure, capital, liability, level of analysis}. In this volume, we will employ this acronym in the current chapter and in the concluding chapter.

CSC: The Structure of Advantage and Disadvantage - II

THE ADDED VALUE OF esc The study of social networks has been the focus of the work of organizational scholars for over a decade. A growing body of studies from various areas and disciplines has focused on the study of social networks in the context of organizations. However, Salancik (1995) recently noted that the wealth of social network studies and methodology has not yet produced an encompassing theory of organization. He therefore suggests that 'a network theory of organization should propose how structures of interactions enable coordinated interaction to achieve collective and individual interests.'22 Theories of corporate social capital present a (partial) answer to this call. CSC theory attempts to explain how social structure is connected to organizational outcomes. It explicitly relates to (the interplay of) structures at the individual level and the firm level, and to how these affect attainment of relevant organizational goals. This invokes questions such as: What kinds of social structures bring what kinds of outcomes? Which networks are good for what actors and under what conditions? When are networks vessels for social capital and when do they render social liability? The fundamental difference between the study of social networks and organizations in general and the overarching agenda of CSC is that, within the framework of CSC, social network structures are the focus of attention in their explicit productive or destructive association with goal attainment of corporate players. With this approach, CSC shifts attention to more daring questions. We move from a descriptive set of findings to a theoretically based orientation with practical implications. A CSC theory provides a different framework for understanding organizations.

Organizational Paradigms The study of corporate social capital suggests an analytical framework that cuts across other theoretical frameworks that are eminent in the literature on organizations--for example, Scou's (1992) paradigmatic typology of organizations as rational, natural, and open systems.23

The rational paradigm perceives organizations as 'collectives oriented to the pursuit of relatively specific goals.' The basic assumption of this paradigm is that of goal-oriented rationality. Rationality suggests a cost-benefit analysis--a balance between the interest of actors, costs, and payoffs. The rationality-based premise of our definition of corporate social capital is straightforward because of the explicit focus on the functional (negative or positive) aspects of social structures in their relation to the goals of corporate actors. Social structure in the framework of CSC has costs and benefits and is directed towards the attainment of goals.

The natural paradigm perceives organizations as 'collectives whose participants share a common interest in the survival of the system and who engage in collective activities, informally structured.' It is largely in the framework of social networks that collectives operate. These social networks can be measured, mapped, and related to outcomes-at the firm and individual levels. The development, creation, and maintenance of these social networks to fit corporate goals are what ultimately transform these networks into sustained corporate social capital. esc is thus explicitly related to the notion of organizations as natural systems. In the natural paradigm, the survival of the system is a goal-the challenge in studying corporate


social capital is to explain which social structures make the achievement of this goal easier or more difficult.24

The open system paradigm sees organizations as open systems 'embedded independent on continuing exchanges with and constituted by-the environment in which they operate.' The open system perspective is a natural context for the theory of corporate social capital, which emphasizes social structure in the relevant environment of firms and their members. An important asset of corporations lies outside and beyond their immediate boundaries. Firms are embedded in interorganizational networks, affecting the goal attainment of corporate players. A firm's environment is made of opportunities and potentially creative relationships that may further the goals and achievements of corporate players. From a social capital perspective, the boundaries of firms fade into their environmentcorporations are embedded in a networked environment. The relationships between (members of) the firm and players outside the firm's formal boundaries affect the efficiency and effectivity of the firm's internal organization and, at the same time, can be modified to suit the firm's goals. The ramifications of the existence of 'external' ties on corporations are at the heart of CSc. 25

Placing the notion of corporate social capital in the paradigmatic discourse of organization studies suggests that CSC contributes a new framework and an additional dimension to the study of organizations and organizational processes.

THE VOLUME

It is common sense that relationships are important, both for individuals and for corporations. It is also intuitively clear that not all relationships are (equally) useful. Unfortunately, it is largely unknown which relationships, under what conditions, are beneficial or obstructive to organizations or their members. It is often argued that part of the explanation of which corporations (individuals) do better than others is found in the relationships these corporations (individuals) maintain. But we are not yet sure what exactly it is in those structures that benefit some and impede others.

The world is becoming increasingly complex and dynamic. Organizations must incorporate even greater diversity to survive and thrive. More complexity compels more organizations to (try to) increase their social capital. Organizations enter into various forms of inter-corporate relationships. We are moving from core competence to core capabilities, increasing the tendency toward hybrid forms of organization. Changing labor contracts are now built on the notion of employability--organizations offer less permanent contracts.

In a competitive and fast-paced world, the differences between the winners and the losers will more strongly be determined by the way actors make use of their opportunity structures. Social capital represents the pipeline to those opportunities. Social capital theory has gained prominence in a wide range of academic fields, particularly in sociology, economics, political science, and managementlbusiness science and is appearing with increasing frequency in intellectual magazines and professional reports.26 Woolcock (1998: 184) even claims that social capital is 'arguably the most influential concept to emerge from economic sociology in the last decade.' He also points out that 'in social capital, historians, political scientists, anthropologists, economists, sociologists, and policy makers-and the various


camps within each field-may once again begin to find a common language within which to engage one another in open, constructive debate, a language that disciplinary provincialisms have largely suppressed over the last one-hundred-andfifty years' (Woolcock 1998: 188).

Notwithstanding the substantial insights in the literature into the advantages and-in a much smaller part of the literature-the predicament of social embeddedness, researchers still lack a coherent theory for explaining how and when social structure transforms itself into corporate social capital or into corporate social liability. We were both disappointed and inspired by this. Trying to develop a more coherent and encompassing framework for studying the interplay between social structure and social capital in the context of corporate actors, we aimed at extending the insights from various fields into a systematic exploration of the concept. We therefore invited prominent scholars to contribute to this volume and suggested they remain within their own field of expertise in writing their chapters. As a consequence, some of the chapters have a slightly different take on corporate social capital than we do in this chapter. Still, we believe that the chapters provide a coherent overview of the field . Some of the chapters are of a theoretical nature, some of them methodological, many of them empirical.

The first chapters in the book present conceptual issues dealing with CSC theory, models, and measurement. These chapters delineate the pressing challenges in these three domains. In the second and third section, CSC is discussed starting from social structure at the level of the individual. The effect of social structure is discussed as it facilitates or inhibits players in pursuit of their professional careersmoving ahead in a competitive environment. The third section explicitly highlights two basic characteristics of CSC: contingency and fragility-not every social network conveys social capital and social networks carrying social capital at one point in time can create social liabilities at another. Social structure at the firm level is considered in the next two sections---highlighting ways In which interorganizational networks bring social capital to the firm.

We had the pleasure of reading multiple versions of the chapters, moving from rough outlines to the texts you find in this book. If, after reading several chapters, the reader has an (increased) appreciation for the theory of corporate social capital, it has all been worth it. We hope you will have as much pleasure reading the book as we had putting it together. Enjoy!

We thank Laura Rittmaster and Judith de Kleuver for their help in producing this book. The assistance and gentle pressure provided by Julie Kaczynski and Elizabeth Murry of Kluwer Academic Publishers were instrumental for the project as well. We thank the School of Management and Organization (Cluster of Business Development) at the University of Groningen and the Davidson Faculty of Industrial Engineering and Management at the Technion for their support and resources.

NOTES

l. The first appearance of social capital in the scientific discourse (see soc-net discussion on the genesis of social capital) was in Marshal (1890) and Hicks (I942)-1hey, however, used the term to refer to different types of physical capital. Hannifin (1920) employed the term in the context of community studies. Again in the context of community studies, social capital was used by Jacobs (1965) and Hannerz (1969). Loury (1977, 1987) used the term in the field of child psychology. Bourdieu (1972, 1980) developed the term in reference to cultural capital. The first researchers who used social capital explicitly in the study of organizations were Flap and De Graaf (l986}-in their investigation of job mobility.


Coleman (1988, 1990) was the most influential in developing a general wide-scale theory of social capital as it is mostly used by scholars today. The work of Bur! (1992) was important in its wide visibility among students of organizations as well as for his explicit emphasis on actors ('players') who are described as competing in the market-place. Putnam (1993ab) has been influential in his application of social capital to macro development policy issues, some of which are used by macro world bank policy makers. 2. Even though most of Coleman's work did not deal with business organizations, it provided a lot of inspiration to scholars of the firm who used his theories in their study of organizations. 3. Third largest was Airbus with 124 customers, fourth largest Bombardier with 107. 4. Koelewijn (1997). 5. Coleman (1990). 6. Also see Gabbay and Sato (1996) 7. During the Second World War, German soldiers were entitled to two liters of beer a week. The Germans were not able to supply their troops that occupied The Netherlands, so they relied on Heineken Company, a Dutch brewery, to distribute beer to the German troops. However, in order for Heineken to distribute the yellow nectar to the Germans, the German army had to inform Heineken about the location and movements of its troops. Heineken then provided the Dutch resistance with this information. The business relationship with Heineken thus helped the Germans quench their thirst for beer, but also negatively affected the German ability to protect strategically sensitive information. 8. With the term 'corporation' we refer to both for-profit and not-for-profit organizations (see, for instance, the chapters by Ferlie and Doreian). Also, we both consider organizations that deal with tangible combodities, and those that provide services (see, for instance, the chapters by Iacobucci, Lazega, Sandefur et al.). 9. This finding is reported by Brass (1985a). 10. Gabbay (1997) calls this 'negative social capital.' II. Daft (1995). 12. Gabbay and Stein (1999), studying infrastructural network projects and their effect on the Middle East, develop the notion of 'regional social capital' inherent in country-to-country networks. 13. Van Engelen et al. (1999). Also see Amason and Schweiger (1994) and Dess and Priem (1995). 14. For an overview, see Leenders (1995b, 1999). 15. See Davis and Newstrom (1985), Hyatt (1989), Simmons (1985). 16. Exemplary is the resistance the Dean of a management school experienced when a new curriculum was introduced that was completely different from the old one. In their fear of having to give up some of their hobbyhorses and relinquish control over the new curriculum, a number of the professors mobilized the social ties they had within the school. They were not fighting the introduction of a new curriculum per se; they were merely fighting having to give up credit points offered for their personal favorite topics. The formal organization set up for the (substantive) development of the outlines of the new curriculum only provided these professors with limited influence. Still, at a meeting with the entire academic staff present, one of them got up and said to the Dean 'why do you even think you can make these decisions? Let the decisions be made where the real power is: with us!' He was (largely) right: the informal network still maintained a lot of the power that, formally, belonged to the dean. Although the dean officially had the authority, he would not be able to pass anything the informal network of professors wished to block. 17. See, Mintz and Schwartz (1985), Stokman et al. (1985), Useem (1979), and Pennings (1980). 18. Nohria, (1992a: 11). 19. Gabbay and Leenders (1999). 20. In some industries, contracts are occasionally signed that prohibit the leaving employee from working with former clients for a set number of years. 21. The discussion of the interplay between social structure at the firm level and social structure at the individual level is beyond the scope of this chapter. 22. Salancik (1995: 348). 23. We use here the most well known typology in the study of organizations. Other typologies can also be connected to corporate social capital. For instance, see Allison's (1971) typology of organizations as 'rational' and 'political' actors, as discussed in Pennings and Lee (this volume). 24. Also see Walsh et al. (1999) on el~onic networks and Gabbay and Stein (1999) on country-tocountry networks. 25. Modem technology and the increasing use of computer mediated communication systems extend the intuitive understanding of corporations far beyond these limited descriptions (Walsh et al. 1999). 26. See, for instance, World Bank (1997).

SECTION I

• Conceptual Issues theory, models, and measurement

•

Organizational Networks and Corporate Social Capital

ABSTRACT

1 David Knoke

Corporate social capital is defined as processes of forming and mobilizing social actors' network connections within and between organizations to gain access to other actors' resources. Following a brief overview of basic network concepts and principles, I discuss alternative theoretical explanations for the origins, spread, transformation, and erosion of social capital. Two sections next investigate how network dynamics have reshaped corporate practices and changed the employment contract between workers and their firms. In conclusion, researchers should conduct more empirical investigations and construct better theories about the mechanisms through which social capital networks change the fates organizations and their participants.

INTRODUCTION

Michael Eisner, the imperious chairman of the Walt Disney Company, hired his long-time friend Michael Ovitz to fill the media giant's presidency in August, 1995. That position had lain vacant for months following Frank Wells' death in a helicopter crash. Ovitz began his Hollywood career humbly, as a tour guide at Universal Studios followed by a stint in the William Morris Agency's mailroom. It soared after he co-founded the Creative Artists Agency in 1975, which soon grew into the entertainment industry's premier talent agency, representing a thousand film personalities including Tom Hanks, Barbra Streisand, and Tom Cruise. Ovitz became the Hollywood's most-powerful and most-feared figure, personally brokering such mega-deals as the Matsushita-MCA merger and SONY's acquisition of Columbia Pictures. In 1995, he lured CBS news executive Harold Stringer to


head Tele-TV, a new video-program joint venture of three regional telephone companies. Tele-TV was relying on the super-agent's connections to procure production deals with the Hollywood studios. But Ovitz's surprise defection to Disney, which was allied with a rival group of telecommunication companies, left CAA scrambling to hold onto its business and entertainment clients.

Industry observers assumed that Disney hired Ovitz to strengthen its ties to top Hollywood talent, to manage its recent $19 billion takeover of Capital Cities! ABC, and eventually to succeed Eisner as chair: 'It provides the company with Ovitz's wide network of contacts and skills,' said one stock analyst. 'Frank Wells was always a sounding board for Eisner. Ovitz will playa similar role' (Wells 1995). Yet after just fourteen acrimonious months, unable to agree with his boss on how best to run the company, Ovitz left the Magic Kingdom 'by mutual agreement.' Denying rumors of a personal feud, Eisner said, We have been doing business together while being friends for many years, and I know that both our professional and personal relationships will continue' (Van de Mark 1996). To salve his wounds, Ovitz took a reported $90 million severance package.

Although exceptionally dramatic in its particulars, the Disney episode exemplifies the use of social capital to achieve individual and collective goals, especially the importance of intra- and interorganizational connections for employee careers and corporate performances. Social actors-persons, groups, organizations, nations--continually interact in pursuit of their interests, and through those processes they generate complex webs of social relations that advance or thwart their goals. Three key issues, regarding the significance of social relations as social capital, permeate this chapter: 1) How do actors in organizations create, reproduce, and change their network ties to other participants? 2) How do actors' positions in multiplex network structures facilitate or constrain their acquisition of social capital and affect personal and organizational performances? And 3) how do these networkformation and performance-outcome processes interact and mutually shape one another over time? That is, selection and influence may operate reciprocally: connections may originate because well-performing organizations prefer ties to other outstanding actors, while superior performances may result from the social capital those network ties generate. Answering those crucial questions draws research attention to temporal sequences of network formation and consequences for corporate social capital at both individual and organizational levels of analysis.

The next section defines social capital as the processes by which social actors create and mobilize their network connections within and between organizations to gain access to other social actors' resources. Then I briefly develop some basic network concepts and principles. I next review competing accounts of the origins, proliferation, modification, and destruction of social capital. The following two sections investigate how network processes are reshaping corporate practices and changing the employment contract. The conclusion appeals for more research and theory construction about how social capital networks alter the fates of organizations and their participants.

Organizational Networks and Corporate Social Capital- 19

SOCIAL CAPITAL EMBEDDED IN SOCIAL RELATIONS

Following suggestions by Bourdieu (1980, 1986) and Loury (1977), James S. Coleman defined social capital as social-structural relationships that an individual could mobilize in actions 'making possible the achievement of certain ends that would not be attainable in its absence' (Coleman 1990: 302; see also Coleman 1986, 1988). In contrast to other forms of capital (physical, financial, human, and cultural capital), social capital resides neither in a person's skills and knowledge nor in such physical instruments as tools and machines. It is embedded in social connections and is 'created when the relations among persons change in ways that facilitate action' (Coleman 1990: 304). Indeed, for many purposes, a network tie is itself a form of capital. People and organizations actively shape their social relationships to obtain better opportunities and benefits. By forging large volumes of connections to numerous, diverse, and well-endowed contacts, a social actor gains potential access to the assets controlled by those contacts:

Social capital is at once the resources contacts hold and the structure of contacts in a network. The first term describes whom you reach. The second term describes how you reach. (Burt 1992: 12)

An actor's social capital encompass many types of relations, including trust and confidence, obligations and cooperation, and information. These 'moral resources' ultimately depend on shared social norms that sustain and strengthen the cooperative bonds among actors in a social system. The dynamics of social capital may parallel other forms of capital growth, with initial investments in riskier relations subsequently paying higher returns to the investors. For example, rural communities and small towns are typically pervaded by extensive webs of mutual social and economic assistance among the residents. Over time, these neighborly exchanges generate high reservoirs of trustworthiness that outstanding obligations to reciprocate will likely be repaid in future times of need. Persons in such interdependent social systems benefit from greater amounts of social capital 'credit' on which to draw, as demonstrated by heroic yet futile sandbagging efforts to prevent the Red River from inundating North Dakota's and Minnesota's riverine towns during the spring of 1997. Similar endeavors occur in The Netherlands, involving periodic evacuations of thousands of residents from lowland areas.

Several social network analysts propose that people mobilize their chains of direct and indirect social connections, particularly their 'weak ties' to higher-status persons in stratified social systems, to accomplish personal goals such as finding jobs and achieving upward social mobility (Granovetter 1973; Lin, Ensel, and Vaughn 1981; Boxman, De Graaf, and Flap 1991; Lin 1995a; Bian 1997; see the Higgins and Nohria chapter in this volume). Job-seekers obtain timely information about openings and necessary qualifications primarily through informal communication channels rather than from formal employment institutions (Rees 1966). An applicant's chances to secure a job or a promotion may depend less on human capital, in the form of school credentials and examination scores signaling achieved skills, and more on obtaining favorable assessments from key advocates and gatekeepers who can vouch for the applicant's reliability. For example,


JANE DICK

1 BANKRIGHT SQUAREBILT

Figure 1. Schematic of a multi-level social capital network

university hiring and promotion procedures typically require letters of recommendation from trusted assessors willing to testify about a candidate's worthiness. Thus, people's chances for improved career outcomes are often enhanced by their ability to tap into (,borrow') the scarce social resources held by actors beyond their immediate social circles. Resources accessed through network relations become an individual's social capital, regardless of whether those ties were explicit investments made by the person in expectation of a future payoff, or were overtly mobilized during instrumental actions (Lin 1995b). The wider the range and the more diverse the contacts directly and indirectly available to a job-seeker, the greater that person's chances for ultimate success. Thriving in a highly competitive labor market favors candidates possessing heterogeneous networks over applicants whose opportunities are restricted by their redundant, impacted networks (Burt 1992: 195; 1997).

The idea that social networks constitute an actor's social capital investments can be readily extended to social groups and corporate entities. Just as an individual can mobilize her personal contacts' social resources for purposive action, so can a formal organization activate various resource networks to achieve its goals. Figure 1 illustrates a hypothetical social system involving complementary exchange networks at macro- and micro-levels of analysis (for simplicity, I omit intermediate groups such as teams and departments). The circles represent the formal boundaries of two organizations: Bankright, a financial corporation, and Squarebilt, a construction company. This example highlights two key employees: Jane, a Bankright commercial loan officer, and Dick, Squarebilt's chief controller. Pairs of directed lines (arrows) represent social exchange relations linking these workers and their enterprises. Within each organization, the employees supply resources essential to their firms' operations, specifically financial expertise and experience in loanmaking and acquisition. As returns on these human capital investments, their firms reward Jane and Dick not only with salaries, fringe benefits, and job-training opportunities, but also with increased social capital via their relationships with other company employees, for example, authority, social status, and collegiality.

The two pairs of horizontal lines depict resource exchanges that span organizational boundaries at both levels. As agents for their organizational principals, Dick and Jane directly negotiate business loans, but apart from these roles they may also maintain personal and professional ties, such as common school,

Organizational Networks and Corporate Social Capital - 21

church, and voluntary association affiliations. At the institutional level, the two companies are linked primarily through their long-standing banker-client relationship. Such financial transactions are often reinforced by solidary connections between their chief executive officers, who belong to the same social clubs and political parties, and whose families may even intermarry. These instrumental networks payoff by turning one actor's social resources into the social capital investments of other individuals and corporations. Squarebilt gains access to Bankright's resources indirectly whenever Dick mobilizes his ties to Jane, who possesses authority to make construction loans. Alternatively, were Jane to reject his loan application, Dick could mobilize his connections to Squarebilt's CEO to tap into the firm-level authority of Bankright's CEO and bring pressure to bear from above on Jane to reverse her decision.

Actors often nurture and manipulate their social relations as deliberate strategies for coping with uncertainties arising from dependence on external environments for many critical resources needed to 'get the job done.' Explaining how people and organizations actually behave requires considering factors beyond the purely arm's length economic transactions occurring in spot-market exchanges between anonymous buyers and sellers, where efficiency criteria are allegedly the paramount determinants. Many economic relations are embedded within larger social, political, and legal contexts (Granovetter 1985), which constrain participants' choices and actions according to normative and political criteria transcending pure cost-benefit calculations. These contexts can be conceptualized and empirically modeled as multiple networks, connecting diverse social actors with varying interests and resources, activated for individual and collective purposes. Hence, researchers ignore network structures at the peril of providing incomplete insights into organizational structures and processes.

Mixed competitive and cooperative modalities suffuse many types of network interactions. At times actors mobilize their social capital to gain personal advantages over their adversaries, while in other circumstances they jointly coordinate actions for collective benefit. For example, corporate employees engage in self-serving career strategies, seeking out mentors or networking with superiors to advance up the promotion ladder. In contrast, new management practices and workplace designs stress teamwork and collaborative responsibility for production, encouraging workers to pool their skills and social capital to improve group performance. Similarly, both modalities operate at the level of organizational strategy where plans to achieve global corporate goals are implemented. Firms operating in the same industry generally compete for customer loyalties and form exclusionary supplier relations, yet they may also collaborate in strategic alliances and joint ventures with expectations of mutual gains. An important task for network theory is to explain under which conditions zero- and positive-sum interactions are more likely to occur.

SOME NETWORK FUNDAMENTALS

This section briefly reviews some fundamental issues facing network researchers. I explicate these concepts and principles in a nontechnical manner, which should suffice for understanding the chapters in this volume. Readers seeking deeper


knowledge should consult such didactic texts as Knoke and Kuklinski (1982) or Wasserman and Faust (1994).

Identifying Actors and System Boundaries Networks are social constructions arising from exchanges and joint activities among participants in a social system, defined as a 'plurality of actors interacting on the basis of a shared symbol system' (Parsons 1951 : 19). The actors in a network may be designated at varying levels of analysis: individuals (children in a kindergarten class); small groups (work teams on an automobile assembly line); formal organizations (corporations in business association); coalitions (lobbying alliances); even nations (members ofthe World Trade Organization).

Identifying a social system's boundaries, and hence its size, requires specifying which potential members are relevant to the system's functioning. Investigators using a nominalist strategy typically achieve conceptual closure by including all actors possessing one or more key characteristics (Laumann, Marsden, and Prensky 1983). Nominal designations often restrict network membership to incumbents occupying formal positions, for example, directors of Fortune 500 companies or middle managers at Apple Computer. The alternative realist approach to boundary specification assumes that system participants themselves can best identify who belongs. Uncovering the network members' subjective meanings requires a researcher to designate 'a substantively defined criterion of mutual relevance or common orientation among a set of consequential actors' (Knoke and Laumann 1982: 256). Typically, potential network members carry out a reputational ranking of other actors according to their importance to the system's performance. Actors enjoying high reputations, indicating that their peers believe they must be taken into account, are included but actors with low or no reputations are dropped because of their marginal importance.

Total and Ego-Centric Networks For a small social system, researchers may be able to obtain data on connections among all system participants, comprising a total network. 'Small' in this context means anywhere from a dozen to several hundred actors. To construct a total network among G actors requires each actor to report on existence of ties to all other system members, typically by checking a previously compiled name list (see questionnaires for U.S. energy and health domains in Laumann and Knoke 1987: 401 -500).

Some social systems are either too large or too weakly connected to collect total networks. In such instances, the only feasible alternative representative is to draw a sample from the target population and to elicit the direct network ties of each sampled actor (an ego-centric network). For example, a high-tech firm studied by Burt (1992) employed 3303 managers who had few direct or indirect connections to one another. He sampled 547 managers and ask them to describe their social and work-related contacts. An ego-net procedure identifies a unique set of network alters, typically through a two-step name-generator protocol. First, an informant ego lists the most important people with whom specific types of interaction occurred. Then, ego describes each alter's key attributes, such as age and gender, the nature


relations with each alter (e.g., closeness and frequency of contacts), and perceived direct ties among the alters. Importantly, because the alters are not subsequently interviewed, ego's self-reported information remains unconfirmed. The ego-net approach is the only plausible network methodology for general population surveys (Marsden 1987) and large samples of diverse organizations (Kalleberg, Knoke, and Marsden 1995).

Network Relational Contents The next task is deciding what substantive relationship to measure. Network' s relational contents refer to some relatively homogeneous tie among actors. Relational contents seem to fall into two general categories: I) A transaction involves exchanges where one actor yields rights of control over some physical commodity or intangible value to another actor, possibly in expectation of eventual reciprocation; 2) Joint actions require actors to co-participate in an event located in specific time and space, without relinquishing control over resources. A minimal joint action is mere co-presence, for example, software firms attending the annual Las Vegas COMDEX trade fair. More intense joint activity involves coordinated efforts to achieve common goals, for example, lobbying the Congress on legislative proposals (Knoke et al. 1996).

Deciding which types of network content to operationalize should be guided by a project's theoretical objectives: what substantive relations are useful for understanding important actor and system behaviors? In practice, multiplex ties link social actors into complex webs, meaning that no such creature as 'the network' exists for any system. Typically, more than one relational content may be studied in a given project.

The following typology roughly classifies relations by increasingly formality:

Recognition Minimal awareness, with actors reporting whether they 'know about' or 'have ever heard of others in the system. Because recognition is typically unreciprocated, mutuality in choices is rare, thus differentiating the 'stars' who enjoy high visibility from their anonymous fans in sports, politics, and science networks. Among organizations, corporate reputations for quality products and services similarly differentiate the well-regarded from the invisible players (Fombrun 1996).

Co-Attendance Common presence at the same events, or membership in the same collectivities, disregarding direct interaction. Mass public assemblies, such as political rallies and athletic contests, anchor one end of a continuum, while co-participation in restrictedaccess enterprises, such as private schools and social clubs, implies the existence of potentially cohesive social classes (Domhoff 1975).

Information Exchange Routine and regular communication of data, whether about scientific-technical matters (a supplier's current catalog of available products and prices) or sociopolitical affairs (claimant organizations' positions on legislative bills). Such information may be widely broadcast through press releases and email-server


networks, or more narrowly targted on specific recipients, for example, a marketing unit providing the sales department with consumer survey results.

Advising Confidential information intended to help a recipient gain advantage over competitors. An advisor transmits her superior knowledge and experience to an advisee, sometimes with no expectations beyond gratitude and deference. Organizational superiors mentor their favored underlings (Keele 1986; Noe 1988b), expert consultants familiarize their organizational clients with R&D opportunities, and venture capitalists scrutinize start-up investment prospects. (See the chapter by Freeman in this volume.)

Trust Confidence in the dependability of another actor's promises, reducing the chances of opportunistic behavior by one's partners. Trust relations run a risk that such reliance will prove ill-founded, but their efficiency in reducing transaction costs is an important precursor for building long-term exchanges between people and organizations. (See the chapter by Nooteboom in this volume.)

Support Expressions of sympathy, empathy, or commiseration in times of need, including instrumental actions that demonstrate solidarity with an afflicted actor. At the individual level, friendship and kinship bonds comprise the most obvious varieties of support. Support relations among organizations involve public legitimation, for example, testimonials regarding one company's willingness to rally behind another enterprise confronting political or legal difficulties.

Financial Aid Transfers of money, credit, or physical facilities, other than arm's-length market purchases of goods or services. Some financial ties seem altruistic, as in corporations' philanthropic donations to charities (Galaskiewicz 1985), while other exchanges attach implicit quid pro quo strings, such as corporate contributions to parties and candidates via political action committees (Mizruchi 1992). Other financial exchanges entail explicit ownership and control connections, such as the debt and equity transactions linking corporate and banking members of Japan's famous keiretsu (Gerlach 1992a).

Authority Legitimate power to expect that commands will be obeyed, backed up by sanctions for failures to comply. Employment relations in corporate and public bureaucracies typify intraorganizational authority networks, while a regulatory agency's power to constrain an industry embodies interorganizational ties. The interlocking directorate, connecting corporations and banks (Mintz and Schwartz 1985; Stokman, Ziegler, and Scott 1985) and nonprofit organizations (Useem 1979), may reflect hybrid authority patterns where agents of outside organizations help set policies binding on participants inside another organization.

Alliances Collaborative arrangements involving two or more organizations that combine resources to pursue common or complementary objectives, often involving applications of uncertain technologies or entry into risky markets. These cooperative strategies vary in the substantive content of interorganizational cooperation and


types of governance structures designed to monitor and control the partners' behaviors. (See the chapter by Stuart in this volume.)

The nine types of ties classified above are illustrative, but not exhaustive, of the diverse relations that network analysts may find useful for their theoretical interests.

Magnitude of Ties Network informants should also indicate the magnitude or value of their relations. At a minimum, only a dichotomous coding-presence (1) or absence (O)-might be recorded, for example, using name list checkoff. The unchecked alters yield important data, since a network's structure depends as much on its gaps as on its direct connections. More detailed magnitude codings assign scalar values reflecting each tie's relative strength. For persons, tie strength usually refers to subjective intensity of commitment, for example, a friendship study requesting egos to indicate which people are their 'casual,' 'close,' or 'best friends' (Leenders 1996). For organizations, relational magnitudes may involve objective data, for example, the dollar amounts of loans from commercial banks to manufacturing firms, or the numbers officers sitting on other companies' boards of directors.

Another important consideration is the time span observed: too short and important but infrequent relations may be overlooked; too long and dormant ties might mistakenly be treated as current. Unfortunately, the temporal dimension hasn't been well-integrated into research procedures. Most network projects yield static snapshots of a long-established network, without revealing their origins, evolution, and ultimate fates. For example, we know little about whether informal ties between employees of different companies subsequently generate organizational alliances, or whether the opposite causal process occurs. Despite evident theoretical payoffs from understanding network dynamics, data collection has not kept pace with recent methods for investigating network changes over time (Wasserman and Iacobucci 1988; Frank 1991; Zeggelink 1994; Snijders 1996; Leenders 1996).

Network Forms Basic network forms describe the patterns connecting system actors regardless of their specific relational contents. Figure 2 displays a hypothetical chooser-by-chosen binary adjacency matrix and its associated graph. Think of the {ABCD} subset as a production department located in one building, while the {WXYZ} subset is a geographically dispersed salesforce. Actors A and W are these units' respective heads. The relational content is regular communications about work. Each matrix row represents a potential sender and each column a receiver of dichotomous social ties. A 'I' entry indicates that the row actor communicated with the column actor, while '0' means that no communication occurred. Graphs depict actors as labeled points and their relations as arrows pointing from sending actors to receiving targets. Because every communication tie is reciprocal, all ten arrows are doubled-headed.

The network volume is the total number of ties and its density is the proportion of observed ties to the number of possible connections not counting self-ties (i -g for a g-actor system). The example has a volume of 20 ties and a density of (20/56) = .357. Actor connectedness counts the number of non-zero entries in a matrix row


PRODUCTION DEPARTMENT SALES FORCE

B X

/' /' c A ... • W I( .y

'" '" Z D

A B C D W X Y Z A 0 1 1 1 1 0 0 0 B 1 0 1 1 0 0 0 0 C 1 1 0 1 0 0 0 0 D 1 1 1 0 0 0 0 0 W 1 0 0 0 0 1 1 1 X 0 0 0 0 1 0 0 0 y 0 0 0 0 1 0 0 0 Z 0 0 0 0 1 0 0 0

Figure 2. Graph and matrix representations of a hypothetical eight-actor network

(out-degrees) and column (in-degrees). Thus, department heads A and W are the best-connected actors, with out- and in-degrees = 4. Next come B, C, and D with three ties each, trailed by the dispersed salespeople with just one connection to their boss. Successively mUltiplying a matrix by itself reveals the minimal path length required to connect pairs of actors. Visually, a path can be traced across directed arrows between pairs, with the length being the number of steps needed to connect that dyad. For Z to pass a message to C requires a path of length = 3: (ZW) + (W A) + (AC). The four production members are connected by one-step paths (direct ties), but the salespeople require 2-paths to reach one another.

A and W enjoy unique and powerful roles in the system, an insight confirmed by measures of actor centrality (see Freeman 1977, 1979). Basically, a central actor participates in a large volume of social relations, with refined centrality concepts differentiating among the type or 'quality' of connections. The simplest centrality measures is an actor's in-degrees, measuring the sheer volume of ego-centric contacts received from alters. Closeness centrality captures the extent to which ego maintains connections to many alters who themselves have many non-overlapping ties, thus enabling ego to reach many others by relatively short paths. A and W each have the highest closeness scores (70), while B, C, and D enjoy somewhat higher closeness (50 each) than the less-connected X, Y, and Z (43.8 each). Betweenness centrality reflects an actor's ability to mediate many connections between subgroups, thereby potentially leveraging greater impact on system activities. Because the salespeople are less connected than are the production employees, W has a higher betweenness score than does A (15 to 12), while the other actors' betweenness centrality scores are O.


The final network forms cluster actors into positions, which simplifies the system's social role structure. Two basic approaches involve cohesion and equivalence criteria. A clique is a network sub-set in which all dyads are maximally connected (all reciprocal direct ties occur, yielding a density of 1.00). By this rigorous definition, only the {ABCD} cluster comprises a genuine clique. Two actors are structurally equivalent to the extent that they display identical or very similar patterns of ties to all other alters, regardless of their ties to one another (Sailer 1978). For example, firms in an industry that buy from the same sources and sell to the same customers are fundamentally interchangeable competitors from the market's perspective. The four equivalent blocks are {A}, {W}, {BCD}, and {XYZ}. Automorphic equivalence identifies actors i and j are automorphically equivalent if, after removing the 'names' of the actors from the nodes, nodes i and j are impossible to distinguish. In the example, the sets of automorphic ally equivalent actors are {A W}, {BCD}, and {XYZ}. In contrast to structural equivalence, which puts A and W into separate positions because they supervise different individuals, they are automorphically equivalent because they are connected to corresponding others-their work-unit subordinates.

INTERORGANIZATIONAL NETWORKS

By the 1980s, converging environmental pressures began restructuring forever the ways organizations would relate to their competitors, employees, customers, and the larger society. While every analyst offers a favor~te list of key factors driving organizational change, the six master trends cited by the Hay Group (Flannery, Hofrichter and Platten 1996) seem particularly concise and comprehensive: rapidly expanding technologies; growing global competition; increased demand for individual and organizational competencies and capabilities; higher customer expectations; ever-decreasing cycle times; and changing skilled personnel requirements. A seventh trend, at least in the U.S., is increased investor pressure on companies to improve their short- and long-term financial performances (Useem 1996). As corporations grew increasingly exposed to international competition, they sought new ways to remain viable by slashing costs and prices, improving production performance, and responding rapidly to technological innovations and fickle consumer preferences. With consumer demand simultaneously globalized and fragmented, niche markets for specialty goods and services supplanted cumbersome mass-production systems run by 'Fordist' principles. Firms perceived performance gains from unbundling their internal hierarchical structures and deinstitutionalizing the product-unrelated conglomerate form (Davis, Diekmann, and Tinsley 1994). These incessant pressures to achieve corporate flexibility and specialization drove organizations to restructure their internal employment systems along much more participatory lines. They also compelled companies to reach outside their traditional boundaries to form long-term collaborative relationships enabling them to stay afloat in an increasingly cutthroat world economy.

One consequence of the strategic search for new competitive advantages was the proliferation of many new interorganizational forms. Figure 3 presents an alliance typology, modified primarily after Yoshino and Rangan (1995: 8). At one extreme are pure market relations, whose transactions require no enduring collaboration by


exchanging parties. At the other extreme are hierarchical arrangements in which one firm assumes full authoritative control over the other, absorbing the participants into a unitary enterprise. Between these extremes fall various 'hybrid' arrangements that are neither clearly markets nor hierarchies but typically blend elements from both types (Jensen and Meckling 1976; Williamson 1975; Powell 1987; Heydebrand 1989). An appropriate label might be the N-form or 'networked' organization, to emphasize that relations are central to these mixed structures. For Yoshino and Rangan (1995: 5), a strategic alliance's critical characteristics are partner firms that: 1) remain independent after the alliance is formed; 2) share benefits and managerial control over the performance of assigned tasks; and 3) make continuing contributions in one or more strategic areas, such as technology or products.

Based on this definition, they classified licensing and franchising as traditional market contracts because one company grants another the right to use patented technology or production processes in return for royalty payments. However, Figure 3 reassigns franchising under non-equity partnerships, since many distribution

Hierarchical relations

Equity partnerships

Non-equity partnerships

Multi-participant alliances

Market relations

Subsidiaries Acquisitions

Mergers

Minority equity investments

Equity swaps

Joint ventures

Business groups

Franchising

Small firm networks

Joint R&D, production development, manufacturing,

marketing, distribution, service & other functions

Data banks and information clearinghouses

Standards-setting consortia

Government-sponsored R&D consortia

Trade associations

Cooperatives for purchasing, marketing

Action sets for lobbying campaigns

Cartels

Spot exchanges

Arm's length buy-sell

Short-term subcontracting

Licensing

Figure 3. Types of interorganizational alliances Modified after Yoshino and Rangan (1995)


franchisers such as McDonald's Corp. exercise centralized coordination to safeguard their corporate interests while leaving ownership and operations to the local entrepreneur (Reve 1990: 148; also Osborn and Baughn 1990).

While some alliances require only a bilateral (dyadic) relations, other forms involve mUltiple participants structured into complex 'alliance networks' (GomesCasseres 1996: 52). Resource and authority commitments also vary considerably among the hybrid types, particularly the extent of equity exchanges among partners. Space limitations prevent detailed discussions of each interorganizational form, but brief distinctions among the most important types may be helpful. An action set is a short-lived organizational coalition whose members coordinate their efforts to influence public policy decisions, for example, the passage of legislative acts affecting the coalition members' collective interests (Knoke et al. 1996: 21). Cartels, or pools, are unstable alliances formed to constrain competition by cooperatively controlling production and/or prices in specific industries (Fligstein 1990: 39). The 'trusts' flourishing in late 19th-century America - in railways, heating oil, steel, aluminum, sugar, salt-were outlawed but cartels periodically arise elsewhere, for example, OPEC.

Small-firm networks (SFNs), are a late 20th century innovation involving large numbers of very small firms (often with fewer than 10 employees) interacting on a long-term basis, 'sharing information, equipment, personnel, and orders, even as they compete with one another' (Perrow 1992: 455). They are complex clusters of raw materials suppliers, producers, financial service, marketing, and distribution firms. SFNs arise primarily in clothing, toys, publishing, motion pictures, construction, light machinery, and electronics industries rather than in heavy manufacturing or extractive industries. For example, Bennetton, an Italian apparel producer, owns very few facilities but parcels out almost all production tasks to hundreds of firms employing thousands of workers (Clegg 1990: 120-125; Kanter, Stein, and Jick 1992: 228). The regional economies of Emilia Romagna in northern Italy (Brusco 1982; Lazerson 1988) and Baden-Wiirttemberg in southwestern Germany (Herrigel 1996) are the most famous SFN exemplars, as is perhaps Silicon Valley in northern California (Saxenian 1994). SFNs should not be confused with the more prevalent small supplier networks dominated by a large industrial firm such occur in Germany's Ruhr Valley (e.g., Grabher 1993) and the American and Japanese automobile industries (Womack, Jones and Roos 1990).

Four main types of equity partnerships occur. In a minority equity investment, one firm buys an interest in another by direct investment, perhaps only a 3-5% stake, 'although the active involvement of the management of the partner-company is retained and the assessment of expertise of the company can be made without complete integration' (Hagedoorn 1993a: 132). Examples abound in high-tech fields where large corporations use minority shareholding to access start-up firms' technologies. An equity swap involves mutual direct investments of partners. A joint venture occurs when 'two or more legally distinct firms (the parents) pool a portion of their resources within a jointly owned legal organization' (Inkpen 1995: 1) that serves a limited purpose for its parents. For example, the well-known NUMMI automobile plant in Fremont, California, enabled General Motors to learn about Japanese management techniques while Toyota gained a U.S. foothold (Adler 1993).


Joint ventures may involve 50:50 ownership between two parents (Lewis 1990: 173-192) or unequal equity shares among mUltiple partners.

Finally, a business group is a coherent collection of firms bound together at an 'intermediate' level between short-term strategic alliances and the unitary corporation (Granovetter 1994: 454). East Asian partnerships among manufacturers, suppliers, and financial institutions-such as the Japanese keiretsu (Lincoln, Gerlach and Takahashi 1992; Gerlach 1992a) and Korean chaebol (Steers, Shin and Ungson 1989)-exemplify business groups spanning multiple fields that are integrated through complex debt, interlocking directorates, and equity ownership patterns. Although some observers argue that Chrysler Corp.'s comeback allegedly resulted from an American weak-tie version of keiretsu, its cooperative relations with parts suppliers involve neither the equity investments nor the management connections that Toyota and Nissan have with their suppliers (Dyer 1996b). (See the chapter by Pennings and Lee in this volume.)

The following subsections examine explanations of three aspects of interorganizational alliances: why they occur, how they develop, and their consequences. In seeking to understand network development over time, we should ask whether any single theory can account for such diverse alliance phenomena or whether distinct theories are required?

Theories of Alliance Formation Two prominent theoretical explanations of why organizations engage in nonmarket relations are transaction cost analysis and resource dependence theories, respectively emphasizing economic and socio-political factors. Transaction costs determine where to draw an economically efficient boundary between an organization (hierarchy) and its environment (market), in other words, whether to make or buy a particular function. Oliver Williamson (1975) argued that the most important factor driving organizational efforts to economize is asset specificity, the extent to which investments are specialized to particular recurrent transactions between buyers and sellers. The greater the specificity, the more likely are the parties to 'make special efforts to design exchanges with good continuity properties' (Williamson 1981: 555), thus effectively locking them both into prolonged bilateral exchanges. For example, a corporation requiring only intermittent legal advice is likely to retain an outside legal firm, while a company with persistent legal problems may create its own in-house legal department. Thus, interorganizational ties arise from specialized investments that would lose their value if transferred to another exchange partner. Otherwise, market exchanges will be more cost-efficient. A second core assumption of transaction cost analysis is that at least some actors are 'given to opportunism' (Williamson 1981: 553), that is, dishonesty and dissembling about preferences and information. Employing a different terminology, Williamson argues that interorganizational relationships that convey social capital, are at risk of being turned into social liability by opportunistic actions of one of the parties. The necessity to monitor partners' performance and safeguard against duplicity can increase interorganizational transactions costs. For example, Seagram's Universal Studios successfully sued Viacom Inc., claiming the latter's launching of a


competing cable TV network violated their agreement on joint operation of the USA Network (Shapiro 1997).

Resource dependence explanations of alliance formation emphasize inherent tensions between organizational resource procurement needs and the desire to preserve freedom of strategic decision making. Intercorporate relations arise from interdependencies and constraints among organizations: situations where one organization controls the critical resources or capabilities-such as money, information, production and distribution skills, access to foreign markets-needed by another organization. Alliances tend to occur more often among interdependent than between independent firms , that is, where complementarity rather than similarity prevails. However, organizational efforts to manage problematic external interdependencies 'are inevitably never completely successful and produce new patterns of dependence and interdependence' (Pfeffer 1987: 27). Dependence theorists argue that network ties arise from managers' efforts to control the most troublesome environmental contingencies through complete or partial absorption (e.g., mergers or joint ventures).

In their drive to acquire critical resources from network partners, organizations risk losing control of their own destinies (social liability). Resource dependence generates interorganizational power differentials that constrain firms' opportunities, since organizations tend to comply with demands from the more powerful actors in their environment. 'Organizations seek to form that type of interorganizational exchange relationship which involves the least cost to the organization in loss of autonomy and power' (Cook 1977: 74). Given an opportunity set of potential alternative providers, a company will optimally choose a partner that can best satisfy its resource needs while imposing minimal constraints on its discretionary actions. For example, confronted with many suppliers capable of providing equivalentquality inputs, a large manufacturing firm is likely to purchase from the smallest supplier, thereby gaining power to impose terms and conditions. Similarly, a small supplier would prefer to spread its business across many customers, thereby avoiding the loss-of-control stemming from dependence on a single partner.

Few analysts have explicitly tested hypotheses about alliance formation drawn explicitly from the transaction cost or resource dependence perspectives. Pfeffer and Nowak (1976) found that resource interdependencies (high exchange of sales and purchases) among companies in technologically intensive industries significantly increased joint venturing at the industry-level of analysis. Zaheer and Venkatraman (1995) tested hypotheses about interorganizational strategies drawn from transaction cost economics and social exchange perspectives, using data from a mail survey of 329 independent insurance agencies. Their two dependent variables were 'vertical quasi-integration' (the percent of total premiums handled by an agency's 'focal carrier,' the company with which an agency conducted most of its business) and Joint action' (a multi-item scale measuring planning and forecasting activities with the focal carrier). Although transaction-specific assets predicted quasi-integration, neither uncertainty nor reciprocal investments were statistically significant. Instead, quasi-integration and joint action were both positively related to mutual trust between agency and carrier, a relationship opposite to the transaction cost hypotheses but consistent with social exchange theory.


Resource dependence principles seemed more helpful than transaction cost concepts for understanding cooperative networks between new biotechnology firms (NBFs) and established corporations in the 1980s (Barley, Freeman, and Hybels 1992; Kogut, Shan, and Walker 1992; Powell and Brantley 1992; see the chapters by Smith-Doerr et al. and Stuart in this volume). Complementary resource needs drove strategic alliances, primarily involving exchanges of financial support for technical expertise. The small, innovative R&D laboratories typically lacked funds, public legitimacy, and in-house capability to market their products and maneuver through the regulatory maze. Hence they allied with diversified, resource-rich pharmaceutical, chemical, and agricultural companies able to provide sustaining resources. In turn, these established firms welcomed collaborative agreements as means to acquire tacit knowledge and to learn new technological skills from their NBF partners (social capital). As relationships accumulated and stabilized over time, the network positions occupied by individual organizations constrained their access to information regarding potential alliance partners (social liability). 'It is the structure of the network, rather than attributes of the firm, that plays an increasingly important role in the choice to cooperate' (Kogut, Shan, and Walker 1992: 364).

Two studies of changing networks patterns in other fields underscore the importance of past ties on future actions. Leenders (1995b) reanalyzed dyadic data from the social service networks of two Pennsylvania counties between 1988-90. Informants named the organizations with whom their agencies maintained relations, such as coordinating client treatments or sharing funds and personnel, including ties mandated by the state. In both counties, estimated dyad-transition models revealed that 'reciprocity both increases actors' inclination of creating and maintaining ties and decreases the inclination of withdrawing ties' (Leenders 1995b: 193). Although he did not use the term corporate social capital, the evolution of these interorganizational networks clearly fits such an interpretation.

In a study of dyadic international corporate alliances, Gulati (1995a) found evidence consistent with both resource dependence (which he called 'strategic interdependence') and social structural explanations. Using a 1980-89 panel of 166 corporations operating in three worldwide sectors (U.S., Japanese, and European new materials, industrial automation, and automotive products firms), he conducted event-history analyses on a variety of dyadic alliances ranging from arms-length licensing agreements to 'closely intertwined equity joint ventures' (1995a: 634). Strategically interdependent firms (i.e., those companies operating in complementary market niches) formed alliances more often than did firms possessing similar resources and capabilities. Previously allied firms were more likely to engage in subsequent partnerships, suggesting that 'over time, each firm acquires more information and builds greater confidence in the partnering firm' (l995a: 644). Beyond a certain point, additional alliances reduced the likelihood of future ties, perhaps prompted by fears of losing autonomy by becoming overly dependent on a single partner. Indirect connections within the social network of prior alliances also shaped the alliance formation process: previously unconnected firms were more likely to ally if both were tied to a common third-party, but their chances of partnering diminished with greater path distances. Gulati concluded that 'the social network of indirect ties is an effective referral mechanism for bringing firms


together and that dense co-location in an alliance network enhances mutual confidence as firms become aware of the possible negative reputational consequences of their own or others' opportunistic behavior' (1995a: 644). His results reflected a logic of clique-like cohesion rather than status-competition among structurally equivalent organizations.

Trust as Corporate Social Capital The formation of successful strategic alliances between corporations hinges on creating and sustaining relationships among the partners based on mutual trust. At the individual level, we consider a person trustworthy if 'the probability that he will perform an action that is beneficial or at least not detrimental to us is high enough for us to consider engaging in some form of cooperation with him' (Gambetta 1988c: 217). At the interorganizational level, trust provides a foundation for one firm to achieve some degree of social control over another's behavior under conditions of high uncertainty. From a transaction cost perspective, the social capital of trust expectations may provide an efficient mutual deterrent to both partners' temptation to opportunism or malfeasance, thereby reducing alliance costs relative to more formal control mechanisms such as written contracts (Gulati 1995a: 88-91). Hence, interfirm trust relations fall conceptually somewhere between the polar logics of hierarchical authority and market price relations (Bradach and Eccles 1989: 104; Sako 1991).

The business-risk view of trust stresses confidence in the predictability of one's expectations hedged by formal contractual means such as insurance (Luhmann 1979). Ring and Van de Ven (1994) emphasized an alternative psychological conceptualization of trust as confidence in another's goodwill, of faith in the partner's moral integrity. In their approach, trust constitutes a fundamental type of organizational social capital, a strong-tie relationship between an ego firm and the alters comprising its organizational field . Organization attributes and network relations interact over time. As a company builds a reputation among its peers for fair dealing and impeccable reliability in keeping its promises, that reputation itself becomes a prized asset useful for sustaining its current alliances and forming future ones. Reputed trustworthiness signals to potential partners that an organization is unlikely to act opportunisticly because 'such behavior would destroy his or her reputation, thus making the total outcome of the opportunistic behavior undesirable' (Jarillo 1988: 37).

The social psychological explanation of trust is rooted in basic social exchange principles, including conformity to such norms as reciprocity, commitment, forbearance, cooperation, and obligations to repay debts (Stinchcombe 1986; Bradach and Eccles 1989: 105; Lewis and Weigert 1985). Because typical interfirm transactions are widely separated across time, trust reinforces these ties by invoking such principles as that exchange values should balance over the long run, and that each partners' payoffs should be roughly proportional to their contributions to any joint enterprise. As trust relations became historically institutionalized in modem industrial societies (Zucker 1986), initial arms-length market transactions grew increasingly suffused with many normative connotations, generating and upholding


STRucrURAL CONDITIONS

COMMUNICATION NETWORK

TRUST )

Figure 4. Trust is an intervening factor in the alliance formation process

ALLIANCE FORMS

the moral communities within which trustworthiness conveyed great importance in members' decisions whether to continue or break off relations.

What are the macro-level sources of trust among organization? Figure 4 proposes that interorganizational alliances emerge over time with trust occupying a pivotal role between antecedent conditions and consequent alliance formations. Note the feedback loop in which trust shapes the form of alliance, while events occurring during the alliance may subsequently transform the interorganizational trust relations, either reinforcing or weakening each partner's belief the other's trustworthiness. Thus, trust and alliance relations mutually change one another as interactions accumulate over time.

As suggested in Figure 4, communication networks structure an organization's ability to screen and evaluate initial information about potential alliance partners. These exchanges involve factual data about alters' interests and competencies, but also provide indirect evidence about other organizations' trustworthiness via path connections to knowledgeable peers in an organizational field. The more central an organization's position within a field's communication network, the greater its visibility and hence more informants are available to testify regarding its reliability and integrity. Organizations located in peripheral positions have fewer opportunities to become familiar with potential alliance partners and for their own trustworthiness reputations to become vetted by the field.

A second set of antecedent factors fostering or thwarting trustworthiness are macro-structural conditions. Imbalances in the resources controlled by each organization (such as their financial size or market shares) may impede trust creation because of unequal partners' inability to satisfy their reciprocity obligations. Pairs of organizations that share similar or complementary characteristics are more likely to develop strong trust relations. Tacit understandings and taken-for-granted assumptions may be rudely violated when partners have little in common. For example, many cross-border alliances, undertaken between foreign partners to gain access to local markets, are fraught with pitfalls stemming from incompatible national cultures (Lewis 1990: 253-278; Lorange and Roos 1992: 177-204; Bleeke and Ernst 1993: 12-13; Gilroy 1993). Even domestic alliances can suffer from clashing corporate cultures. A major instance was the office-network software


producer Novell Inc.'s disastrous effort to integrate its subsidiary WordPerfect Corp.'s 'close-knit and insular' staff with the parent organization's profit-driven style (Clark 1996). After two years of plummeting market share and stock prices, Novell sold WordPerfect to Corel Corp. at one-tenth its original $1.4 billion acquisition price.

The feedback loop between trust and alliance depicted in Figure 4 implies a temporal dynamic to changing governance forms through accumulating interorganizational experiences (Smith et al. 1995). Many alliances begin with formal linkages that expose the partners only to small risks. Because the organizations as yet have few bases for trusting one another, equity-based contracts (,hostage-taking') predominate as legal protection against potential opportunism. But after partners gain confidence in one another through repeated testing, then 'informal psychological contracts increasingly compensate or substitute for formal contractual safeguards as reliance on trust among parties increases over time' (Ring and Van de Ven 1994: 105). This substitution process is succinctly summarized in Gulati's (1995b) affirmative answer to the question 'does familiarity breed trust?' Because strong-tie trust relations can counteract firms' fears of the partner's betrayal of confidence, governing alliances through legal documents yields to relations governed by interorganizational trust. Reduced transaction and monitoring costs make informal social control the preferred cost-effective alternative to both market pricing and hierarchical authority. Consistent with these expectations, Gulati's (1995b) analysis of multi-sector alliances found strong evidence that formal equitysharing agreements decreased with the existence and frequency of prior ties to a partner. Domestic alliances less often involved equity mechanisms than did international alliances, supporting claims that trust relations are more difficult to sustain cross-culturally.

Another crucial developmental issue concerns the relative agency of organizations versus individuals qua persona in creating interorganizational trust. In general terms, network analysis needs to resolve its quandaries about the role of human agency in social action, 'the capacity of socially embedded actors to appropriate, reproduce, and, potentially, to innovate upon receive cultural categories and conditions of action in accordance with their personal and collective ideals, interests, and commitments' (Emirbayer and Goodwin 1994: 1442). Applied to the present context, a central question is whether trust relations operate at the organizational level, or whether trust encapsulates purely interpersonal phenomena? As noted above, some theorists emphasize that trust originates in the social psychology of interpersonal interactions, and thus often evokes strong emotional overtones of sharing and caring for the welfare of one's partner (McAllister 1995). As the employees who occupy key boundary-spanning roles try to cope with their organizations' environmental uncertainties, they socially construct strong bonds of mutual confidence and trust with their counterparts in other organizations that may affect interorganizational behavior. For example, a study of company decisions to switch auditing firms found that the individual attachments of such boundaryspanners as the company's chief executive, financial, and accounting officers attenuated the pressures arising from changing resource needs (Seabright et al. 1992). If only people can manifest beliefs and emotional attachments, then trust may


reside wholly within the individual fiduciaries who establish and nurture trust relations on behalf of the organizations they represent. The potential for intermingling the reputational social capital of people and organizations spawns some knotty dilemmas for intraorganizational control: exactly who legally and morally owns the trust relations in which both employers and employees have invested? This question is not a trivial concern for firms, as reflected by such practices as 'noncompete' clauses restricting local television news personalities from working for rival stations after severing their employment ties, and in law-suits against lawyers and talent agents who defect to rival firms, taking along their client lists (Tevlin 1997). In the most extreme instances of trust violation, agents may pilfer major corporate secrets for their new employers, as in Jose Ignacio Lopez's alleged transfer of General Motors procurement data to Volkswagen.

Alliances Outcomes The belief that interorganizational networks offer corporate social capital in the form of performance benefits superior to both markets and hierarchies is widespread among social scientists and corporate managers. Networks are allegedly 'lighter on their feet' than hierarchies (Powell 1990: 303). They enable organizations and their agents to respond rapidly to emerging contingencies, particularly gaining timely access to swiftly changing technological knowledge and data essential for survival and prosperity. Yet the evidentiary basis for such claims remains remarkably slim. Researchers have proposed numerous criteria for judging alliance 'success,' ranging from mere organizational survival to economic performance levels above industry norms. One difficulty in assessing performance outcomes is that most alliances explicitly seek only limited purposes and are intentionally short lived, so duration alone may be an inappropriate yardstick. When an alliance terminates in one partner's acquisition ofthe other, as in the majority of cases (Bleeke and Ernst 1993: 18), does that outcome constitute a failure of the alliance? A success for one organization but a failure for the other?

Embeddedness in interorganizational alliances seems to contribute to participants' survival chances compared to organizations engaging only in arm'slength market transactions. Uzzi (1996a) used both ethnographic and quantitative methods to study the impact on firm failure of the mUltiple network ties among 23 New York better dress apparel ftrms. 'Social capital embedded ness' indicated whether a contractor had a network tie to a business group, typically formed around CEOs who were kin or colleagues from previous jobs. Other measures involved the proportion of work exchanged between organizations and the degree to which the ego firm maintained arm's-length or embedded ties with partners. Uzzi's logit analyses showed that 'firms that connect to their networks have greater chances of survival than do firms that connect to their networks via arm's-length ties' (1996a: 694). But optimal networks were a mix of both types of relations:

A crucial implication is that embedded networks offer a competitive form of organizing but possess their own pitfalls because an actor's adaptive capacity is determined by a web ofties, some of which lie beyond his or her direct influence. Thus a firm's structural location, although not fully constraining, can significantly blind it to the important effects of the larger network structure, namely its contacts' contacts. (Vzzi 1996a: 694)


Organizations enter alliances with many motives and strategic objectives, including: speed of entry into new product or geographic markets; faster cycle times in developing or commercializing new products; improved product or service quality; gaining technical skills, tacit knowledge and competencies; sharing costs; spreading risks and uncertainties; monitoring environmental changes. Bleeke and Ernst (1993) relied on unpublished reports and interviews with insiders of 150 top companies in the U.S., Europe and Japan to determine that, in 49 cross-border alliances, 51 % were successful for both partners while 33% were mutual failures. Alliances were 'more effective for edging into related business or new geographic markets' (1993: 18) while acquisitions worked better for core businesses and existing areas. Other conditions leading to success included alliances between equally strong partners, evenly split financial ownership of the joint venture, and autonomy and flexibility for the joint venture to grow beyond the parents' initial expectations and objectives.

Empirical evidence regarding the financial outcomes of strategic alliances is scarce, with network studies of investment banking and the stock exchange a notable exception (Eccles and Crane 1988; Baker 1990; Podolny 1993). For example Chung (1996), analyzing cooperative exchanges among 98 top investment banks involved in new stock issues in the 1980s, found that the best long-term performers (measured by amounts underwritten) were involved in a strategy of exchange initiation, which also led to subsequently higher popularity and expanded participation in stock deals. However, few researchers have studied whether joint venture partners recover their capital investments, or whether such collaborations yield a higher returns than available from alternative resource allocations. Theorists tend to emphasize only the social capital emerging from networks, while ignoring potential social liability inherent in interorganizational relations, specifically that social embeddedness may exert a drag on market efficiency. For example, Sako (1991: 239) speculated that a major disadvantage of obligational contractual relations is '[r]igidity in changing order levels and trading partners [and] potential lack of market stimulus.' Similarly, the impact of trust on alliance success remains uninvestigated. Trust presumably fosters goal attainment by facilitating the favorable resolution of conflicts inevitably cropping up during joint operations. Given its subjective basis, high mutual trust is likely to correlate with feelings of satisfaction about the partner's performance and contributions. Researchers might inquire whether collaborators feel their venture is worthwhile and whether they would repeat the alliance for other purposes or to recommend their partner to other firms seeking to form strategic ventures. On the negative side, trust and other obligational norms may attach organizations too strongly to their partners, carrying relations beyond rationally efficient limits by resisting swift dissolution of inefficient or inequitable situations. Clearly many opportunities await for imaginative research on the outcomes of interorganizational alliances. .

INTRAORGANIZATIONAL NETWORKS

The macro-change forces noted above that reshaped interorganizational relations also wreaked enormous transformations inside factories, offices, and clinics. During a prolonged and painful decade of downsizing, reengineering, and restructuring exertions, more daring or desperate corporations implemented flexible new designs.


Rigid bureaucratic hierarchies yielded to experiments in cross-functional teams that devolved increasing volumes of information, technical skills, and managerial responsibilities down to the front-line worker level (Katzenbach and Smith 1993). Employees were prodded to contribute to restructuring decisions by such schemes as job enrichment, quality circles, job rotation, gain-sharing, and stock ownership plans. The social control of organizational performance became increasingly internalized through corporate cultures based on Deming and Juran's total quality management principles, 'a set of powerful interventions wrapped in a highly attractive package' (Hackman and Wageman 1995: 339). TQM emphasized the never-ending collaboration between management and workers for continuous learning and quality improvements, assessment of customer requirements, scientific monitoring of task performance, and process-management to enhance team effectiveness. These high-performance innovations were all intended to lower supervisory costs and increase employees' work-life morale, thereby raising corporate productivity, quality, and profitability (Levine 1990; Lawler 1992).

During the last half century, the implicit employment contract binding workers and firms changed from a virtual guarantee of long-term job security to one emphasizing employability (Cappelli et al. 1997). In the insightful words of Intel Corp.'s vice president for human relations, 'You own your own employability. You are responsible' (O'Reilly 1994: 47). By flattening managerial hierarchies and outsourcing formerly internalized staff functions, firms shortened or eliminated many traditional internal labor markets that had provided career ladders for regular promotions to ever-higher levels of responsibility, prestige, and pay. Instead, jobs evolved from fixed positions into flexible bundles of tasks that were subjected to periodic restructuring to grapple with organizational contingencies in tumultuous world economic markets. Jobs metamorphosed into project-based appointments through which multiply-skilled employees rotated in short-term assignments on their way to newer projects inside the firm or with other employers. Temporary and subcontracted workers became the fastest growing segments of the U.S. labor force by the 1990s (Belous 1989; Parker 1994). The proliferation of computerized communication (Internet and intranets) and production-control systems (CADI CAM), coupled with escalating customer demands for made-to-order goods and services, drove the relentless quest for continual upgrading of employees' technical and interpersonal skills. Firms deployed a multi-track approach, searching for new workers with requisite competencies, training current employees in-house, and forging ties to external vendors of job-training services such as junior colleges (Kalleberg, Knoke, and Marsden 1995).

From these gales of creative destruction emerged a new corporate form-the network organization, whose external alliances were discussed in the preceding section (Miles and Snow 1995). Its distinguishing internal features are multiplex exchange ties among the firm's loosely coupled divisions, departments, work groups, and the individual managers and employees. It breaks down hierarchical and functional barriers, replacing them with task-specific units connected through communication, advice, and trust networks (Krackhardt and Hanson 1993). The


network organization

creates autonomous units, but it increases the volume, speed, and frequency of both vertical and horizontal communication within the organization to promote collaboration . ... The result is an organization with superior performance characteristics for the 1990s. Network management is, in the end, management by empowerment. (Limerick and Cunnington 1993: 61)

Intraorganizational networks operate according to a logic of economically efficient asset allocation. Rather than transferring goods and services by centrally administered prices, the quasi-autonomous units are subjected to internal market discipline when buying and selling resources, thus assuring they will continually seek to improve their performance (Snow, Miles and Coleman 1992: 11). But, corporate networks also function politically and socially in ways that defy strict economic utility maximization principles. In particular, network relations offer employees a prime source of social capital for developing rewarding careers under the new employment contract terms which stipulate greater personal responsibility.

Networks and Career Capital Employees have always used networking activity as important strategies for getting ahead in their companies. A worker's personal networks comprise crucial social capital investments that are as essential for career development as her or his human capital assets of knowledge, skills, and experience. Employees survive and thrive by learning how to construct and manipulate ego-centric networks that provide advantages in the competitive scramble for jobs, project assignments, promotions, and rewards. Networking abilities assume an even greater significance for employees of the new forms of network organizations, where formal positions are ill-defined and perpetually changing.

In a study of senior managers of a computer firm, Burt (1997) examined how social capital affected rapid promotion. He measured social capital as constraints on personal networks, that is, concentrated on fewer contacts. Persons whose networks span more 'structural holes' are well-positioned to broker the flow of information and to coordinate and control interactions between unconnected people on opposite sides of the hole. Not only were managers with less-constrained networks promoted relatively early, but the effect varied with the number of competitiors. The correlation between network social capital and promotion was stronger for those managers with few peers compared to those in positions where many people did the same work. The social capital payoff was higher for people in unique corporate roles 'because such managers do not have the guiding frame of reference provided by numerous competitors, nor the legitimacy provided by numerous people doing the same kind of work' (Burt 1997: 356). Thus, their entrepreneurial networks offered access to more rewarding opportunities.

Research on gender differences in network dynamics sought to explain how personal networks are converted into corporate advantages. In a New England advertising and public relations firm, Herminia Ibarra (1992) found differential patterns of homophily (tendency to form same-sex ties) among the 80 male and female employees. Men tended to concentrated their ties across multiple networks (communication, advice-seeking, support, friendship, and influence) primarily on


other men. Women employees differentiated according to network contents, obtaining social support and friendship from their female co-workers and instrumental access through ties to higher-status men. That is, expressive and instrumental ties coincided for men, but were inversely correlated for women. Consequently, men seemed to receive higher returns than women on their social capital investments, in the form of greater network centrality.

Similar gender-differentiated network propensities occurred among 63 managers of four large corporations (Ibarra 1993a), with men relying more on weak-tie homophilous networks and women forging more strong expressive ties to other women. The relationship between managers' ego-net strategies and their potential for promotion, as judged by supervisors and human resources staff, were also conditional by sex. High-potential women and low-potential men placed greater relevance on expressive networks, such as trust and reciprocity, while high-potential men and low-potential women stressed instrumental ties. Ibarra concluded that women's preferred network strategies placed them at a disadvantage relative to their male peers: The 'entrepreneurial' network pattern characteristic of successful male managers is less effective for females who require stronger network ties to achieve the same level of legitimacy and access to resources' (Ibarra 1993a: 27).

Networks as Power Resources Social power is a structural property of the relationships among actors in a social system, rather than inherent in individuals' formal roles or personalities (Pfeffer 1981: 3; Knoke 1990: 1). Hence, power and political action in organizations are rooted in the multiple intraorganizational networks connecting the participants. Even formal authority that assigns legitimate rights to control corporate human resources, as exhibited in an organization chart displaying supervisor-subordinate positions, should be viewed as just one of several networks conveying political implications, including such informal ties as communication (Pfeffer 1992: 111-125), advice, support, trust, friendship, and horizontal workflow (Brass 1984). Knoke (1990) argued that the primary analytic power relations of every social system are reducible to two basic exchange networks that follow differing logics under which actors affect one another's behavior. Influence occurs when 'one actor intentionally transmits information to another that alters the latter's action' (1990: 3), while domination involves controlling another actor's behavior 'by offering or withholding some benefit or harm' (1990: 4). For example, an employee may induce a coworker's collaboration by persuading the co-worker that cooperation is in the mutual interests of both firm and employee (influence), or by offering resources essential for the co-worker's project (domination). Over time, these informal political relations become institutionalized as the company's intraorganizational power structures, with the employees occupying the dominant and influential positions affecting their less-powerful colleagues' perceptions, cognitions, beliefs, and behaviors. Intra-organizational power structures are highly stable and resistant to change as the persons in power seek to perpetuate their advantages. Structural change occurs mainly as the result of major external shocks, for example, corporate takeovers or technological innovations that drastically rearrange existing political relationships (Burkhardt and Brass 1990: 105).


A core theoretical proposition, derived from resource dependence principles, is that actors who occupy the more central positions in intraorganizational networks can exercise greater political power. While centrality may generate power for an employee, the reverse causal process may also operate over time: people seek to establish connections to the most powerful organizational players, in expectation of enhancing their own power through these contacts. Incumbents in central locations enjoy a variety of advantages over peripheral positions: through their proximity to others in communication exchanges they can acquire more timely and useful information; can better control the flow of resource exchanges and mobilize support for initiatives; can mediate and broker deals between interested but unconnected parties; and, through boundary-spanning ties to external organizational actors, they can direct the organization's strategic objectives (e.g., Kanter and Myers 1991). In short, 'network centrality increases an actor's knowledge of a system's power distribution, or the accuracy of his or her assessment of the political landscape . ... Those who understand how a system really works can get things done or exercise power within that system' (Ibarra 1993b: 494).

As noted above, network methodologists developed alternative measures of network centrality, including in-degree, closeness, and betweenness scores (Brass 1992). Rather than treating all relations as making equal contributions to each person's centrality, a prominence index assigns higher centrality scores to employees who are also highly central within the organization (Knoke and Burt 1983; Bonacich 1987). Thus, network centrality as prominence extends the principle that 'it's not what you know, but whom you know,' to emphasize that your own power depends importantly on 'whether the whom that you know has power.'

A few empirical investigations have uncovered compelling evidence that the attribution of power covaries positively with employee centralities in intraorganizational networks (see overviews by Krackhardt and Brass 1994; Brass 1995b). In a study of communication, friendship, and workflow networks among 140 nonsupervisory employees of a newspaper, Brass and Burkhardt (1992) reported numerous statistically significant correlations between three types of centrality scores and reputations for power (as attributed by supervisors and by peers). The indegree measure proved to be a stronger predictor than closeness or betweenness, suggesting that a large volume of direct contacts may be necessary for coalition formation and also provide the best mechanism for 'learning the network' (1992: 211). Krackhardt (1990) investigated the effects of betweenness centrality and cognitive perceptions of both friendship (trust) and advice-giving networks on the power reputations of 36 employees of a small entrepenurial firm. Controlling for formal position in the company, persons who were more central in the friendship network and who had more accurate cognitions of the advice network were rated as more powerful by others. Neither advice centrality nor friendship accuracy had statistically significant bearings on reputed power.

Finally, network centrality appears to affect some work-related perceptions and activities. Ibarra's (1993b) analysis of which employees adopted problem-solving innovations a New England advertising firm showed that centrality (prominence in five types of network relations--communication, advice, support, influence and friendship) 'was the most significant predictor of administrative innovation roles . ..


and mediated the effects of various of various non network variables on innovation involvement' (1993b: 492). But centrality was not a statistically significant factor in the adoption of technical innovations. Burkhardt and Brass's (1990) longitudinal analysis of computer adoptions in a federal agency also found similar patterns, with early adopters' power and centrality increasing more than later adopters. In further analyses of the advertising firm data, Ibarra and Andrews (1993) showed that advice network centrality and friendship network proximity to varying degrees each affected perceptions of such organizational conditions as risk-taking, acceptance, information access, interdepartmental conflict and autonomy.

CONCLUSIONS

The preceding review of research and theory construction about organizational networks and corporate social capital suggests that we are collectively investigating several critical issues. Researchers are probing the social forces that lead to the formation of intra- and interorganizational ties, their persistence, and their severance. We have fragmentary understanding of how global network structures simultaneously facilitate ('social capital') and constrain ('social liability') the opportunities available to people and organizations in pursuit of their interests. And we now better appreciate corporate social capital as both a generator and an outcome of strategic actions embedded in complex social structures. Still missing is a comprehensive framework to coordinate and accelerate the efforts of numerous scholars toward a more coherent and cumulative research program that could integrate the diverse facets of these elusive phenomena. After decades of network analysis developments, we have abundant conceptual and methodological tools with which to forge such a synthesis.

Two generic tasks should be intensified in tandem. First, researchers should track social capital across multiple levels within and between organizations. At the intraorganizational level of analysis, research designs could examine the concatenation of multiplex relations among employees, work groups, departments, and divisions into complex yet reproducible assemblages that maintain the identity and integrity of the corporation as a bounded social actor. At the interorganizational level, investigators must examine the detailed mechanisms through which social network investments yield individual and collective benefits to alliance members. Until we gain a clearer picture of how relations between firms shape economic and political outcomes, our perceptions of the emergent N-form organizations will remain fuzzy. The second major task for corporate social capital researchers should be to collect and analyze longitudinal data about changing network structures and processes. Current knowledge is cramped by the cross-sectional nature of most research designs. Many methodological advances promise boundless opportunities to expand the temporal dimension of social capital dynamics. We need to learn how seemingly minor changes in specific connections, involving a handful of critical ties, can cascade rapidly through a network, radically transforming its shape and functions. And we need to integrate unique events into the actor-relation dualism, thereby increasing our capacity to capture the historical forces changing corporate social capital.

•

Social Capital of Organization: Conceptualization, Level of Analysis, and Performance Implications 2

ABSTRACT

Johannes M. Pennings Kyungmook Lee

In this chapter we explore the benefits of social capital and the harmful effects of social liabilities. Following Allison (1971), two models of the organizations are juxtaposed: those of the Rational and Political Actors. The issues of social capital require different perspectives when its implications for performance are addressed. The mediation through individuals takes a prominent place in the Political Actor, and moves to the background in the Rational Actor. The issue of aggregation from the member to the organization is primarily an issue when we view the organization as a Political Actor in which the members' social capital aggregates to that of their organizations. Two illustrative cases that fit the two models are then presented, the industrial business groups in Japan and Korea on the one hand, and the popUlation of professional services firms in the Netherlands on the other. In the case of business groups we point to both the benefits of social capital and the drawbacks of social liability. When we shift to the study of professional services firms, we demonstrate that social capital as a distinct organizational resource diminishes the likelihood of dissolution. The implications for social capital and social liability are exposed and reviewed.

INTRODUCTION

Organizations are presumed to have boundaries. They are endowed with various kinds of assets on which they make ownership claims, and which are protected with isolating mechanisms such as patents and contracts. They are liable for their products and services. Also, they have members whose inclusion in the organization is usually beyond dispute. In fact, the firm as a collection of individuals is often


bracketed when considering the competitive game it is playing with other firms. Yet, organizational boundaries are precarious and permeable. Organizations have exchange relationships with suppliers and clients, collude with competitors, and forge all kinds of alliances because they cover only part of the value added in their value chain. In their positioning across the chain they face such decisions to 'make or buy' components and supplies, whether to share or even outsource R&D efforts, or to operate on a stand-alone basis. Their coherence and integrity might decline and bundles of resources often unravel into discrete parts, but these resources might also become combined-for example in divestments and acquisitions, respectively.

Organizations are embedded in a web of relational ties. In the present chapter, the term social capital captures important aspects of this relational web. Social capital of organizations constitutes a distinctly collective property that might be mediated by individuals, yet is uniquely organizational. Social capital complements financial and human capital as assets that are more or less valuable, scarce and imperfectly tradable (Barney 1991). Social capital is even more unique and difficult to appropriate than these other types of assets as it hinges on the continued involvement of two or more parties. Firms, as repositories of unique resources require complementary assets in order to compete successfully. Social capital is crucial in bundling intangible assets and provides the absorptive capacity to merge proprietary knowledge with that of others. Organizations need to coordinate their interdependencies in the value chain and negotiate a position in their industry. By forging external networks, the organization maintains optimal boundary conditions and remains in tune with external trends and events. At the same time, its boundary structures preserve an organizational modicum of identity and protection against erosion of its assets.

The social capital benefits seem beyond doubt; less intuitive might be the cost of social liability. Social embedded ness endangers a firm's appropriability regime, and might also envelop the firm too tightly into a web of ties that stifles its ability to change or impedes its innovative capability. While network relationship is often viewed as conferring various benefits, we should therefore also examine its undesirable consequences.

As numerous chapters in this book indicate, social capital refers to resources inherent in sustained long term relationships and associations. The concept originates in sociology, with two writers standing out: Bourdieu (1980, 1994) and Coleman (1990). In this chapter we extend their representation of social capital by treating it as a unique organizational resource. We will further reflect on the nature of organizations, and ask how such human aggregates or their social organization are capable of possessing social capital. As with human capital, we need to dwell on the tension between individual and organizational levels of analysis. While it is tempting to 'anthropomorphosize' the firm as a human aggregate and impute an ability to mold its surrounding network, we need to ask how such semblance comes about, who the agent is, and what collective motives are operating. After having dwelled on these issues, we explore the implications for organizations of having accumulated social capital. We do so by contrasting two contrasting settings, i.e., business groups and professional service firms, as these stylized forms might respectively illustrate the firm as rational and political actor, and by implication, the

Social Capital of Organization - 45

sort of aggregation issues that color the reason we depict their social capital. Below, we belabor these two metaphors to highlight aspects of corporate social capital. We conclude by spelling out implications and future research opportunities.

CONCEPT OF ORGANIZATION AND ITS SOCIAL CAPITAL

It has not been customary to view organizations as embedded in a network of relationships, although person based networks have been used to describe a firm's external linkages (e.g., Levine 1972). Much of the pertinent literature has focused on individuals (e.g., Burt 1997; Coleman 1988; Granovetter 1985; Uzzi 1997a), their place in some larger network, and the impact it has on their behavior and attitudes. Many views stand in sharp contrast with an 'over-socialized' view of man. Economists tend to couch transactions in personal, self-interest seeking terms. As parties in a market, people engage in 'arm's-length' relationships and their interaction is solely conditioned by the need for exchange. Contrary to a utilitarian tradition, norm theory in modern sociology assumes that people are overwhelmingly sensitive to the expectations of others (Wrong 1961). Sociologists often stress the structural context within which parties meet, and such a context might give rise to a small number of conditions in which actors develop personal bonds, based on trust and mutuality. Uzzi (1997a, this volume) calls such links 'embedded ties.') Within such bounds, utility maximization is often suspended for the sake of preserving reciprocal, even altruistic relationships. The next issue involves the extension from the individual as a party onto himself versus the individual as an 'office holder.' Size also matters; for example a market with single proprietorships entails rather different inter-firm networking than the US banking world in which firms are tied together, for example, through interlocking directorates.

Entrepreneurs, new ventures, and small firms differ markedly from large corporations in terms of the links they maintain. The links that bind them might vary from those that are heavily endowed with trust to those that fit the arm's length relationships. The large corporation is prone to have arm's length relationships with external actors, but as we will see, they often invest in boundary spanning systems in which personally mediated links are discernible. Small firms are more likely to develop bonds of trust and mutual adjustment with external actors such as suppliers and clients, although some conditions give rise to arm's length relationships.

We need to position these distinctions against the 'model' of the firm, which is often implicit (Allison 1971; Simon 1957; Thompson 1982). Organizations have often been viewed as 'rational actors' (Allison 1971) or have otherwise been treated as unitary economic agents. As a singUlar agent, the firm might be embedded in a multiplex web of inter-firm relationships as manifest in contracts, joint ventures, stock cross-holdings, etc. As units with clear legal boundaries and other 'isolating mechanisms' firms complement each other in the value chain. The ties that bind them can be viewed as social capital for coordinating inter-firm activities. If we, however, view organizations as human aggregates, as Allison (1971), for example, stipulates in his organization as 'political actor,' we might attribute to that organizational social capital by virtue of the aggregate social capital of its members. The presumption of firms endowed with social capital appears non-problematic but


the implications are rather different in the two scenarios thus depicted. In this chapter we visit the issue of firm as rational versus political actor in greater detail.

In this chapter, for the sake of the argument, we juxtapose the rational actor caricature with its political actor counterpart and examine the social capital as an integral part of these models.2 In the case of the firm as rational actor, we treat individuals as a component in what often appears to be a multi-layered network; partly mediated by individuals and partly by other linking vehicles. In the case of the firm as political actor, the link will often be personal and fit the characterization of simple tie, based on trust and tacitness.3

We want to stretch the concept of social capital such that it might become an extension of the individual as an office-holder in an organization and, consequently, become an accessory for his firm's functioning. For example, an early study by Pettigrew (1974) on the 'politics of organizational decision making' narrates the position of an information technology specialist as a boundary spanner between his firm and external vendors. As office-holder his significance derives from the quality of internal and external embeddedness. We might then ask whether the office holder's network connections can be combined with that of others into an index of organizational social capital. Furthermore, inter-firm links might also be discernible beyond the IT specialist, for example, by the long term outsourcing of data storage and retrieval services, or the presence of a hot line with the IT consultants. Such a link is not 'simplex,' but what might be called 'multiplex.' The Pettigrew example illustrates the transition from the firm as a human aggregate to the firm as a coherent, singular entity where the issue of aggregation becomes bracketed, or remains altogether outside the purview of the observer.

LEVEL OF ANALYSIS

It is problematic to move from the individual to the organizational level of analysis when analyzing inter-firm networks. The issue of aggregation from the member to the organization is primarily an issue when we view the organization as a Political Actor in which the members' social capital aggregates to that of their organization. Nevertheless, people associated with the organization as Rational Actor carry out deeds on behalf of their firm, and while the model is agnostic about their integrity, we could focus on their role as distinct linking mechanisms as well.

At face value, the individual-collective distinction seems merely conceptual, not 'real.' The issue oscillates between two frames: do individuals as agents or officeholders connect organizations and other human aggregates? Or do organizations and other human aggregates connect individuals? In this chapter, we are mostly concerned with the first type of framing. Nonetheless, we recognize that many interfirm links condition the intermediation of individuals. In abstracting away from individuals as mediators of inter-firm links we shift from the view of the firm as a 'political' actor to that of a 'rational' one (Allison 1971). The level of analysis becomes moot and little need exists for acknowledging cognitive, cultural, or strategic differentiation-whether in the organizational core or at its boundaries.

To the extent that aggregation surfaces as a salient feature, we should abandon the neoclassical notion of the firm as a unitary actor with a well defined preference ordering and whose strategy betrays a clear and unambiguous preference ordering.


Its membership has a singular identity. The challenge for firms is to consolidate divergent identities into a coherent one such that they might even approximate the firm as a unitary integrated actor. The members are assigned to interlocked sets of roles and they develop informal sets of hierarchical and horizontal relationships with other people inside and outside the organizations. A large chunk of organizational social capital exists by virtue of the individuals whose relationships span organizational boundaries.

Some organizational participants are more contributory in their social capital than others, depending on their involvement in the focal firm and its transacting partners. Indeed, not all members are equivalent in their ability to leverage their social capital for the firm. Members vary not only in their contribution to external ties but also in their participation in the organization (e.g., Cohen, March, and Olsen 1972). When aggregating the social capital of members to arrive at a stock index of firms, there is also the issue of redundancy. A network link is redundant if the marginal increase in benefits from acquiring or maintaining that link equals zero. Redundant ties have been well documented at the individual level, e.g., Granovetter's (1995) 'weak' versus 'strong' tie and Burt's (1992) presence or absence of 'structural holes.'

The aggregation of the networks of organizational participants is prone to have redundant contacts. The number of members maintaining contact with representatives of other organizations might produce 'stronger' ties that are particularly beneficial for the transfer of sophisticated knowledge. For the transmission of information or what might be called 'explicit knowledge,' such strong ties are hardly efficient (compare Hansen 1997). Furthermore, not all social capital of members aggregates to the social capital of the organization. The social contacts of certain organizational members may have little or no instrumental value for their organization.4 Only overlapping membership in groups and organizations, that are operationally or strategically relevant, matter when aggregating individual social capital to that of the organization; the most common example is interlocking directorates (Pennings 1980; Stokman, Ziegler, and Scott 1985).

Boundary Spanner or Multiple-Group Membership The concept of overlapping membership as a way to represent an individual's social capital should also be invoked to revisit the issue of a firm's boundaries. If members vary in their inclusion in the focal organization, their external contacts should vary in value as well. Even if organizational members have valuable external ties, they become a valuable component of the firm's social capital only if the members enjoy access to certain peers-for example, those with power, information, and other resources. If inclusion is highly partial, their individual social capital becomes marginalized for the firm as well.

For simplicity's sake, organizational members might be stratified into a core group, a regular or associate group, and temporary or marginal workers. The core group consists of essential employees who are long-term employees and owners. Their fate is usually tied to that of the organization. The regular or associate group consists of rank-and-file employees who have been involved in the organization for 'some' time and face good prospects to join the core group. Many members who


participate in that tournament will 'plateau,' become sidetracked or might even be terminated, however. The temporary or marginal category include temporarily hired workers and employees of sub-contractors, i.e., workers who fill the jobs not requiring firm-specific skills and who have little chance of moving into another category of members.5

It follows that the social capital associated with the core group is more important for the organization than that of the regular group. The reason is two-fold. First, members in the core group are more likely to use their social contacts on behalf of the organization. Consistent with the garbage can model (Coh~!1, March and Olsen 1972), these members have the highest 'net energy load,' as their fate is closely tied to that of the organization. Second, they are likely to maintain more valuable social contacts for the organization. They are more central to the access structure, and enjoy higher positions with more power and authority. Many of the firm-relevant social contacts are based on the job and title of individual members. A CEO becomes a board member of a peer organization, supplier or some other organization; a partner in a consulting firm befriends senior executives in the firm he works for, etc. Compared to the employees in regular or temporary groups, members in the core group tend to have social ties with people who occupy higher, more visible and more prestigious positions in their organizations. In other words, people who have social contacts with members in the core group of a focal organization tend to have more valuable resources at their disposal for the focal organization than do the people who have primarily social ties with members in its more peripheral ones. Core members also stay longer with their organization such that their organization stands to benefit more from their social capital. Overall, we need to focus on the nature of the employment relationship to weigh an individual's ability to link his firm with other ones.

Figure 1 provides a graphical display of organization stratification in terms of magnitude of personal inclusion.

There are also other ways to compartmentalize the firm as a community of people who are endowed with human capital, and who are differentiated by skill, function, types of markets, products, or technologies. Firms have either a functional

Marginal: free agents, sub contractors, contingent workers, strategic alliance guest employees

Associate: temporary employees, 'in-transit' workers

Core: long term employees, partners, owners, managers, residual claimants

Figure 1. Stratification of firms based on partial inclusion of their members


or divisional (and in many cases some hybrid) structure whose boundaries define identities. In fact, although finns proclaim to be a hierarchy that economizes on transaction costs (Williamson 1975), they in fact comprise numerous sub-cultures, with their own identity and parochial interests. While hierarchy and lateral linkages integrate disparate units, they often face major hurdles in consolidating their skills or knowledge, or more generally in bundling their contributions to the common good (Brown and Duguid 1997; Kogut and Zander 1996). A firm's internal networks such as heavy duty project managers (Clark and Fujimoto 1991), overlapping teams, and interdepartmental career paths become vehicles for knowledge migration, but such networks are often comparatively deficient because specialization impedes knowledge transfer, especially knowledge that is difficult to package. Ironically, communities of knowledge within the firm have often easier access to likecommunities in other firms than they do with the sister departments within their own firm. The implication is that such external networks are often more efficacious in bridging the firm with external actors than do networks that embrace the total organization. By way of example, we might consider a firm's participation in an 'invisible college' less problematic than its participation in a trade association (Powell 1990; Lazega, this volume).

Multiplex versus Personal Forms of Organizational Boundaries At the level of inter-organizational relationships, we could make an even stronger argument about the individually anchored social capital of organizations. When the vendor of a software firm leaves, he might appropriate the connections with clients that he has built up during his tenure. One might thus argue that the social capital of organizations is tied up in the individuals they employ.

Yet, as with all intangible assets, social capital can also be treated as an intangible asset that is not exclusively buried in personal networks. Social capital is often 'depersonalized' or is couched in mUltiplex forms. Interorganizational links established through individuals might begin to lead a life of their own. Or such links become embellished by other glue such as contracts, traditions, and institutional arrangements. The members who are then a complement to a system will in fact also be governed by the norms and beliefs that are endemic to local social arrangements. When links become multiplex, they cease to be dependent on individuals who act as brokers. By way of examples, patent citations signal proximity of knowledge among organizations and can be examined as a conduit for inter-firm knowledge transfer. Cartels amount to a clique with shared norms where the members are firms rather than people. A set of firms might be tied through mutual share holdings. Affiliation among organizations, such as keiretzu in Japan, chaebol in Korea, or business groups in Sweden illustrate bundles of inter-firm connections that cannot be reduced to middleman-members.

Strategic alliances such as joint ventures, R&D partnerships, and minority investments embody nodes in webs of inter-firm networks in the telecommunication, micro-electronic and biotechnology industries (e.g., Ajuha 1998, Hagedoorn and Schakenraad 1994; Omta and van Rossum, this volume). Severing some of these linkages might be impossible. For instance, Microsoft has extensive lock-in agreements with PC makers and their suppliers and PC manufacturers in fact have


contracted for the pre-arranged installation of Microsoft's operating system in what used to be called 'IBM-compatible' personal computers. Biotechnology firms' entrenchment can be inferred from patent citation networks in which their intellectual property is more or less linked with that of other firms; the tightness of their links is derived from the proximity as measured by relative citation frequencies (Stuart, Hoang and Hybels 1997, see also the chapters in this volume by Stuart and Smith-Doerr et al.).

All of this requires us to dissect the ingredients of inter-firm networks into at least three categories:6

Link Any sort of association between two or more firms, including equity cross-holdings, patent-ties, licensing agreements, R&D partnerships, equity joint venture agreements, gatekeepers, or interlocking directorates.

Ties Human mediated links, such as interlocking director or guest engineer. Ties can be 'neutral,' reflexive (Pennings 1980) or even universalistic versus parochial and particularistic.

Relationships Human mediated ties that are particularistic, as for example the guest engineer who has an OEM employment status but resides on the premises of a supplier.

In short, corporate social capital bifurcates into personalized and depersonalized forms, with relationships often augmented with ties and links; while in other instances, the link might persist without the benefit of a relationship. This distinction often corresponds to a simplex versus a multiplex web of network connections. Multiplex 'links' appear to be more congruent with the rational actor metaphor of Allison, while 'relationships' feature prominently in treatments of organizations as political actors. Table 1 furnishes some examples. First, the organization itself can have a link with other organizations that is instrumental for its functioning. Affiliation among organizations, such as keiretzu in Japan or chaebol in Korea is a social link of the organization itself rather than of organizational members. As a legal entity, the firm is capable of contracting, of acting as a partner in any market relationship, induding the setting up of joint ventures, the acquisition of another firm, or the shedding of a business unit to other firms, etc. Indeed, independent of their members, the organization often maintains social capital through the repetitive exchanges with other organizations. The pattern of exchanges has stabilized, even if the individual members who participate in the process have been changed (Chung 1996). Investment banks perpetuate their collective efforts when they syndicate public offerings (Chung, Singh, and Lee 1995). Semiconductor firms joined SEMA TEe when they sought to acquire greater economies of scale.

Whether one assumes a personal or impersonal link (or a hybrid form comprising both links and relationships) between organizations, links constitute the ingredients of arrangements that govern the firm-environment interface. In some cases the arrangements can be viewed in their own right, but their efficacy in


.........................................

...

Figure 2. Boundary transaction system comprising four individuals among two organizations

managing external dependencies depends critically on the quality of the relationship with internal and external decision makers. Adams (1976) was one of the first writers to review such arrangements. He refers to so called 'boundary transaction systems.'

Boundary Transaction System Social capital fits with the notion of more or less permeable boundaries of organizations that become spanned by a 'boundary transaction system' (Adams 1976). Figure 2 provides a graphic representation. As Table 1 indicated, such systems diverge into pairs of individual dyads such as the interlocking director or guest engineer whose role in maintaining the firm's network depends critically on a balanced overlap between the inside and the outside. Or boundary transaction systems are larger and more elaborate entities-for example kereitzus and R&D partnerships. In the latter case the inter-firm link is not nearly as dependent on the presence of boundary-spanning individuals such that the significance of their mediation is comparatively minor. The personal ties often complement non-personal ones such as reciprocal ownership arrangements and R&D partnerships.

Furthermore, the relative salience of the system hinges on the duration of links that are maintained by individuals that make up the system. The longer the tenure,

Table 1. Examples of social capital among organizations

Mediated by Individuals (Simplex)

Interlocking directorates (Pennings 1980)

Guest engineers (Dyer 1996a)

Social register (Useem and Karabe11986)

Revolving door syndrome (Pennings, Lee

and Witteloostuyn 1998)

Alumni (McKinsey)

Double agent

Gatekeeper (Tushman 1978)

Emissary

Mediated by Systems (Multiplex)

Business groups (Acevado et al. 1990)

Chaebol (Kim 1997)

Keiretzus (Gerlach 1987)

Investment bank syndicates (Chung, Singh, and Lee

1995)

Joint ventures

R&D partnerships

Guanxi-chia-jen (Tsui and Fahr 1997)

Electronic clearing house (Pennings and Harianto

1992)


the more distinct the boundaries of the transaction system and the greater the likelihood that its members 'go native,' i.e., acquire an identity almost different from the firms they span. Consider boards of directors, or executive councils of Japanese business groups who over time might become closely knit teams. Employees originate from leading universities, where they have already formed friendship networks, and synchronically move upward through equivalent organizational ladders, such that the 'old boy network' remains intact from university years until retirement. The implication is that succession patterns further strengthen the boundary system's identity (Yoshino and Lifson 1986).7

The boundary transaction system is useful in that it points to the role of member's social capital in producing organizational social capital. Likewise, by recognizing that the system often evolves into a system that cannot be reduced to the participating members, social capital might become depersonalized. The system might become part of a business group, cartel, a joint venture, a long term licensing agreement, or R&D partnership. Such systems are bound to become semifreestanding entities when three or more firms decide to participate. For example, SEMA TEC and ESPRIT are consortia of semiconductor firms that joined forces at the behest of the US and European Union governments respectively to create what we might call a boundary transaction system.

A key difference between a simplex and multiplex boundary system involves the notion of trust. In a simplex system, trust is anchored in a dyad of trustor and trustee who maintain a form of personal trust of what Simmel calls 'mutual faithfulness.' Bradach and Eccles (1989) refer to expectations that the other side will not behave opportunistically. It accords with the definition of trust by Mayer, Davis and Schoorman (1995: 712)-a willingness of a party to be vulnerable to actions of another party based on the expectations that the other party will perform a particular action important to the trustor, irrespective of the ability to monitor or control the other party. This definition excludes the social context of the dyad.

In multiplex systems, the social context becomes central and will in fact color the nature of the relationships between individuals who are part of that system. The context includes not only traditions, ties inherited from individuals who are no longer present, contracts and financial leverage, but also forms of institutionalized trust (Luhmann 1979; Shapiro, Sheppard and Cheraskin 1992; Zucker 1986). The institutionalization evolves both temporally and spatially. Firms have often recurrent contacts with other firms, and the history of their relationship provides a platform for the current boundary system. Firms are also entrenched in larger entities, most notably business groups. The firms that make up a business group share norms about inter-firm transactions, have developed routines for contracting, and enjoy a groupderived reputation that molds the dynamics of interpersonal relationships within a boundary transaction system between two member firms. And history matters here, too: the member firms have collectively gone through actions that resulted in shared practices, mutual stock ownership, exclusive supplier-buyer relationships, or investments in transaction specific assets (Dyer 1996a). The historical and spatial context for two individuals who span their respective firms is therefore critically important in comprehending corporate social capital.


The fact that building up social capital requires time was nicely illustrated in the recent difficulties between Ford and its suppliers. Ford sought to redesign its Taurus model, while at the same time redesigning its boundary transaction system (Walton 1997). For example, the firm attempted to move from multiple, arm's length ties with suppliers to single source relationships. Having made few investments in social capital, its 'relational competencies' (Lorenzoni and Liparini 1997) for managing such supplier relationships were grossly inadequate. The boundary system included individuals such as Taurus project managers and representatives from 235 suppliers. The project's social architecture was to embrace a Japanese-style long-term cooperative relationship with suppliers. Yet, the culture of the system could be described as 'You could not trust them.'

The boundary transaction system should not be confined to individuals who gave rise to the system or were involved in its perpetuation. It ranges from dyads of individuals to complex social, economic, and technological arrangements. It evolves from individuals who interact frequently so that the firms become familiar with each other. Familiarity alleviates transaction costs, improves coordination across organizational boundaries, and reduces agency problems-in short the familiarity that comes with organizational networks confers benefits. Familiarity also produces group-think, cuts the firms off from important external stimuli, and renders it increasingly inflexible. More specific benefits of social capital and the harmful effects of social liability are discussed next.

PERFORMANCE IMPLICATIONS OF SOCIAL CAPITAL

At the onset of this chapter, social capital was mentioned as an integral part of the organization'S intangible assets. The reference to assets suggests a rent producing potential. However, social capital as such cannot produce rents, but it contributes to greater rent maximization of other resources that complement social capital.

Burt (1992) points out that social capital is owned jointly by the parties to a relationship whereas financial and human capital are the property of individuals or firms. In other words, social capital is embedded in the positions of contacts an organization reaches through its social networks (Lin, Ensel, and Vaughn 1981). Second, social capital is related to rate of return in the market production function whereas financial and human capital pertain to the actual production capability. We should ask: What is the role of social capital in economic transaction? Under perfect competition, social capital cannot generate any economic rents (Burt 1992). The market however is hardly perfect and information is not costless. The member's social capital strengthens his firm's ability to retain clients, perform market intelligence, and learn about new technologies. This is particularly true in our knowledge economy where many industries are characterized by abstract products or services, whose quality and other dimensions are difficult to articulate and where delivery of output is highly coupled with reputation (cf. Burt 1992). Clients resort to their social contacts to screen their suppliers because assessment criteria for quality might be hard to come by. While social capital is not part of the production function, it has profound impact on the benefits that firms derive from their productive capabilities. Putting it differently, social capital brings the opportunities to exploit financial and human capital at a profit.


In the next two sections, we belabor these implications by reviewing two examples with rather different manifestations of interfirm ties: industrial firms that make up business groups such as Keiretzus and Chaebol and professional services firms that comprise the audit industry. We have hinted that these two examples present different manifestations of a firm's external networking. Firms that belong to a business group are typically depicted as ('rational') actors in a conglomerate-type setting with mutual equity ownership, long term supplier-buyer transactions, and shared directorships. The relational structure that business groups have is assumed to furnish social capital to member firms. We impute such benefits to the firm without confronting aggregation issues or delving into internal factions. Individuals are merely one of the threads that make up the fabric of networks of business groups. Thus the member-firms of business groups are depicted as integrated, unitary actors who might benefit from their inclusion. The groups furnish interesting data on the benefits of social capital and costs of social liabilities among firms that come close to the stylized Allison-type Rational Actor.

In contrast, professional services firms belong to a sector that resembles a cottage industry, where individual professionals appear to be the most salient participants. While many professionals join a partnership and thus become coowners of the firm, these organizations are very flat and by dint of the professionalization comprise members whose loyalty might be as strong to their firm as it is towards the profession. The social capital of the firm might in fact be the social capital of individual professionals. Even if we aggregate their social capital to that of the firm they belong to, there always remains the issue as to whether it is the partner, his peers as co-owners, or his firm who can make claims on the social capital that is mediated by the professional. The professional has his own roster of clients and might feel more loyal to those clients than to his brethren with whom he makes up the partnership. His ties, and by implication his firm's links, often fit the notion of embedded ties. Arm's length transactions are incompatible with the rendering of services, although some emotional distance with the client is often deemed appropriate. Since partnerships often break-up, or witness an exodus of partners, the caricature of Allison's political actor might sometimes be quite appropriate as a general descriptor. Yet, as we will see we often have to qualify this caricature.

Business Groups Social networks have been a pervasive feature of Asian socIetIes in general. According to Hofstede's (1980) landmark study, Asian societies stress collectivist values and cherish loyalty and commitment to family, organization, and community. At the corporate level we also discern a preponderance of networking-most visibly in business groups. Business groups include Japanese keiretzus, or their pre-war predecessors, called zaibatzus, and Korean chaebols. These groups contain a myriad of firms held together by ownership links, supplier-buyer relationships and mutual guarantee for each other's bank loans. Other countries, most notably Sweden (e.g., Hakanson and Johanson 1993; Sundqvist 1990; Berglov 1994) and Argentina (e.g., Acevado et al. 1990), harbor business groups, but take on a local, idiosyncratic form. Therefore it is prudent to limit ourselves to a relatively homogeneous class of cases


(cf. Guillen 1997). Furthermore, some other Asian countries manifest distinct forms of social capital among organizations; we could mention bamboo networks that are depicted as a guanxi (relation)-based cluster of Chinese firms (cf. Tsui 1997; Weidenbaum and Hughes 1996). In these cases, the individual as family member performs a primary role in forging inter-firm links, and the family rather than the firm appears to be the most salient unit of analysis. Unlike more centrally coupled business groups in Korea and Japan, these Chinese forms of organization are octopoid and opportunistically diversified (Tam 1990). In this section, we restrict ourselves to keiretzus and chaebols.

Chaebols Korean business groups manifest several features that set them apart from Westernstyle business groups (Kim 1997). They display family ownership and management, controlled by a powerful chair. The chair's power derives from stockholdings and from being the father or senior family member who are heads of member companies. Kim (1997) even refers to unquestionable filial piety and patriarchy based family control within modem multinational firms. A founder ' s descendents actively participate in the top management of the chaebols. When the founder dies, his descendents succeed as heir. When the founder with multiple descendents dies, 'his' chaebol sometimes divides into several mini-chaebols as the case of Samsung indicates. Still, the kinship and family networks link the member firms of those mini-chaebols.

Chaebols also exhibit high flexibility in mobilizing financial capital, technology and human resources. Unlike keiretzus and zaibatzus (although the same Chinese character is used to denote this extinct type of Japanese business group as well as chaebol!) that are governed through consensus building and psychological commitment, chaebols are nimble in their deployment of resources and the patriarch can implement strategic decisions without consulting others. There is widespread rotation of key personnel, R&D efforts are pooled across companies and transfer of cash can be arranged through financial services firms, and the member companies can guarantee each other's borrowings from financial institutions.

Finally, the complex set of networked firms that make up a chaebol are exceptionally broadly diversified. Kim (1997) shows that a chaebol like Samsung operated in light and heavy manufacturing as well as in financial and 'other' (e.g., construction, media, hospitality, and advertising) services. Presumably, such diversification allowed chaebol to offset lack of high-tech skills by exploiting semiskilled and unskilled labor in a way that would not be feasible to a non-networked competitor (Amsden 1989), while at the same time produce products that are pricecompetitive rather than quality-competitive in the global markets. Compared to keiretzus, chaebols are basically shaped on the basis of the founding family. The financial institutions are less utilized to form the relationship within a chaebol's member firms, because chaebols are blocked from owning more than 8 % of shares in commercial banks.

Keiretzus Chaebol should thus not be confused with keiretzus or even with their name sake zaibatzus although the degree of contrast is a matter of controversy. After the second


world war, zaibatzus were dismantled but reappeared in a different form called 'keiretzus.' As a result of the transformation, the founding families of zaibatzus lost their shares and power and thus were no longer a source of connections. The insurance companies are at the keiretzus' apex, and from them cascades a transitive pattern of equity cross holdings-the implication being that the insurance firms and their executives are the ultimate center of power and influence (Nishiyama 1982). Keiretzus' governance is much more decentralized with decision making among firms by consensus rather than through fiat by the keiretzus' insurance firm's executives. The zaibatzus provided a template and became mimicked by Korean entrepreneurs and in any event evolved into a prominent form during Korea's industrial revolution. Zaibatzus and chaebols share characteristics such as family ownership, management by patriarch, and unrelated diversification. However, unlike the chaebol, the zaibatzu also controlled commercial banks, giving them access to capital markets.

Keiretzus are laterally federated with transitive stock ownership arrangements that induce minimal interference in between-firm interactions, rather than resembling a chaebol-like holding with a vertically arranged governance structure.s Gerlach (1987: 128) refers to them as 'business alliances,' which he defines as the 'organization of firms into coherent groupings which link them together in significant, complex long-term ownership and trading relationships.' They are distinct in the manner in which they have established coordinative mechanisms to govern their relationships. These include high level councils of executives, the shaping of exchange networks, and the external presentation as a coherent social unit, for example, through advertising and product development activities. Prominent, but largely invisible in the structuring of network links is the role of financial institutions, which unlike the chaebol are an important component of the Japanese style alliance. The member firms are heavily indebted to the keiretzu's main life insurance company and bank. The cross equity holdings constitute an important link over and beyond the relationships that could be uncovered if one were to have access to their inner circles. Unfortunately, no research exists on the power structure within such circles, and the sort of collective decision making processes that ensue. Thus we are also deprived from making strong conclusions regarding the stock of social capital among keiretzus firms. These links are not merely leverage tools, but in fact might acquire a significant symbolic meaning on their own and complement other media of networking such as exclusive R&D projects. The keiretzu as a somewhat hierarchical network is therefore multiplexdebt holdings, cross-equity holdings, supplier-buyer links, and personnel bonds are part and parcel of the connections that bind the firms into a tight and relatively unified alliance.

Social Capital of Business Groups Firms that are part of chaebols and keiretzus are presumed to benefit from the social capital that ensues from their membership in these alliances (e.g., Kim 1997; Lawrence 1991). Social capital is manifest in two ways: First, business groups provide its member firms with access to resources from other firms. As a quasiholding or federation of businesses, they can furnish superior access to financial


capital through a member financial services firm and cross guarantee each other's bank loans. Similarly, business group specific suppliers and their Original Equipment Manufacturers belonging to the same group display shorter lead times in new product development because they circumvent transaction costs, for example, by making significant asset specific investments that in the absence of a business group context would incur significant hold-up problems (e.g., Dyer 1996a; Gerlach 1992b). The inclusion in the keiretzu reduces the outsourcing to one or at most two suppliers, and the relationship is typically based on trust and mutuality. By way of contrast, Toyota relies often on a single, keiretzu-anchored supplier, while US auto manufacturers such as GM usually rely on as many as six suppliers, with whom they interact opportunistically and at arm's length (Dyer 1996a; Nooteboom, this volume). The suspension of the hold-up problem results also in joint R&D and in the geographic clustering of OEM and their suppliers, thus economizing on value chain coordination costs, transportation distance, and inter-firm transfer of tacit knowledge (e.g., Hansen 1997; Nooteboom, this volume).

The social capital of business groups, however, is not confined to intra-group relationships. Since their boundaries are also salient at the group level, they have enjoyed scale advantages, not unlike those accorded fully vertically-integrated firms. Such assertions question the saliency or distinctiveness of boundaries, and in particular the issues associated with vertical integration, governance, and transaction costs (Williamson 1996; Powell 1990). Even though inter-firm links are not exclusively mediated by individuals-as we have argued they are mUltiplex, to say the least-the links that bind them might be so strong that the focal attention often shifts to that level of analysis when discussing social capital. They maintained levels of flexibility in moving around human resources and other assets, and because of superior access to cheap and unskilled labor, were able to claim cost leadership positions in their world of multi-point competition. (Kim 1997: 180-195). Yet, on the next higher level of analysis, these groups commanded clear benefits that surpassed inter-firm arrangements, as reviewed by Powell (1990).

Empirical Evidence In Korea, there is the often documented 'cozy' chaebol-government interface. Chaebols as groups are often endowed with a good deal of social capital because of the support they have extracted from the South Korean government. Compared to non-chaebol firms, chaebols have had better access to state-controlled resources, and were thus able to exploit governmental powers for their own benefit (Kim 1997). The chaebol dominated segment of the economy grew much faster than the economy as a whole.

The reasons that chaebols have received a great deal of governmental support are two-fold. First, the sheer size of chaebols has made them very important for the development-oriented Korean government. For instance, the value added by the 30 largest chaebols has been around 15 % of GNP and their sales volume has been around 80 % of GNP (Cho, Nam, and Tung 1998). Since chaebols have been used as a tool for the government's industrialization policy, the Korean government has provided a great deal of favors including soft loans, import prohibition, tax breaks, etc. Second, the relationships of elite university graduates strengthened the


relationship between the Korean government and chaebols. People who graduated from elite universities have occupied major positions in the Korean government, banks, and parliament. As a result, chaebols appointed elite university graduates as CEOs to lubricate their relationship with external entities. For instance, 62% of CEOs of the seven largest Korean chaebols in 1985 graduated from Seoul National University (Steers, Shin, and Ungson 1989).

There is also some provisional evidence that member firms within a chaebol or keiretzu might encounter the adverse effects of 'over-embeddedness.' In Korea we have the case of the Kukje chaebol and recent bankrupcies of major chaebols, while in Japan the differential learning of keiretzu versus non-keiretzu suppliers provide testimony to the harmful effects ('social liability') of social embeddedness. The Kukje case emerged in February 1985 and evolved from an ordinary bankruptcy into a scandal when the Chun government disbanded the chaebol due to 'reckless management, and exceedingly high debt rates.' It is most relevant for our argument because of 'nepotic management by the sons of the founder' (Kim 1989). The bankruptcy case is somewhat ambiguous and opinions varied as to whether it was over-embeddedness among member firms or deficient external social capital that accounted for the disbanding of Kukje. Yang, the chaebol president, claimed favoritism on the part of the Chun government. In any event, further research should identify whether it was social capital at the group level or at the group-state level that explains the demise of Kukje.

Due to the risk-sharing role of chaebols, Korean chaebols enjoyed very high survival chances and thus only a few chaebols experienced bankruptcy. During the period of January 1997-January 1998, however, nine chaebols among the 30 largest chaebols experienced insolvency. The mutual guarantee of bank loans made whole member firms rather than some of them insolvent. In some cases, the failure of one member firm became the reason of the bankruptcy of the chaebol. Overembeddedness to other member firms rendered profitable and financially sound member firms bankrupt, thus revealing the 'dark side of social capital' (Gargiulo and Benassi, this volume).

Keiretzus in Japan also function as a tool for risk sharing among member firms (Nakatani 1984) and thus they enjoy a lower bankruptcy rate (Suzuki and Wright 1985). However, criticism has surfaced regarding their traditionally claimed advantages. Gerlach (1992b) sees the potential unraveling of keiretzus now that their benefits have appeared to wane. Nobeoka and Dyer (1998) have recently completed a survey of OEM-automotive supplier relationships and produced evidence indicating that suppliers that diversify away from a single keiretzu based OEM are more profitable compared with firms who are locked in a close single-source relationship. They interpret this finding as being due to either superior bargaining power, or to a broader exposure to technological know-how; such firms diminish their dependence on a single OEM or they witness learning benefits in that their know-how is likely to be more generic and less firm-specific.

Similarly, Lincoln, Ahmadjian and Mason (1997) provide evidence of Toyota the auto manufacturer and Toyota the keiretzu member, which diversified away from keiretzu-based automotive suppliers. These authors report that intra-keiretzu knowledge was not only limited, but that Toyota did not even attempt to elevate its


'internal' suppliers to the standard that would meet its needs. The implication is that, in spite of trust and inbred capabilities, the firm begins to question the benefits of traditional arrangements. Such precedents might lead the keiretzu on a path of further unraveling its stale social capital and the substitution of a fresh one.

Summarizing, business group's endowment of social capital should be differentiated into that social capital that is discernible at the group level versus that which resides at the interface between the business group and external actors. The beneficiary of social capital is the firm or a group of firms who are portrayed as unitary actors, operating in their economic-political arena. The evidence so far has focused on the social capital inherent in social structure, but more recent evidence shows also that over-embeddedness might lead to social liability.

Audit Firms The accounting sector presents another setting in which the costs and benefits of social embeddedness are evident. Unlike markets with industrial firms, as is the case with industries comprising business groups, the accounting sector produces largely intangible and abstract services. The measurement of product quality is elusive, the production flow is exposed to the client who is often an active co-producer of the services rendered. The firm has some degree of hierarchy but is usually much flatter. In fact most firms are stratified into partners (i.e., owners) and employees, some of whom expect to join the partnership. Their close exposure to the market place and their intense involvement with clients makes social capital a central feature of operations and a key driver of organizational performance. This sector resembles numerous cottage industries where personally mediated ties predominate, not unlike the settings of garment district members (Dore 1983; Uzzi 1997a), or investment bankers (Burt 1997).

Ironically social capital can be viewed as a substitute for objective criteria of quality, reliability and consistency. In the absence of objective, verifiable and measurable product attributes, clients might rely on their networks to select auditors or to remain loyal to them even after the honeymoon period has passed (cf. Podolny and Castellucci, this volume). The endowment of social capital is therefore a critical resource in such sectors. Absent social capital, the firm might not extract much rent from its human capital. Furthermore, social capital allows the firms to leverage their human capital thus extracting more quasi rent from that asset. Social capital is not only valuable as rent producing potential, but is also scarce and difficult to appropriate. These aspects suggest social capital as a resource not unlike brand equity, reputation and goodwill, and should be further explored here.

As we indicated at the onset of this essay, social capital fits Barney's (1991) criteria of the resource-based-view of the firm. Resources that provide a competitive advantage should be valuable, rare, hard to imitate, and imperfectly substitutable. Applying these conditions to accounting firms and other professional service sectors, it appears therefore obvious that the social capital of an audit firm forms a major source of competitive advantage in this 'knowledge' sector. Social capital of audit firms has a rent-producing potential, in that it is valuable and scarce (product market imperfectness) as well as imperfectly tradable (factor market imperfectness). Araujo and Easton (this volume) employ a similar list when they conceptualize


social capital through a 'relational' lens. Let us review these aspects of social capital in closer detail in order to reveal their role in explaining the benefits of embedded ties.

Valuable As far as the value argument is concerned, a substantial number of studies in sociology have shown that social ties transfer influence and information (e.g., Burt 1992, 1997; Coleman, Katz, and Menzel 1966). At the individual level, the benefit of having supportive relations has been welI established. Supportive relations contribute to getting a job (Granovetter 1995), high compensation (Boxman, De Graaf, and Flap 1991), and promotion (Burt 1992). We argue that this argument pertains to the (audit) firm level as well. Burt and Ronchi (1990) and Burt (1992) applied the notion of social capital to organizations. Burt (1992: 9) pointed out that 'the social capital of people aggregates into the social capital of organizations.' Social capital amassed in the organization's members is among the firm's most valuable productive assets (Burt and Ronchi 1990). Unlike the setting of business groups, in this sector we can define an organization's social capital as the aggregate of the firm members' social capital. An individual member's social capital is captured by his connectedness with client sectors.

Why would audit firms with social capital enjoy competitive advantages and higher survival chances? That is, what is the role of social capital in the economic transaction of providing audit services? Under perfect competition, social capital cannot generate any economic rent (Burt 1992). However, the market for auditing services is hardly perfect, and information about audit services is not costIess. The owner's social capital strengthens his firm's ability to retain and attract clients. This is even more true in the audit industry, where information with respect to qualities of professionals is hardly perfect (cf. Burt 1992; Polodny and Castelluci, this volume). Clients resort to their social contacts to screen their service providers, because assessment criteria for auditing quality are hard to come by. Crucial contacts include those that involve the client sectors that an audit firm serves. There are three reasons why network ties with client sectors may well facilitate the building and retention of clientele.

First, people tend to rely on their current social relations to alleviate transaction cost (Ben-Porath 1980). A stranger who does not anticipate an enduring exchange relationship, has an incentive to behave opportunistically. To curb this malfeasance, ill-acquainted exchange partners typically rely on elaborate, explicit, and comprehensive contracts. These contracts, however, are difficult to write and hard to enforce (Williamson 1975). Mutual trust between the actors, developed through repetitive exchanges, obviates the need for writing explicit contracts. If the creation of trustworthy social relations were costless, however, the existing network ties would not confer benefits to those who nurtured them. In reality, individuals and organizations have to invest substantial time and energy in forging durable relations with others (Burt 1992). Variations in networking among firms should then contribute to differences in the firms' ability to attract clients. Second, trustworthy relations produce information benefits for the linked actors (Burt 1992). Information is not spread evenly across all actors. Rather, its access is contingent upon social


contacts (Coleman et al. 1966; Granovetter 1985). An actor cannot have access to all relevant information, nor can he process and screen all important information single-handedly. Being embedded in a network of relations allows a particular actor to economize on information retrieval. Second hand information, at least, serves to signal something to be looked into more carefully (Burt 1992). Personal contacts also make it possible for the involved actors to acquire the information earlier than others. Third, trustworthy relations enhance the possibility for an actor to refer his contact person ( for example, an auditor, physician or management consultant) to a third party (i.e., 'tertius'). Burt (1992: 14) puts the benefit this way: 'You can only be in a limited number of places within a limited amount of time. Personal contacts get your name mentioned at the right time in the right place so that opportunities are presented to you.' The counterpart in a dyadic relation can playa role as a liaison to link the social actor to third parties.

Scarcity The argument as to the scarcity issue is, again, specific to the CPA profession. The CPA profession is there to attest financial outlets of organizations. In effect, this was the very reason for the origination of the profession. In away, this is comparable to other public professions. For example, police officers are trained to perform their public, and legally protected, role of preventing and bringing action against violations of the civil order. In a similar vein, CPAs are expected to prevent and bring action against violations of the 'financial order.' Therefore, CPAs are trained to perform their public attesting role-this is the core of any CPA education program. This very nature of the profession implies that the majority of CPAs are employed in public practice, working within audit firms rather than client organizations. Only a minority is attached to internal control jobs within client organizations. Hence, social ties that come with current (or previous) partners or associates with previous (or current) employment outside the audit industry - i.e., through jobs in governmental bodies or private enterprises - are not abundant. For example, in 1920 roughly 80 % of Dutch CPAs worked in public practice. In the period from the 1960s up until the 1980s, this percentage dropped to slightly above 50 %. Hence, there is much room for audit firm heterogeneity in this respect, both in time as well as over time.

Nontradability Apart from market imperfection (resource value and scarcity), nontradability is needed to guarantee the sustainability of rent appropriation. Social capital is tradeable, however, though all but perfectly. Within audit firms, an individual CPA handles a set of client accounts. That is, from the perspective of the client there is a double tie to the audit service supplier-i.e., to both the audit firm and the individual auditor. For one, client loyalty to the audit firm is rather high. This is particularly true for large companies, which rarely switch from one audit firm to the other (Langendijk 1990). Among small and medium-sized client firms, audit firm switching may well be common, though. Additionally, however, a client's financial reports are attested by an individual CPA. This introduces a tie to the individual auditor, too. In many cases, the auditor' s position involves confidentiality and trust.


In a way, the auditor develops into a mediator who plays an advisory role in a wide array of financial and even non-financial issues. So, social ties are partly linked to the audit firm, and partly to the individual auditor. This implies that by moving to another firm, an auditor only depreciates part of this social network, because client sector ties are both an integral part of the firm as well as linked to the trust relationship with the individual CPA. Of course, the partner-associate distinction is relevant from the observation that ownership is associated with limited mobility.

Finally, we should mention that during the last half century partnership contracts have further diminished the portability of social capital. In both the US. Europe and elsewhere, partnership agreements typically contain a clause that blocks partners from taking clients with them in the event they leave the firm. Needless to say, such contractual constraints bolster the non-tradability assumption of a firm's social capital. Such clauses have also become standard since the second world war and diminish the mobility of a partner's roster of clients.

In sum, a firm of which partners are tied with potential clients is better positioned to build clientele since a potential client can 1) actually become the firm's client, 2) provide valuable information about potential markets, and 3) refer the firm to other potential clients. These aspects should strengthen a professional service firm's survival chances

An Empirical Test An empirical study of the Dutch accounting industry over a period of 110 years (1880-1990) was used to test the proposition that social capital diminishes the likelihood of firms getting dissolved. Social capital was proxied by various measures. For example, 'partner from client sectors' was the proportion of partners who worked in client sectors (i.e., other industries or governmental agencies). They are assumed to have more valuable network ties with potential clients than partners without job experience in client sectors. When departing partners find employment in client sectors, they are likely to have an affiliation that can utilize their professional knowledge. 'Controller' and 'chief financial officer' are examples. As a result, they are likely to be in a position to choose a professional service provider. Because they have strong incentive to take advantage of their social capital, they are likely to choose the professional service firm they worked for (Maister 1993; Smigel 1969). To reflect this effect, the study included a 'partner to client sectors' variable. This is the proportion of partners who left the firm within the previous ten years in order to work for other industries or governmental agencies. A ten-year span was adopted for two reasons. First, the strength of network ties may decrease over time, as the departed partners develop new network ties. Second, the departed partners are ultimately bound to retire from the business world and thus no longer provide economic opportunities to the firm. Note that these proxies of social capital derive from the mobility of professionals who move through a revolving door between two firms. Much of the social capital literature assumes stationary individuals who link two or more organizations through overlapping membership, for example, interlocking directorates.

To test the hypothesis regarding social capital and dissolution, a hazard analysis


Table 2. Complementary log-log regression of firm dissolution

Variables

Intercept

Partners 'from' client sectors

Partners 'to' client sectors

Heterogeneity in partners' origin

Heterogeneity in departed partners' destination

Partners ' industry-specific human capital (Graduate school education) Partners' industry-specific human capital (Industry tenure)

Partners ' industry-specific human capital (Industry-tenure)2

Partners' firm-specific human capital

Partners' firm-specific human capital2

Log-likelihood (Degrees of Freedom)

chi-square compared with previous model (d.f.)

Notes: *p < .10, **p < .05, and ***p < .01 (two-tailed test) Data: 1851 firms, 8696 firm-intervals, and 1164 firm failures .

Model 2

b (s.e.)

-.450 (.934)

-.090*** (.033)

-.0\3*** (.001)

-.061 (.062)

.024 (.076)

-.138*** (.040)

-.\06** (.050)

.145*** (.032)

-.236** (.107)

.226*** (.034)

-2060 (39)

118*** (9)

Regulatory, historical, industry level (e.g., density), firm level (e.g., size, age) and control variables not displayed (compare Pennings, Lee, and van Witteloostuyn 1998). Model 2 includes control variables, but not the variables involving associates' human and social capital.

was conducted on these firms, while controlling for numerous other variables (e.g. , industry level variables such as density, size distribution, history and regulation; and firm level variables such as firm age and size). The rent producing potential of human capital is conditional on the firm possessing social capital. Further details are provided in Pennings, Lee and Witteloostuyn (1998). A partial display of the results is provided in Table 2.

The results were supportive of the hypotheses. In Table 2 we present the results involving the human and social capital of owners, i.e. the partners without showing the simultaneous effects of numerous control and other variables, including those that are associated with firm and industry characteristics. Consistent with the hypothesis, all coefficients of the social capital proxies were negative, indicating that a firm's social capital statistically significantly decreases firm mortality. The effect was statistically significant for two classes of social capital: ties that derive from the recruitment of professionals out of the accounting firm's client sectors, as well as ties that are associated with a firm's 'alumni' who after the tenure in the firm have moved to client sector firms. The heterogeneity in bundles of social ties, as derived from the interfirm mobility of professionals did not statistically significantly affect the firms. When we add the proxies of human and social capital involving associates to the model of Table 2, it was found that associates' social capital does not seem to benefit their firm. In short, this study provided some important findings regarding the beneficial effects of social capital.


What Further Implications Regarding Social Capital? This chapter illustrates the benefits and drawbacks of social capital that are either mediated by individuals or become formed through an array of linking vehicles such as cross-stockholdings and long-term buyer-supplier relationships. We have suggested that the model of the firm conditions our conceptualization and operationalization of social capital and the consequences associated with social capital. Firms are conceived of as unitary actors that interact with other actors (e.g., peer firms in business groups), or they can be conceived of as a community of practices and aggregates of individuals with their distinct objectives and unique agendas (e.g., professional services firms). Allison's (1971) labels of rational and political actor correspond with these stylized forms of organization. In the former case, social capital inheres in the multiplex arrangements that bind a firm to other actors. In the latter case, we focus on individuals and their ties that aggregate to organizational social capital. We then set out to review the benefits of social capital as a distinct organizational (intangible) asset.

Mediated by individuals, social capital nonetheless can be viewed as an organizational property. The individuals might be stationary (as illustrated by the linking pin (Likert 1961) or double agents) or they might migrate between firms (as illustrated by the revolving door syndrome). The relative inclusion of the individual defines its functionality for information and knowledge transmission: the personal needs to be available for external linking, yet requires also sufficient proximity to internal members and groups who can convert the flow of knowledge and other resources into some competitive advantage.

Individuals can also mediate social capital in the case of business groups. In fact, some of the pertinent literature has focused on individuals as transmitters of knowledge between firms they span-for example, so called guest engineers who are employed by the OEM or its supplier and are assigned to work in the partners' site, or civil servants who have been recruited by a chaebol firm and join their ranks. For example, in the above Kukje bankruptcy, it has been suggested that the chaebol management shunned participation in semi-public sectors such as the I1hae Foundation, thus depriving themselves from individually mediated social capital. The Pusan based chaebol neglected to maintain part of its boundary transaction system. What sets business groups apart from partnerships, among others, is that business group links are typically mUltiplex, comprising both personal and impersonal means for maintaining durable links.

In spite of such differences, this chapter has indicated that network embeddedness can have both positive (social capital) and negative (social liability) consequences. The links that bind provide access to competitively critical resources, but they can also be so binding that they are stultifying and rather harmful. The case of Kukje illustrate the deleterious effects of embedded ness that becomes fractured as a result of governmental interventions. The inclination of Toyota to reduce its embeddedness within its keiretzu signals a desire to increase the flow of novel information that current links cannot furnish; its conventional supplier links might be too limited in contributing potentially innovative ideas. The negative first order effect and positive second order effect of social capital on performance in the


apparel industry might be the most robust finding to date regarding the paradox of embeddedness (Uzzi 1997a).

Uzzi (1997a, this volume) makes the important observation that embeddedness is a two-edged sword. Embeddedness ranges from 'under-embedded,' via 'integrated' to 'over-embedded networks.' As was shown, this distinction hinges largely on whether links are 'arm's length'(i.e., contacts based on selfish, profit seeking behavior) versus 'embedded' (i.e., contacts based on trust and mutual intimacy). A firm's network that comprises largely arm's length links does not confer much advantage in knowledge transfer, coordination, or strategic alignment. Conversely, a firm that is strongly entrenched in embedded networks might become so insular that it suspends exposure to markets and technologies that reside outside its immediate environment.

It appears that these distinctions do not readily map on the two contrasting cases we presented in this chapter. The partnerships in a professional services sector fit the conceptual distinctions between arm's length and embeddedness, together with their functionality such as trust, tacitness of knowledge being transferred, and mutual adjustment (Thompson 1967) as coordination mode. At face value, partnerships are internally personalized and anchored in trust, and so we would expect some of the relationships to be among professionals and their clients. Uzzi's (1997a) case involves similar Gemeinschaft-like firms, i.e., small entrepreneurial firms, mom and pop, a trade making up a cottage industry-in short, organizations in which face-toface relationships predominate and which often become extended externally. The apparel world resembles the Chinese 'bamboo network' (Tsui 1997) and Dore's (1983) description of the Japanese textile industry, which he labels as 'cottage industry' and in which goodwill becomes the central feature in describing the prevailing trust and mutuality. The network ties are largely mediated by individuals.

How do we map these descriptions onto the social capital of firms in business groups that tend to be mUltiplex? Are such links more Gesellschaft-like in their appearance and functionality? What sort of processes can we envision in a boundary transaction system in which personal ties complement contracts, equity crossholdings, and traditions that outlive their instigators? We should ask such questions particularly when the individuals in the boundary transacti0n system 'do not go native,' and continue to link up with people and groups in the firms they span, together with other elements that define their inter-firm context. The issue is germane to our earlier review of the firm as a layered entity in which the boundary spanning system resides largely in the more peripheral bands. Such networks abound with actors possessing 'structural autonomy' (Burt 1992) and creating opportunities for opportunism, information asymmetry, and knowledge hoarding--opportunities which Uzzi considers antithetical to embedded ties.

The implication of these observations is to recognize the two faces of organizations and to develop divergent frameworks for capturing the performance implications of network embeddedness. Without forcing us onto a meso-level of research, by artificially integrating face-to-face and small group dynamics with large scale firm-interface arrangements, we might develop a middle range theory of social capital that fits the specific questions we might ask. Whether organizations have at least two faces, or whether we invoke two cognitive models of organizations might


be an issue left to philosophers and epistomologists. Empirically, we might envision a continuum in which organizations range from highly cohesive, well bounded aggregates that are tightly coupled and present few if any intrafrrm hurdles for coordination, knowledge sharing, and strategic positioning. We can also envision organizations that are loosely coupled, with permeable boundaries and few isolating mechanisms, barely holding themselves together and maneuvering on the brink of dissolution. In either case, the firm is part of a larger context. How they position themselves onto this continuum, and what image we impose on them remains a never-ending challenge. The research on social capital will shed further light on how they negotiate their embeddedness, and what sort of advantages and shortfalls they derive from that social structure.

We appreciate the comments from Jon Brookfield, Jeff Dyer, Giovanni Gavetti, Jim Lincoln, Lori Rosenkopf, and Brian Uzzi.

NOTES

1. Embedded ties could have two (if not three) rather divergent meanings: I) ties that are reinforced by mutual feelings of attachment, reciprocity, and trust; and 2) ties that are a link within a larger set of links and nodes. Since Uzzi's work is confined to dyads, the first meaning applies. When members of a dyad become affected by third parties who envelop their tie, as in Burt (1992), the second meaning applies. In both cases, the concern is with a focal person. If one moves to an even higher level of analysis 3), as for example the internet, transactions among textile traders in 15th century Florence and Flanders, or community power structures, then the network takes primacy over the ties between individuals who are embedded in those networks. A person's or firm's 'centrality' conveys relative access to other actors in the network such that a focal actor's social capital hinges partly on the direct and indirect ties that the tied partners possess (e.g., Levine 1972). Empirically the effect of centrality on firm behavior or performance has not been studied adequately (an exception is Freeman, this volume). 2. Note that the rational model of the firm does not presume anything about its embeddedness here. In either the rational or political scenario, we do not assume organizations to behave as if they are atomized from the impact of their relations with other organizations, or from the past history of these relations. If we were to extend methodological individualism to the embeddedness of firms, we would not be able to furnish an adequate account of how firms' actions combine up to the level of the value chain, markets or institutions. We only make the analytical distinction based on the relative saliency of aggregation when examining social capital as a firm-specific asset. Hence, our reluctance to include Allison's second model, the 'organizational actor model' in our review. In the extreme, over-socialized individuals would reduce to mentally programmed automatons who mechanically replicate the routines that the organizational socialization process has imprinted onto them. As role incumbent, they would have no discretion to embellish their position or protect personal interests, nor could they be construed as the personal authors of their social network. 3. Some examples might illustrate the issues at hand. Firms are tied to each other through trade associations, business groups, consortia, cartels, joint ventures, and directors who sit not only on their board but on the boards of other organizations as well. They are locked into licensing agreements and long term supplier-buyer arrangements, and might have made significant investments in specific interfirm relationships. The presence of such links and their benefits seem obvious, when that capital is treated as firm level or individual level phenomena.

For example, Boeing's 747 aircraft requires the input from numerous contractors and subcontractors-only certain chunks of the cockpit and wings are developed and produced by Boeing. Such inter-firm transactions result in long term links that become independent of the members who forged them originally. Many firms occupy positions in the value chain with interdependencies so dense that one might consider the value chain to be a more salient unit of action than the firms that exist within the value chain. A simple illustration from the computer industry might further illustrate this observation.

During the main-frame computer era, it was common for firms like mM and Hitachi to control all the steps in the value chain, from silicon, computer platform, system software, application to distribution and service. The firm was the value chain, and competition between corporations matched competition


between value chains. In the late nineties, we observe a fragmented horizontal competition between finns, but vertically dense complementarities have surfaced. Microsoft competes with Apple and Unix, but is symbiotically linked with upstream PC manufacturers and their suppliers, such as Intel . Downstream, the finns relate to distribution and service finns such as computer stores and mail order firms. Microsoft has been a shrewd exploiter of network externalities: the various technologies require complementary products, lead to the fonnation of virtuous cycles such as software developers writing more Microsoft Windows applications, and when these become available, more customers adopting Microsoft Windows. Increasingly all firms in the value chain become 'locked-in' (or locked-out!) resulting in a complex string of links that are straddled around a dominant computer design (e.g., Yoffie 1996). In such a value chain, links are often de-personalized and it is the organizations that become the salient unit of the network. The ties in such networks are critical for the finns involved as their products and technologies become heavily intertwined with those of others.

Much of the social capital literature has an individual slant (e.g., Burt 1997) and finn attributes have often been examined as an individual manifestation. Burt's (1997) recent study examines investment banks but really focuses on its traders and the 'structural holes' that benefit the size of their perfonnance based bonuses. One might also focus on their banks' tombstones and the social capital that could be inferred from them. Coleman's (1988) classic example involves the tight social circle of diamond traders in New York whose smooth and paperless transactions hinge on the social ties that they maintain with other traders. The trust that is sustained within such a network results in a substantial reduction of transaction costs. Likewise, he (Coleman 1988) shows that children whose parents know other parents and teachers are better embedded in their school community and show lower dropping-out rate. Finally, Uzzi (1997a) recounts the linkages among individuals who make up the New York apparel industry. In such instances, the issue of aggregation and presumption of finn as a unitary actor is rather moot: the entrepreneur is the finn. In these and many other contributions, social capital is a resource that belongs to the networking or interacting individuals and that might affect the venture with which the embedded individual is associated. 4. By the same token, an individual who is neutral to the bridging between two finns cannot easily be incorporated in the organization's social capital. Referee, arbitrator, or mediator roles are sharply different from those we associate with ambassador, spy, or guest engineer. The fonner's neutrality might depreciate or sanitize whatever infonnation or knowledge the 'middle-man' furnishes to the linked organizations. His neutrality also precludes intimacy and creates social distance. We assume that organizations have discrete bundles of knowledge and infonnation whose rents will be augmented by the development of 'proprietary' social capital. 5. Sherer (1995) identifies three major types of employment relationships. The first is the employment relation coupled with ownership. It includes employees who share the risk of organization via various incentive systems which link their earnings to the perfonnance of the organization. Employees in that relation constitute the core group in our analysis. The second is the traditionally described employment relation in which employees receive a fixed amount of earnings, provide a fixed length of time, and perfonn work based on the direction from the supervisor or job description. Employees in these types are designated regular group in the present discussion. The third embodies relationships that involve temporary employment or contracting out. Employees in this type fonn the temporary or marginal group. Note that with the rise of temporary employment agencies, outsourcing and sub-contracting, this latter group has acquired huge proportions. Analogous distinctions have been made by Jensen and Meckling (1976) and Milgrom and Roberts (1992). 6. The classification was suggested by Jon Brookfield. 7. For example, Toshiba and Tokyo Power maintain close buyer-seller relationships; they both draw graduates from Tokyo University who get promoted in their respective companies, and they move in tandem, their roles might change but their mutuality stays intact. The demography of the system coevolves with that of the respective organizations. Such evolutionary arrangements ensure network continuity throughout the firms' history. 8. The tenn transitive cross-equity holding refers to a string of keiretzus finns between which ownership is mutual yet unequal. Nishiyama (1982) reports the pattern of large block holdings in the Sumitomo Business Group, with Sumimoto (S.) Life Insurance owning a larger percent of shares in S. Bank, S. Metal, S. Chemical, S. Electric, etc. than vice versa; it augments its power over these finns because these finns in tum own shares in each other, such that cumulatively, S. Life Insurance scores highest on the 'comprehensive power index.'

•

A Relational Resource Perspective on Social Capital

ABSTRACT

3 Luis Araujo

Geoff Easton

This chapter reviews the notion of social capital from a resource based perspective. We argue that the notion of social capital relies on a metaphorical mapping of features associated with economic notions of capital or assets into the social domain. We start from the notion that not all economic resources can be classified as assets in the way the term is deployed within legal and accounting language, and argue that social capital shares many features with other less understood and intangible resources. By employing a framework to examine the multifaceted and relational dimension of resources, we examine in detail the entailments of the social capital metaphor and relate to current applications within the business and management literature. We conclude by reflecting on the characteristics of social capital as an economic resource and caution against the dangers of engaging in facile prescriptions based on a cursory understanding of the logic of accumulation and use of social capital.

INTRODUCTION

The notions of embedded ness (Granovetter 1985), social capital (Bourdieu 1986; Coleman 1988, 1990), and social resources (Lin 1990) have, in recent times, contributed to a rekindling of interest in the interaction between the economy and society. At the heart of this revival is an attempt to trace the mutual influences between economic exchange and the social structures in which the economy is embedded.

The notion of embeddedness relies on the insight that economic life is shaped and constrained by norms, social networks, institutions, and a variety of motives

A Relational Resource Perspective on Social Capital - 69

other than the unconstrained pursuit of self-interest. The revival of the interest in the social has also got to do with the increasing awareness that modern economies, while relying on impersonal forms of exchange and complex forms of contracting with third-party enforcement, cannot dispense with other forms of support in the guise of moral, social rules or codes of conduct (Platteau 1994ab).

The emergence of market economies does not diminish the need for social solidarity and trust. On the contrary, as Macneil (1986: 592) argues, market economies have acute problems regarding social solidarity. The embeddedness of economic exchange in social structures very often dictates complex legal structures remote from, though essential to, the exchange relations themselves. Thus markets cannot be regarded as a spontaneous order or a primitive state of nature, but a convergent network of actors and institutions mixing different forms of exchange and where order is generated through translation processes and rules that are reproduced across exchanges and over time.

Market order is partially generated by institution building, to establish and enforce sanction systems and solve coordination problems involving the risk of freeridership and dilemmas of collective action. But, as Bates (1988) argues, formal rules enforced through third parties can also be subjected to free-riding, and the role of institutions as impartial rule makers and enforcers can be questioned. The problem cannot be resolved through appeal to a further tier of institutions to monitor the performance of the first tier. In short, order can emerge only in the presence of both institutions promoting and enforcing formal rules, and informal norms such as a generalized morality that draws on a society'S social fabric and culture. These informal norms can thus act as substitute for or a reinforcement of formal rules and control mechanisms, with the consequence that coercive enforcement of formal norms becomes either redundant or of secondary importance. 1

The objective of this chapter is to revisit some of the issues concerning the ways social structures impact on economic exchange. In particular, we are concerned with the ways the notion of social capital, residing both in concrete, interpersonal relationships inside and outside formal organizations as well as in wider social structures, can be deployed to break down some of the artificial divisions between the economy and society. The structure of the chapter is as follows: in the first part we look at the notions of embedded ness and social capital and the way they have been used to understand the coordination of socioeconomic life. In the second part of this chapter we use a framework we developed to dimensionalize economic resources (Easton and Araujo 1996), to look at the characteristics of social capital as an economic resource. We conclude with some speculations on the role of social capital in the coordination of socioeconomic life.

The Notion of Embeddedness The key issue that arises in discussions of embeddedness and that is germane to the notion of social capital is the extent and form of the embedded ness that we might expect to find in a modern economy. Below we review briefly some of the more recent offerings to give a flavor of the debate.

The embedded ness argument is given short shrift in traditional neoclassical, undersocialized conceptions of human action and also in neoinstitutionalist


arguments that, although acknowledging the role of social networks and private orders borne out of repeated social contacts, still regard the role of abstract, formal rules as solely responsible for market order (Platteau 1994a). Greifs (1994) analysis of the contrasting solutions adopted by Maghribi and Genoese traders in the eleventh century to trade expansion demonstrates the advantages and limitations of embedded ness as a mechanism for governing economic life. The collectivist system adopted by Maghribi traders, where order was enforced through moral sanctions, worked well in the case of intraeconomy agency relationships, but was incapable of supporting intereconomy relationships and of allowing for the division of labour necessary to take advantage of new trade opportunities. By contrast, the Genoese introduced formal enforcement institutions to support impersonal forms of economic exchange and promote further division of labour, thus enabling their society to capture the efficiency gains stemming from the expansion of trade.

Hardin (1993: 510) regards the thick relationships that the embeddedness argument prescribes as yielding only a part of the knowledge we have of others. But, of course, one might learn from the experience of others, through reputational effects and a variety of other indirect means. As Hardin (1996: 31) argues, there are two modal categories of controls operating in society. There are geographical associates-the group of friends, family, and associates with whom one is inevitably bound up in repeated interactions and long-term relationships. And there are the elaborate large-scale controls associated with institutions, such as the legal system, relying on formal rules and coercive enforcement. But between these two modal categories there are a variety of mixed devices such as broad social norms, mixing elements of both modal categories, that provide important elements of social control (Hardin 1996).2

In his oft quoted critique of conceptions of human action in economics and sociology, Granovetter (1985) rejects both notions of undersocialized actors as behaving atomistically and oversocialized views of human action, where actors simply follow a script attached to the intersection of the social categories to which they belong. Instead, Granovetter (1985: 490) revives Polanyi's (1957) notion of embeddedness and stresses 'the role of concrete personal relations and structures (or 'networks') of such relations in generating trust and avoiding malfeasance.' Granovetter's argument revolves around the notion that the production of trust in economic life is mainly accounted for by concrete social relations rather than institutional arrangements or norms of generalized morality.

He further argues, however, that the existence of strong social relationships may contribute both to the production of trust and trustworthy behavior as well as, perversely, mistrust and malfeasance. In short, embedded ness can both contribute to the resolution and the collective dilemma implied by the Hobbesian position as well as introduce the possibility of disruption on a larger scale than the one that is possible in a truly atomized, state-of-nature social situation (Grano vetter 1985: 493). Zukin and DiMaggio (1990: 15) elaborate on the notion of embedded ness by defining it as the contingent nature of economic action with respect to cognition, culture, social structure, and political institutions. Cognitive embedded ness is defined as the equivalent of bounded rationality, the set of heuristics and biases that pervade all forms of reasoning. Cultural embedded ness refers to yet more limitations


on economic rationality imposed by the constraints of shared values and understandings in shaping economic goals and strategies. Structural embedded ness is for Zukin and DiMaggio (1990: 18) what embedded ness tout court is for Granovetter-the contextualization of economic exchange in ongoing patterns of social relationships. Finally, political embeddedness refers to the power struggles in which different types of actors (e.g., business firms, institutional actors, and the state) vie for the power to shape the rules that govern economic life.3

Uzzi (l996a, 1997a) elaborated the concept of structural embeddedness describing it as a specific logic of exchange or coordination in which trust, borne out of repeated interactions and the prospect of a continuing relationship, pushes the logic of calculativeness and monitoring to a secondary role. In this mode of coordination, thick relationships provide more fine-grained, tacit, and holistic information transfer with a significant and positive impact on problem solving and conflict-resolution arrangements as well as innovation.4 Uzzi's conclusions are largely unsurprising given the plethora of studies over the last twenty years on buyer-supplier relationships in Europe (Hakansson 1982; Axelsson and Easton 1992; HAkansson and Snehota 1995), the U.S. (Helper 1991), and Japan (Sako 1991; Smitka 1991; Nishiguchi 1994). Furthermore, the literature on flexible specialization and industrial districts has long since underlined the role of embeddedness in promoting flexibility, innovation, and adaptability to changing demands as alternatives to mass production of standardized goods (Lazerson 1988, 1995).

One central conclusion that can be drawn from this brief review is that the form and extent of embeddedness seen in an economy is strongly influenced by the discipline and the school of thought the writer subscribes to. While this phenomenon is a commonplace, as any sociologist of knowledge would acknowledge, what is less usual is that one line of attack on the issue is the result of the borrowing of a concept from one of these disciplines for use in another.

Economics has been described as the study of the allocation of resources, sociology as the study of human aggregations and their behavior. Clearly, these definitions already imply an overlap. Resources cannot be allocated in an economic system without the intervention of human agency. Nor can the behavior of social systems be unaffected by the resources they create, consume, and exchange, directly or indirectly. However the crucial link here is that of resource. What has happened is that sociologists have appropriated the concept of resource, or more narrowly capital, from economics and have used it to capture the notion of the embeddedness of economic behavior, itself described in terms of resources, in social structures by way of the notion of social capital. While the basic idea has been around for a long time, it is only more recently that it has become popular.

Social Capital The notion of social capital is an intriguing one and raises interesting possibilities as a counterintuitive metaphor, mapping an economic domain where the notion of capital is well established and tying accounting and legal conventions to a social domain, where accounting for value, investment, depreciation, and so on poses a number of difficulties. One of the functions of metaphors in theory building is an exploratory one (Easton and Araujo 1993). This is not much removed from an


explanatory function, but it emphasizes research directions and possibilities rather than assumes a comprehensive mapping of the base into the target domain. Before proceeding to explore the relationship between the base and target domains, we focus briefly on the notion of social capital that has been used in the social sciences and the business literature. Portes and Sensenbrenner (1993), while recognizing the value of Granovetter' s embedded ness argument, make a case for focusing on the more manageable concept of social capital as developed by Pierre Bourdieu in the Francophone world and by James Coleman in the Anglo-Saxon literature.

In Bourdieu's hands the notion of social capital is deployed in a rather broad and undifferentiated sense. Bourdieu (1986: 249) defines social capital as 'the aggregate of the actual or potential resources that are linked to a possession of a durable network of more or less institutionalized relationships of mutual acquaintance and recognition ... a 'credential' which entitles them to credit, in the various senses of the word.' Further on, Bourdieu characterizes these connections or obligations as the product of investment strategies aimed at establishing or reproducing social relationships that can be leveraged either in the short or long run. Different forms of capital (e.g., cultural or political) derive their forms from the fields in which they are deployed, but the notion of economic capital remains the master metaphor for all other forms of capital.

Coleman (1988) starts from the rationalist assumption that each actor has control over certain resources and that social capital is one type of resource available to an actor. Coleman (1990: 302) distinguishes between physical, human, and social capital as three separate forms of capital, all of them having a potential to facilitate action, or as Coleman puts it 'social capital is productive, making possible the achievement of certain ends that would not be attainable in its absence.'5 If physical capital is regarded as wholly tangible, being embodied in observable physical forms, and human capital is less tangible but still embodied in the skills and knowledge possessed by an individual, social capital is even less tangible and nonlocalizable for it is a property of social relationships and social structures.

On closer inspection, Coleman's notion of social capital encompasses both properties of dyadic relationships and the social structures in which these relationships are embedded. Social capital takes the following forms: obligations, expectations and trustworthiness of structures, information channels, norms, and effective sanctions. So, for example, when in a dyadic relationship one of the parties does a favor for the other, there is an expectation of reciprocity at some unspecified future date. Coleman (1990: 306) refers to this situation as one of the parties issuing a credit slip that can be called on at a future date. But other forms of social capital are clearly properties of social structure that impinge on and constrain specific, dyadic relationships. An example of this would be a moral norm prescribing the foregoing of self-interest in favor of collective welfare.6

Coleman (1990: 317) also argues that the public good aspect of most forms of social capital is, perhaps, its most distinctive feature vis a vis other forms of capital. As in Olson' s (1965) collective action dilemma, self-interested individuals free-ride or seek to reap the benefits of public goods while evading the costs of participation in their production, and so rational actors will consequently underinvest in the production of these public goods.7 Many of the benefits flowing from actions that


create social capital can be enjoyed by people other than the ones who have contributed to those activities in the first place. In short, those who produce them cannot necessarily appropriate the benefits flowing from investing in the creation of social capital.

Smart (1993: 393) elaborates on the problems concerning the use of the term capital and offers a few interesting avenues to preserve the distinctions Bourdieu makes while clarifying the interrelationship between different forms of capital. Smart notes that social capital by its very nature is vague and unmeasurable: it lacks a currency and a space of calculation where debits and credits can be accumulated and compared. An obligation, for example, becomes concrete only once it is liquidated, and until then there is no certainty that will ever be reciprocated. For Smart (1993: 393), if social capital cannot be possessed but can be converted into other forms of capital, then it is entirely contingent on the reproduction of the social structures in which it is embedded. If social debts-such as obligations-can be recovered through enforcement by third parties then, according to Smart, we are talking about economic and not social capital. 8

Smart relies on a similar logic to argue that social capital is a resource that resides in dyadic, specific social ties and that more generalized resources such as honor or reputation that are valued within society or subgroups within it, are best characterised as symbolic capital. Smart goes on to distinguish between Bourdieu's notion of capital-in-general and the notion of power. Whereas power is seen as resting on the authority to command the actions of others, capital in the sense Bourdieu uses it, is the ability to induce others to act in one's interests through the leverage of resources available to the agent. Smart (1993: 394) distinguishes between the different forms of capital invoked by Bourdieu in the following way:

Economic capital involves ownership of objects, but property ownership entails claims that others may not interfere with your property without your permission, and exclusive ownership may be used to induce others to act in particular ways (such as hiring or firing them). Cultural capital is a claim to having the ability to engage in certain types of practices, and in the strongest forms it accords a monopoly over such practices (for instance, medical doctors or accountants). Symbolic capital involves claims by the possessor that he or she be treated in particular ways by classes of others. Social capital consists of claims to reciprocation and solidarity from particular others. What is fundamental to social capital. however. is that explicit claims are normally excluded from the performances within which they are made, so that power over the action of others is radically distinct from exercises of power utilizing the discourse and apparatus of command. (emphasis added)

Smart uses the example of gift giving and guanxi in China as an instance in where explicit recognition of instrumental goals is excluded from the performance, and incompetent performances results in loss of face and dissipation of the very outcome that the performance intended to achieve. Gift exchange must, by its very nature, be devoid of any explicit reference to a calculative logic, lest they be devalued as failed gift performances or understood as a bribe. In essence, Smart's reinterpretation of Bourdieu's social capital consists of locating it within concrete, dyadic relationships and insisting on the absence or accomplished concealment of explicit claims by the possessors of social capital when attempting to cash in their credits as well as the


nonexistence of third-party enforcement to reclaim bad debts. Although we are less keen on the fine distinction that Smart adopts between social and symbolic capital, his definition has at least the advantage of relating social capital to contextual performances that have to be interpreted in the light of experience and further cues.

What we conclude from this discussion is that while there is agreement that there are different forms of social capital, there is disagreement as to what conceptualizations of these forms might be most useful. One of the ways we can explore the notion of social capital is by examining in more detail how the base domain of the metaphor (economic capital, or more generally economic resources) map into the object domain (social interaction). In the next section we briefly review the notion of economic resource before proceeding to examine the properties of social capital as an economic resource.

The Notion of Economic Resources The notion of resource has recently been deployed extensively within the strategic management literature under the guise of the resource-based theory of the firm. These developments can be traced back to the pioneering work of Penrose (1959) and pursued in different guises since Wernerfelt (1984).

Resources, in the traditional language of business strategy, are strengths, and there is little attempt here to suggest how and why they become so and how that might relate to one another. Despite its growing popularity, the resource-based theory of the firm has come under attack for its lack of a clear conceptualization of what constitutes a resource. In the absence of a clear definition, we are left with a tautology: firms must discover which resources contribute most to their competitive advantage and develop or acquire more of these (Porter 1991).

This chapter draws on a specific notion of economic resources, inspired by the industrial networks research program into interorganizational relationships-see Axelsson and Easton (1992) and Hakansson and Snehota (1995) for recent examples. We define economic resource as any entity that can be deployed by an actor that is capable of continuing independent existence, has futurity, and can, or may, meet an economic need. In this sense, an economic resource is a stock that can be drawn on either through ownership and control or through indirect access. Resources stand in contrast to activities. Activities comprise current actions and have no direct continuity. Resources have a durable character, even if they are the products of past activities, and acquire their meaning and value when mobilized within specific activity structures.

We make a further distinction between the generic notion of economic resources and the terms asset or capital as used in management accounting theory. We reserve the use of the term asset or capital to describe a resource whose ownership, control, and market valuation can be determined. Thus assets are simply a subset of economic resources protected by property rights enforceable by third parties, whose valuation has some claims to be measurable which, in turn implies the construction of calculable spaces and technologies able to enforce standardized measures of value (Miller 1994).

An obvious corollary to our distinction between resources and assets is that we are not constrained by legal and accounting definitions of the firm. We draw on the


early work of Chamberlain (1968) to emphasize the distinction between the firm as an entity defined for legal and accounting purposes and the firm as an entity for coordinating a set of business activities, a key aspect of the industrial networks approach. Chamberlain employs a broad notion of asset and argues that most of the assets that constitute the firm's instruments of action are not the ones described on its balance sheet. In other words, Chamberlain does not assume that a firm's capability for strategic action resides within the boundaries of what it owns and controls. Instead the firm's capabilities are seen as embodied in evolving networks of interdependence both within the firm and with aspects of its environment. A firm's strategic capability depends on two factors: 1) its capacity to generate resources from the its current operations and 2) its capacity to mobilize support and resources from entities and institutions within its environment. More recently, Barney (1991) argues that a wide variety of the firm's resources may be complex to the extent that they reside in relationships among people and may be socially constructed. Among these Barney includes the interpersonal relations among managers, a firm's culture, and the firm's reputation among suppliers and customers. And, from an industrial networks perspective, we would add social bonds developed in the course of economic exchange relationships between two organizations (Hakansson and Snehota 1995).9

The above discussion has some relevance to the use of the concept of capital-ingeneral and, in particular, the distinctions between different forms of capital. Economic capital, in the cursory way most authors in the social capital literature seem to understand it, is only a small subset of all the economic resources that are necessary to conduct economic life. Social capital patently lacks most of the characteristics that define economic capital. In particular, the absence of accounting conventions, property rights enforceable through third parties and a system of exchange devalue the metaphor somewhat. But, on the other hand, it shares many other features with other economic resources that cannot be qualified as capital, in the way defined above.

In the following section we attempt to compare the notion of social capital with that of an economic resource by relying on the framework adapted from Easton and Araujo (1996). The adaptation consists in changing the focus from resources as an organization-based phenomenon to resources as a property of intra and interorganizational, individual-based dyads and social networks.

Social Capital as an Economic Resource This framework dimensionalizes resources in four separate categories relating to their existence, evaluation, relationship to actors, and relationship to other activities and resources. These dimensions are further subdivided into separate categories each (see Figure 1). Pennings and Lee (this volume) employ a similar list in their discussion of the benefits of embedded ties for audit firms.


Creation Existence • Depreciation

Durability

Relationships to actors

Valuation

Evaluation • Evaluability Scarcity Value

Relationships to other activities and resources

• Controllability Accessibility Tradeability

Integrity Versatility Complementarity Understandability

Figure 1. Dimensions of resources (adapted from Easton and Araujo 1996)

The Existence of Social Capital Starting with the existence group, the processes involved in the creation, depreciation, and durability of social capital are difficult to understand, and there is no obvious causal mechanisms that relate investment to a logic of accumulation of social capital. The same is true of other intangible resources such as organizational competencies in producing innovative products, for example.

Coleman (1990: 317-318) argues that much social capital arises or disappears as a by-product of other activities and without anyone's willing it into or out of being. Furthermore, some of the conditions that may foster the creation of social capital can also contribute to its destruction. The closure of social networks, for example, where judgements of trustworthiness may depend on intersecting patterns of relationships and transitive judgments (A trusts C, because A trusts Band B trusts C) may lead both to inflationary and deflationary spirals of trust placement in a community (Coleman 1990: 318). Stability in social structures may become an important precondition for the creation and maintenance of social capital, although strangely Coleman makes an exception for formal organizations where social capital is assumed to be vested in role occupancy rather than on the identity of the occupant. 10

Finally, Coleman (1990: 321) argues that all factors that help to minimize social interaction and increase individuals' self-reliance, such as affluence or public aid, tend to contribute to the depreciation of the stock of social capital, through bypassing the very activities-e.g., the exchange of obligations and favors-through which social capital is produced and maintained. Like other economic resources such as reputations or brand names, social capital is depreciated through the lack of use, and only through use can its stock be maintained and rejuvenated.

Coleman's account of the creation, destruction, and maintenance of social capital undervalues instrumental actions directed at creating social capital and elides some of the dilemmas we alluded to earlier-namely the relationship between social capital as a property of dyads, small-knit face-to-face groups, or wider social structures. For example, Burt's (1992, 1997) notion of structural holes describes how social capital is created and leveraged as a result of brokerage opportunities in a social network. Uzzi's (1996a, 1997a) account of the New York apparel industry


documents how social capital is created by one-sided investments in business relationships anticipating future reciprocity.

Pennings and Lee and Lazega (this volume) discuss the relationship of dyadspecific social capital to wider structures at intra- and interorganizational levels. Putnam (I993a), in his celebrated work on modern Italy, elaborates on this point in the case of trust at the societal level. lI For Putnam (1993a: 171-176) the generalization of trust from closed social networks to the wider society can be traced back to two sources: norms of generalized reciprocity and networks of civic engagement. Norms of generalized reciprocity, or diffuse reciprocity to use Keohane's (1986) expression, are seen as a catalyst for the development of social capital. As in Gouldner's (1960) account of reciprocity, social capital is built not solely on the basis of exchanges of obligations that are perfectly balanced but on mechanisms that induce further debts and credits of obligations and avoid the expectation that all exchanges will ever return to a zero state in which none of the parties is indebted to the other.

Keohane (1986) makes an important connection between the norm of specific reciprocity in dyadic relationships, where the exchange of obligations and duties is roughly equivalent and performed in a strictly delimited sequence, and the norm of diffused reciprocity, where equivalence is less precise and sequence of events less clearly bounded. The repayment of debts and obligations in specific reciprocity situations may lead actors to take a broader view of their interests and engage in diffuse reciprocity. Conversely, the decay of diffuse reciprocity may lead actors to revert to exchanges narrowly bounded by specific reciprocity. As Keohane (1986: 25) puts it: 'specific and diffuse reciprocity are closely interrelated. They can be located on a continuum, although the relationships between them are as much dialectical as linear.'12

In short, the creation and maintenance of social capital as an economic resource is often a by-product of distributed activities and actor orientations that may be reinforced or hindered by the institutional context in which these activities are embedded. Whereas the role of institutions and formal rules has been underlined in helping to supersede collective action dilemmas, by providing a stable background of enforceable norms and duties and thus lowering the risks involved in engaging in cooperative relationships, less has been made of their role in destroying social capital.

Recently, Pi Ides (1996) has argued that the intervention of the state through the provision and enforcement of legislation can destroy social capital through an insufficient appreciation of the role of social capital that underwrites the successful enforcement of laws. Pi Ides (1996) advances an argument on the role of the state in contributing to the destruction of social capital that can be extended to all attempts at formalizing rules. There are three ways in which social capital can be destroyed through state intervention: 1) indirectly, by attacking the structural foundations on which social capital can be created and maintained-e.g., by prescribing spaces in which social interaction mayor may not take place, 2) directly, by attacking norms of reciprocity through efforts to rationalise conventions, and 3) indirectly, by attempting to incorporate social norms into legislation without taking due account of the role of remedial flexibility in the enforcement of social norms.


The Evaluation of Social Capital In this section, we attempt to relate the notion of social capital to four dimensions associated with the evaluation of economic resources: valuation, evaluability, scarcity, and value (positive/negative) (Easton and Araujo 1996). All economic resources are generally held to possess a value, even if that value cannot easily be expressed in the metric of money. More often than not, it is the intangible and more difficult to evaluate resources that constitute the most valuable resources a firm possesses (Itami 1987). Furthermore. the value of a resource is intrinsically tied to the context in which it is used. or to put it differently, to describe the content and value of a resource is also to describe its context of use.

Social capital is usually hailed as an important resource for promoting both effective government and prosperous economies. In particular, trust has been singled out as a key resource in promoting economic growth and competitiveness and as having a positive impact in intraorganizational relationships (Breton and Wintrobe 1982; Miller 1992) as well as in interorganizational contexts-from buyer-supplier relationships (Barney and Hansen 1994; Nooteboom 1996) to industrial districts (Lorenz 1993) or the relationship between capital market institutions (Neu 1991).

In general. as Gambetta (1988a: 221) observes: 'societies that rely heavily on the use of force are likely to be less efficient, more costly, more unpleasant than those where trust is maintained by other means. In the former. resources tend to be diverted away from economic undertakings and spent in coercion. surveillance, and information gathering and less incentive is found to engage in cooperative activities.'

Social capital can thus be regarded as a valuable. infrastructural resource to economic life but one that cannot be evaluated easily. Sabel (1995) contrasts two opposite views on this issue, one inspired by Hayek and the other by Durkheim. Whereas the Hayekian view stresses the distributed and partial nature of knowledge in society. and the way norms evolve as a spontaneous. unintended outcome (Bianchi 1994), the Durkheimian perspective stresses the role of institutions in curbing the freedom of rational. selfish agents to act as they see fit. From a Durkheimian perspective, the economy can perform well only if is embedded in a well-integrated society and that society is willing and able to impose normative constraints on the pursuit of self-interest (Streeck 1997: 199).

The role of social capital as an economic resource is clearly less important under a Hayekian perspective. For Hayek (1978) the market with the price mechanism, sets of property rights. and enforceable rules is the best example of an evolved institution. Social institutions are instrumental, problem-solving mechanisms that evolve spontaneously under the pressure to find ways to coordinate the activities of individuals facing a complex world equipped with limited knowledge. Under a Durkheimian perspective the value of social <;apital to economic life is much clearer. For example. social constraints on the pursuit of self-interest in the form of moral norms or formal laws can reinforce trust and thus increase the efficiency of economic transactions by reassuring potentially suspicious parties of continued adherence to norms of reciprocity and normal business practice. As Streeck (1997: 202) puts it: 'Credible information that the other side has noneconomic in addition to


economic reasons not to defect accelerates and consolidates the growth of trustful relations. '

But if a Durkheimian perspective exalts the value of social capital in promoting economic efficiency, it also warns about the dangers of attempting to design social institutions with the sole purpose of economic efficiency in mind. Streeck (1997: 217) warns that 'the kind of social embedding good economic performance requires can be built only for reasons other than good economic performance, enabling it to support rational-economic action by containing it.'

These considerations lead us to reflect on two other dimensions of the evaluation of social capital: scarcity and value (positive/negative). The issue of scarcity of social capital is important and related to the issue of valuation. As a resource, most forms of social capital are not scarce in the sense that it is a stock depleted through use. On the contrary, social capital depreciates through lack of use and appreciates through extensive use and imbalanced exchanges. For Putnam (1993a: 178-179) both reciprocity/trust and dependence/exploitation can become stable equilibria in particular communities depending on initial conditions-namely, stocks of social capital and paths of evolution. As in North (1990), path dependence is deemed responsible for the self-reinforcing nature of a particular equilibrium. Distrust, for example, once set in has the capacity to become self-fulfilling and lock a particular community into a low-level equilibrium of authoritarian government, patron-client relationships or putting them at the mercy of criminal organizations (Gambetta 1988a).

Sabel and Zeitlin (1997) effectively reject the notion of path-dependence and claim instead that actors are able through reflexivity and choice to transform the conditions under which they live and to be less discriminating between the two equilibria Putnam describes. Economies often contain what Sabel and Zeitlin (1997) describe as moral borderlands where moral rules are only partly observed or significantly relaxed. In short, the cycles of trust and mistrust that Putnam describes as being the resultant of initial endowments of social capital and path-dependent evolution may be less sharply defined and delimited in historical epochs and equally dependent on stocks of social capital.

The issue of the scarcity of social capital as an initial condition to foster economic development has been hotly debated for some time. Recently, Evans (1996) reflects on the role of social capital in promoting economic development and suggests that endowments of social capital cannot be seen as the major obstacle to foster development projects in the Third World. In most Third World communities prior endowments of social capital, in the form of social networks, norms of solidarity and trust at the micro-level are seen as adequate, but the difficulty lies in scaling up such capital to generate solidarity and diffuse norms at a politically and economically efficacious level.

Lastly, we must consider the valence of social capital as a resource. Throughout most of this chapter we have implied that social capital is an asset to political and economic life, rather than a liability. But, as Portes and Sensenbrenner (1993) remind us, the same social mechanisms that account for useful resources appropriable at the individual, dyadic, or community level may also set important limits to action and economic development.


Gargiulo and Benassi (this volume) illustrate how social capital can, in the face of changing circumstances, turn into social liability constraining opportunities for organizational change. Recent research on entrepreneurship and immigrant communities in the U.S. documents both the negative and positive aspects of social capital and embeddedness. Portes and Sensenbrenner (1993) argue that the greater the social capital produced by solidarity and community ties, the more likely that particularistic demands will be imposed on successful entrepreneurs, thus limiting the possibilities of individual expression and the expansion of opportunities outside the community boundaries. Waldinger (1995), in his study of immigrant communities in the New York construction industry, argues that in this instance, the embedded ness of economic behavior in ongoing social relations among a myriad of social actors impedes access to outsiders. Embeddedness contributes an exclusion effect from social networks, breeding a preference for established players with track records. However, the overlapping of economic and ethnic ties has a further undesirable effect, since outsiders also fall outside those networks that define the industrial community.

Lastly, Gambetta (l988b) reminds us that there are situations where the public interest might be better served if trust and the social capital built through exchange of obligations and favors is collapsed rather than reinforced. Baker and Faulkner's (1993) study of social networks involved in collusive practices in the American heavy electrical equipment industry demonstrates the point.

The Relationship of Social Capital to Actors The third group of dimensions to classify economic resources is labeled relationship to actors. In this case, actors might be individuals, groups, business firms, or other collective actors such as business interest associations. Three dimensions characterize the relationship of a resource to actors: controllability or the extent to which actors can appropriate and control a resource, accessibility or conversely the existence of barriers to use a resource and tradeability, or the ability of actors to convert resources through exchange processes.

Starting with controllability, we recall that social capital can be regarded as a public good in the way Coleman (1990) defines it. Public goods are usually characterized by jointness of supply and nonexcludability of benefits (Olson 1965). But clearly this is not always the case with social capital. For example, Portes and Sensenbrenner (1993) describe an informal financial system operating in the Cuban immigrant community in Miami in the 1960s, whereby Cuban bankers were prepared to provide 'character loans' to newly arrived refugees, based on the personal reputation of the recipient in Cuba. This system worked well until a new wave of immigrants, no longer known to the Cuban banking community in Miami, arrived in 1973 after which character loans were discontinued.

Hechter (1987) discriminates between public and collective goods, on the basis of the degree to which benefits are excludable. Most forms of social capital such as trust cannot be characterized as public goods, in the sense that its benefits are distributed homogeneously within a population. One might describe generalized trusting within a community as a form of social capital but trust itself, is a specific property of a dyadic relationship: A trusts B to do X (Hardin 1993). My ability to


deepen trust within a dyadic relationship--A trusts B in a range of situations, rather than in a specific instance X--or to extend it to other relationships, is a learned response. As Hardin (1993: 508) puts it: 'Trust has to be learned like any other generalization. Insofar as my trust is a generalization in the face of new persons, this merely means that the capacity to trust, the optimistic Bayesian estimate of trustworthiness, is learned perhaps from long experience.'

But my propensity to trust is also dependent on factors outside my own experience-namely, my family upbringing and the community in which I have been brought up. The amount of generalised trusting I have encountered will undoubtedly affect my estimates of the benefits of trusting and others' trustworthiness. Whereas Hardin is right to contend that trust is not a form of human capital, it is also true that the generalized trusting that is so critical to foster trust in dyadic relationships could itself be seen as a form of social capital. My ability to trust someone is partly dependent on my experience with similar people and possibly also the experience of people I trust with that third party. Thus, Uzzi's (1996a) study of the New York apparel economy found that, as in third-party referral networks, the stock of social capital existent in a business relationship was often used in a new, connected relationship by applying expectations and norms from an existing relationship.

In short, social capital cannot be owned, controlled, or appropriated by an actor. But that doesn't mean to say that social capital is a simply a resource uniformly distributed and awash in the social structure of a particular community. It is at once located in concrete, identifiable dyads and as a generalized resource that actors can selectively draw for their social interactions.

Accessibility is an associated dimension to controllability. It denotes the ability of actors to access a resource even when they are unable to secure control over it. Information, for example, is one type of economic resource that presents specific problems in terms of control and access. Access to social capital requires active participation in the social networks and engagement in specific relationships. Krackhardt (1992), Lazega and Lebeaux (1995), and Lazega (this volume) provide good examples of how in formal organizations social capital is differentially distributed among different social networks and how individuals might leverage their social capital differently depending on specific circumstances. Krackhardt (1992) is concerned with the strength of strong ties or what he describes as phi/os relationships that have are based on long-standing friendships and affection. In a casestudy, exploring an unsuccessful attempt to unionize an electronics firm in California, Krackhardt attributes the failure to unionize the plant to the power of individuals, based on their friendship networks, in opposing the move. Whereas the advice networks reflected routine and workflow knowledge and influence, the phi/os network became more important in a crisis context, when advice was sought from those one trusted (friends).

Lazega and Lebeaux's (1995) study of a law firm uncovered partly overlapping social networks based on different factors (friendship, advice, coworker) and showed that actors leverage their social capital rather selectively, depending on the particulars of a situation. Lazega and Lebeaux (1995) research focused on how a focal actor used an intermediary to influence a target actor in a range of situations.


The results of the research demonstrated how focal actors drew selectively on their social capital by using different social networks for different purposes and were concerned about limiting the negative effects of leveraging that social capital. However, this research also showed how the focal actors were often forcing intermediaries to draw on their social capital in less discriminating and selective ways.

Smart' s (1993) study of the use of guanxi in China is at once a good description of how social capital is produced and reproduced within close-knit networks of relationships, and also a good example of how it takes time and investment in relationships to build up social capital in a particular network. As knowledge of the rules of guanxi in China are difficult for an outsider to comprehend, many Hong Kong investors use social connections as intermediaries to contact local brokers with reputations for being able to solve problems (Smart 1993: 403).

Lastly, social capital is not a tradeable resource in the sense that it can be easily exchanged for other resources. Although it is tempting to use the language of debits and credits in the exchange of obligations, favors or trust, these exchanges are characterized by the absence or accomplished concealment of explicit claims by the possessors of social capital when attempting to cash in their credits as well as the nonexistence of third-party enforcement to reclaim bad debts. The logic of social capital is thus the logic of making calculativeness, self-interest, profit, accumulation, and so on taboo (Bourdieu 1994).

So one party might build social capital in a particular relationship by a gift presentation or trusting gesture, but there is no guarantee that the debt will ever be repaid or, if repaid, the exchange will be equitable. But, to the extent that one invests in gift giving and disinterested gestures, this will have positive outcomes via building up status and reputation in a particular community or group. If reciprocation is not direct, benefits can still be accrued by others' recognition of one's generous disposition. But as Smart (1993: 396) remarks, we have had experiences of unreciprocated gifts and invitations or noted the absence of acknowledgment of social debts.

In economic life, the notion of gift presentation and exchange of obligations is more nuanced and less clear cut than in the examples given above. Often, the presentation of gifts in the social exchanges that accompany economic exchange are part of specific, particularistic relationships between roleoccupants (e.g., salesmanpurchasing agent). The social capital built in these social relationships may be recognized as institutional as well as personal social capital. However, this form of social capital may only be recognised when individuals leave roles or take with them a range of relationships (e.g. , customers) to a different organization. 13

In practice the picture may be more complicated since organizations are related not simply through economic exchange but also through business-interest associations, joint suppliers or customers, and their members' activities in professional associations, industry forum, and technical standards committees (Hakansson and Snehota 1995). Studies of information transfer and social networks demonstrate how social capital is created and consumed in a variety of industry contexts. Saxenian' s (1994) study of the evolution of the computer and semiconductor industries in Silicon Valley and Route 128 ascribes the comparative


success of Silicon VaHey to informal and densely patterned communication networks between communities of practice, based on professional aHegiances and reciprocity of information exchanged, that provide an efficient and rapid source of learning.

In short, social capital as a resource in economic life is characterized by a continuum ranging from areas where the absence of calculativeness and self-interest is paramount to areas where the tradability of favors, obligations, and reciprocity in information exchange is more readily accepted. In all cases though, the borderlands between specific reciprocity and equity in the exchange, and performances where instrumental ends and calculativeness are either absent or successfully concealed, are contingent performances subject to continuous negotiations within concrete relationships, rather than fixed and socially prescribed rules.

Relationship to Other Resources and Activities The first dimension in this set is that of integrity. It measures the extent to which resources are simple or compound; that already comprises a mix of resources or else representing some fundamental unit of resource that cannot be subdivided. Social capital provides an excellent example of a compound resource that is impossible to disentangle from the context in which it is produced and consumed. As Smart (1993: 393) argues social capital is social in the sense that it cannot be possessed and its existence is contingent on the reproduction of the concrete social relationships within which it resides. At the same time, social capital in the form of obligations or connections constitutes the very raw material through which social relationships within groups or communities are reproduced and maintained.

The compound nature of social capital is intrinsically linked to the nature of its creation and maintenance. The creation of some forms of social capital such as trust can be regarded as fortuitous by-products of experiences over which the individual may have little control or even did not undertake (Hardin 1993: 525). An individual may grow up trusting and reproducing trust in his or her social relationships, as a result of growing up in a supportive environment and learning through repeated experience that trusting pays off handsomely. Similarly, Putnam's (1993a) argument is that social capital is created and reproduced in communities where associations proliferate, membership overlaps, and multiplex relationships are formed. 14

A dimension widely acknowledged to be important in understanding resources is that of versatility. Resources may be versatile or specific in nature. In this context versatility refers to the ways in which a resource may be combined with other resources or activities. In practice, versatility is unequally distributed among resources and, in any case, is not regarded as an intrinsic property of resources but a function of their use-namely, in the context of exchange activities.

Social capital provides an excellent example of a resource that may be both specific and versatile, depending on the context of use and its form. For example, the existence of an obligation in a dyadic relationship is narrowly confined to the context of that relationship and the ways in which it can be repaid forms part of the atmosphere and tacit understandings prevailing in that relationship. Similarly, placing trust in somebody to do something is often a highly particularistic judgment that is embedded within the joint history and the atmosphere of a relationship.


However, belonging to a particular community, group, or clique and drawing on the social capital accorded to members may be a highly valuable and versatile resource. Examples of uses to which social capital acquired in this manner include getting a job, landing a contract, or getting a client's trust (Waldinger 1995).15 In particular, social capital built across overlapping relationships between the same individuals (e.g., as buyer and seller, friends, professional colleagues) may provide a highly versatile resource that can be leveraged across a range of contexts. Coleman (1990) argues that social capital is often the by-product of membership in voluntary associations and the social capital thus formed can be harnessed for other collective action ends. Taylor and Singleton (1993) claim that the endogenous solution of collective action dilemmas is greatly facilitated by the existence of social capital lodged in multiplex relationships.

Similarly, clear-cut typifications of an individual as member of a group or organization, may shut off the very same opportunities it affords generously to insiders. Thus Waldinger' s (1995) study of the New York construction industry and its ethnic niches showed that black contractors often had no choice but to restrict themselves to public works contracts since all the other contracts were stitched up in social circles and activities (e.g., golfing, boating) from which they were excluded.

The third dimension in this set is that of complementarity. Some resources are easier to combine with other resources, and with activities, than others are. Resources may be competitive as well as complementary even within the same firm. Complementarity is a measure of the processes of combination. It is concerned with the ease with which combinations with a specific resource may, in general, be carried out. The complementarity of human capital and physical capital, for example, is well documented in the resource-based theory of the firm, starting with the pioneering work of Penrose (1959).

However, the complementarity of social capital with both physical capital and human capital has received less attention. Coleman (1988) uses family upbringing and relationships within the community in which the family is located as examples of how social capital influences the creation of human capital. Waldinger (1995: 560) argues that Coleman' s examples only illustrate the indirect effects of social capital on the development of human capital and that a more compelling case could be made of how social capital is constitutive of all human capital. Learning through socialization and participation in a community of practice such as the building workers that Waldinger studied, are good examples of how the formation of human capital is enmeshed with social capital. Ostrom (1994) provides a rare example of a study where the interaction of social and physical capital is explicitly considered in collective action problems.

Burt (1992) provides the most forceful argument for the belief that social capital is both complementary to other forms of capital as well as being capable of being leveraged for the benefit of an individual or organization. For Burt (1992) in firms that provide services, for example, there are people valued for their competencies in providing expertise and ability to deliver a quality service and those (the rainmakers) who are valued for their ability to deliver and handle clients. The former represents financial and human capital, whereas the latter represent social capital. In short, different forms of capital are highly complementary and divisible on the basis of a


sharp division of labor and endowment of competencies even if, as Burt states, 'social capital is the final arbiter of competitive success.' In a world where players can easily be matched on their endowments of human and financial capital, social capital narrows down the pool of individuals who can compete successfully for new opportunities. However, Burt (1992, 1997) privileges the notion of structural holes or a relationship of nonredundancy between two contacts, and thus, in his view, social capital is effectively reduced to the instrumental use of social relationships for the pursuit of entrepreneurial opportunities to transform network structures to one's own advantage.

The final dimension in this set is understandability. It is a measure of how easy it is to comprehend the nature of a resource or even to recognize a resource at all. Some resources are easily recognized as such and are well understood. Most forms of physical capital come into this category. They are tangible, controllable, and easily evaluated. Other resources, human capital included, do not share these properties. They are intangible and cannot easily be valued. As mentioned previously, organizational competencies such as routines fall into this category. Exchange relationships and managerial skills are examples of this kind of resource.

If a resource can readily be understood, then it is more likely that the ways in which it can be created and maintained as well as combined with other resources will be well understood and hence exploited successfully. Virtually all forms of social capital fare badly on this score. Coleman (1990: 312-313) argues that there are few cases where social capital is well understood as a resource and can be created as a direct result of investment of actors who have the aim of receiving a return from their investment. The creation of formal organizations and the associated authority structures with well defined obligations and expectations associated with role occupancy is presented as a case of investment in social capital, in the same way investment in human capital is associated with the appointment of individuals to specific roles. But Coleman's formulation seems to conflate notions of power, vested in authority structures, and social capital as the ability to induce others to act in one's interests through the leverage of resources available to the agent. Whereas power can ultimately be seen as resting on the authority to command the actions of others and enforce sanctions for noncompliance, social capital tends to rely on implicit rather than explicit claims to reciprocation and cannot resort to formal sanctions on recalcitrant targets.

In hierarchies, as Miller (1992) cogently argues, managers do not spend their time writing and enforcing contracts defining obligations and expectations associated with specific roles anymore than employees spend their time maximizing self-interest within the constraints imposed by those contracts. Norms of cooperation and trusting relationships are as critical to the functioning of an organization as a supplement, but not necessarily as a substitute for incentive systems based on formal contracts. In short, notions of power and social capital are as crucial to the functioning of hierarchies as private and public orders are vital to the functioning of markets.

In summary, the character of most forms of social capital makes it a poorly understood resource. Its creation is often dependent on by-products of other activities, mostly escaping the direct control of actors, its leverage is dependent on


contextual performances, it cannot be easily transferred from context to context and its linkages with other resources cannot easily be disentangled. Moreover, the concealed or implicit nature of leveraging social capital, often attached to the history and atmosphere of particularistic relationships, makes it even less understandable as an economic resource.

CONCLUSIONS

The increasing attractiveness and popularity of the notion of social capital can be understood by reference to the appealing logic of the capital-in-general metaphor and for the possibilities it brings to breach some of the artificial divides between the economy and society. This appeal rests as we have argued, in the often insufficiently understood mechanisms of investment, accumulation, and benefits accruing from the use of capital. In this sense, as others have noted (see, e.g., Tarrow 1996) social capital has an appealing normative component. Creating, maintaining, and growing social capital, so we are told, is the key to healthy societies, high-performing organizations, and prosperous economies. In this sense, social capital can be seen as important resource since its existence obviates the need to allocate other resources to the formalization of rules, coercion, surveillance, and information gathering to supplement private norms.

On the other hand, social capital provides a vehicle to breach the divide between the social and economic worlds and its attributes make it an all-encompassing notion, connecting dyads to wider social structures and back again. Thus social capital-although conceived as a property of wider social structures in the form of organizational structures, occupational communities, trade or civic associations, and so on or based on categories such as ethnic groups-is manifested through concrete outcomes in ordinary social practices and relationships. In this sense, social capital can provide an antidote to more generic notions of embeddedness, understood simply as a static backcloth of social structures encapsulating economic exchange but themselves immune from reciprocal influences.

The argument in this chapter has been that neither the normative nor the structural attractiveness of social capital as a notion can mask the dangers of engaging in too cursory an application of the capital-in-general metaphor to the study of the interaction of social structures and economic action. Social capital understood as a resource that can impinge on economic action is perhaps not unique in some of its attributes, but its heterogeneous manifestations, intangibility if not purposeful invisibility, its compound nature, and poorly understood logics of accumulation and use make it a counterintuitive metaphor.

In this chapter we have attempted to be more specific about the nature of economic resources and subject the social capital metaphor to a more systematic appraisal of its attributes as a resource impinging on economic action. In so doing, we effectively moved the notion of social capital away from the notion of capital-ingeneral and instead stressed its intangible, dynamic, and relational nature. This allows us to move beyond notions of social capital as an abstract property of social structures to explore its concrete outcomes and contextual performances and link these to the logic of production and reproduction of those social structures where social capital is deemed to reside. This last point is perhaps worth stressing anew,


even if a full discussion of this point is outside the scope of this chapter social capital consists of claims to reciprocation and solidarity contained in concrete episodes and contextual performances. What governs these episodes, however, are that explicit, calculative claims are normally excluded from the performances within which they are made and thus the leverage of social capital has to be read and interpreted anew in every single episode.

Finally, we wish to issue a cautionary note to ourselves as much as to others, regarding the need to resist the facile prescriptions that inevitably spillover from too many of the romantic accounts of the role of social capital in socioeconomic life. Even if we begin to understand a little better the notion of social capital, the operation of policy instruments directed at building, maintaining, or destroying social capital remain as slippery, open-ended and uncertain as any other policy instrument.16 Hopefully, this volume will contribute to advancing our understanding of these issues.

NOTES

I. Recent work in the sociolegal field has recently taken an interest in the interaction between the law and infonnal social nonns in law making and enforcement (see, e.g., Pildes 1996; Posner 1996). 2. On this topic see also the belated discovery of the existence of a variety of blends of trust and contract in buyer-supplier relationships in some sociolegal and economics literature (e.g., Burchell and Wilkinson 1997). 3. See F1igstein (1996) for a comprehensive treatment of the role of political embeddedness in the evolution of markets. 4. See also Podolny (1994) and Uzzi and Gillespie (this volume) on this topic. 5. Ostrom (1994: 527-528) defines social capital simply as the arrangement of human resources to improve flows of future income for at least some of the individuals involved in the production of that capital. Similarly, Portes and Sensenbrenner (1993: 1323) define social capital as the relevant expectations concerning economic action within a collectivity, affecting the goal and goal-seeking behaviour of its members. 6. For a parallel and insightful distinction between social capital as lodged in personalised, dyadic relationships and depersonalized, institutional relationships, see Pennings and Lee (this volume). See also Knoke (this volume). 7. For a similar argument on social capital as a public good see Putnam (1993). 8. On the role of third parties in exchange see Nooteboom (this volume). 9. See also Pennings and Lee (this volume) on the importance of relationships that span organizational boundaries for corporate social capital. 10. See Pennings and Lee (this volume). See also Burt (1997) for an argument on how social capital is becoming increasingly important in organizational life and managerial perfonnance. II. For critical reviews of Putnam's perspective on the impact of social capital on political and economic life see Levi (1996), Tarrow (1996), and Kenworthy (1997). 12. See Lazega (this volume) on the notion of the finn as a multiplex, generalized exchange system and on the role of reciprocity in this exchange system. Scharpf (1993: 153-154) makes a related point concerning generalized trust whose existence 'pressuposes a generalized willingness to cooperate even in constellations where cooperation is not advantageous and is easily destroyed by the pursuit of self-interest at the partner's expense. But where it exists, generalized trust is enonnously advantageous. It will enable rational actors to enter into vulnerable positions, and to engage in high-risk (and potentially high gain) mixed motive transactions under conditions of incomplete infonnation.' 13. See Pennings and Lee (this volume) for a more comprehensive discussion of these issues. 14. The importance of multiplex relationships for the emergence and maintenance of social capital is highlighted in several chapters in this volume-see, in particular Pennings and Lee, Knoke, and Lazega. 15. See also Flap and Boxman (this volume). 16. See also Leenders and Gabbay (this volume).

•

Social Capital by Design: Structures, Strategies, and Institutional Context

ABSTRACT

4 Wayne E. Baker

David Obstfeld

We examine social entrepreneurship from a structural perspective, distinguishing between two structures of social capital and their associated entrepreneurial strategies: structural holes and the 'disunion' strategy versus social cohesiveness and the 'union' strategy. These two strategies represent alternative ways social entrepreneurs access and mobilize the resources inherent in the structure of a social network. The disunion strategist exploits structural holes between alters by keeping them apart; the union strategist creates value by bringing together disconnected alters. The frequency, legitimacy, and success of each strategy depends on the 'design' of the institutional context in which social entrepreneurs operate. Disunion strategies tend to occur in organizations and markets characterized by sparse, disconnected, and differentiated networks, coupled with competitive rules of exchange, opportunism, and an individualist orientation; union strategies tend to occur in organizations and markets characterized by dense, connected, and undifferentiated networks, coupled with cooperative rules of exchange, norms of reciprocity, and a collectivist orientation. We illustrate the distribution of triadic strategies in a specific institutional context by taking a triads census of alliances in the global automobile industry and testing the structural hypothesis about the use of disunion and union strategies.

INTRODUCTION

Ever since Schumpeter (1934: 156) identified entrepreneurship as a 'vehicle of continual reorganization of the economic system,' entrepreneurship has been recognized as playing a key role in catalyzing change, promoting innovation, and

Social Capital by Design - 89

enhancing productivity in the economy at large. For example, the continuous process of 'breaking away' from existing firms to create new ones is a prime engine of the growth and economic development of cities (Jacobs 1965, 1970). The role of entrepreneurs as intermediaries between corporate actors (firms, associations, governmental bodies) is well known (e.g., Coleman 1990: 180-188). However, entrepreneurship is also recognized as a creative activity that occurs inside organizations, an activity that is critical to their survival and health (e.g., Burgelman 1983a; Kanter 1983a). Entrepreneurial action within organizations is called intrapreneurship (Pinchot 1985), corporate venturing, or corporate entrepreneurship (Burgelman 1983a, 1983b).

Structural sociologists shifted the study of entrepreneurial behavior away from a focus on traditional entrepreneurial activities, such as using economic capital to start new ventures, to the analysis of the strategic use of 'social capital' both inside and between organizations (e.g., Burt 1992; Bourdieu and Wacquant 1992; Coleman 1988, 1990). Our chapter follows in this structural tradition. We offer a new theoretical distinction between two structures of social capital and their associated entrepreneurial strategies-structural holes and the 'disunion' strategy versus social cohesiveness and the 'union' strategy. These two strategies represent alternative ways social entrepreneurs access and mobilize the resources residing in a social network.

We use the phrase 'social capital by design' in two senses that together capture the essence of our argument. First, social capital is related to the 'design' of an institutional context. A context characterized by sparse, differentiated, and disconnected networks, for example, yields social capital in the form of structural holes. A context characterized by dense, integrated, and connected networks, in contrast, yields social capital in the form of social cohesiveness. Second, social capital can be created by intention, that is, by the strategic moves of an individual entrepreneur or the deliberate manipulation of organizational and interorganizational structures. Change agents, for example, can change the structure of social capital (and the entrepreneurial strategies used to access it) by altering the design of the institutional context.

STRUCTURES AND STRATEGIES

The Structural Approach to Entrepreneurship Most attempts to understand and promote entrepreneurship have examined entrepreneurs and their behaviors at the individual level, striving to define the key traits and characteristics of successful versus unsuccessful entrepreneurs (Gartner 1989; Low and MacMillan 1988). For example, McClelland (1967) argued that the 'need for achievement' is a key psychological characteristic of the successful entrepreneur, but empirical research has not supported a link between the need for achievement and, say, the decision to start a new business (Sexton and Bowman 1985). Locus of control and propensity for taking risks have been proposed as possible distinguishing characteristics, but research has not provided much empirical support of these hypotheses (e.g., Brockhaus 1982; Sexton and Bowman 1985; Gasse 1982). Tolerance of ambiguity does appear to distinguish entrepreneurs from managers (e.g., Schere 1982; Sexton and Bowman 1985). In general, the concerted


attempt over the past decades to build a personality profile of successful entrepreneurs has not yielded insights into the unique personalities of entrepreneurs. Based on their review, Low and MacMillan (1988: 148) conclude that, ' ... at a ... fundamental level, it can be argued that the wide variations among entrepreneurs make any attempt to develop a standard psychological profile futile. One is struck by the appropriateness of Gartner's (1985) observation that 'there is as much difference among entrepreneurs as between entrepreneurs and non-entrepreneurs."

The failure of personality research to identify key characteristics suggests that the individual level of analysis may be inappropriate for understanding entrepreneurial action (Gartner 1989). These efforts suffer from the neglect of social structure and the complex relationships between the individual, corporate actor, and environment (Martinelli 1994). Others have proposed a behavioral focus for entrepreneurial research, emphasizing the processes associated with entrepreneurial behaviors (e.g., Gartner 1989, 1990). The structural approach, which we use here, is consistent with such a behavioral focus (e.g., Aldrich and Zimmer 1986; Burt 1992; Krackhardt 1995; see also, Freeman, this volume, for a similar perspective).

The entry of structuralists into the study of entrepreneurial behavior shifted the focus away from traditional entrepreneurial behaviors, such as starting new businesses or founding firms, to the examination of the strategic creation and use of 'social capital' both inside organizations and among organizations (e.g., Burt 1992; Bourdieu and Wacquant 1992; Coleman 1988, 1990). 'The central proposition of social capital theory,' summarize Nahapiet and Ghoshal (1998: 243), 'is that networks of relationships constitute a valuable resource for the conduct of social affairs .. .. ' We use the term 'social entrepreneur' to describe individuals or corporate actors who access and mobilize the social capital inherent in organizational networks, as opposed to the 'traditional entrepreneur' who uses economic capital to create new businesses and firms (of course, traditional entrepreneurial activities often involve the use of both social and economic capital). The social processes we discuss apply to activities both inside organizations and between organizations (what we call 'institutional contexts' below). We focus on 'corporate social capital'-social capital that resides inside, between, or among organizations-in contrast to other locations of social capital, such as families and communities (see, e.g., Coleman 1988; Putnam 1995a).

The definition of social capital, along with the role and activities of the social entrepreneur, have been the subjects of considerable debate. l Clarifying the dimensions of social capital is a high research priority (Putnam 1995a). Nahapiet and Ghoshal (1998), for example, make useful distinctions between three dimensions of social capital: the 'structural dimension' (the configuration of social networks), the 'cognitive dimension' (shared systems of meaning, narratives, language), and the 'relational dimension' (norms, trust, obligations). Our structural view of social entrepreneurship emphasizes the first dimension, exploring the basic structures of social capital (structural holes versus social cohesiveness), their corresponding entrepreneurial strategies (disunion versus union), and the relationship between institutional context and strategy. Our distinctions, as we elaborate below, help to clarify the debate and resolve some of the confusion about the structural approach to social capital and social entrepreneurship. Though we


emphasize the structural dimension of social capital, we acknowledge the importance of the cognitive and relational dimensions.

Two Structures of Social Capital The triad of social actors, composed of 'ego' and two 'alters,' is the basic unit of analysis in social entrepreneurship. Structures larger than the triad are possible, of course, and often occur; however, they are based on the triad as the fundamental building block. For example, 'communities of trust' are generalizations of the threeactor structure (Coleman 1990: 188-189). Social actors can be persons or corporate actors: people acting as individuals, people acting as agents or representatives, organizational subunits (such as teams or departments), organizations, and even governmental bodies, states, and nations. (For example, the triad is a basic unit used in analysis of geopolitical relations.) Given our focus on corporate social capital, however, we are interested primarily in two types of social actors: individuals who are members of organizations and organizations themselves. Our concepts apply to both types.

The structural basis of entrepreneurial action was suggested by Simmel (1950: 154-162), who stressed the importance of the 'third element' in group dynamics. Simmel argued that the introduction of a third party fundamentally alters the social dynamics of dyadic ties (see also Nooteboom, this volume). Of particular interest is the triad type Simmel (1950) called tertius gaudens-'the third who enjoys' benefits by his or her position between two disconnected parties. These two parties, because of their unfamiliarity with each other, can be manipulated to the third party's benefit. Simmel's tertius gaudens is the basis of Burt's (1992) influential theory of structural holes. Burt argues that a sparse egocentric network with few redundancies (few members of the network know each other) is a social structure rich in structural holes. A hole exists between two people (alters) if they are not connected to each other but share a tie with a common third party (ego). This structural arrangement puts the third party in the role of the tertius gaudens who can take advantage of the two disconnected persons (or corporate actors) for private gain. Burt (1992) found, for example, that managers in a large corporation who have networks rich in structural holes were promoted faster and at earlier ages, compared with otherwise similar managers whose networks lacked structural holes. Burt (1992) also found that firms with interorganizational networks rich in structural holes earned a higher rate of profit, compared to firms without these structural advantages.

Some argue that Burt's (1992) structural holes theory is an alternative to, rather than an example of, a theory of social capital (e.g., Walker, Kogut, and Shan 1997: 112). These arguments emphasize the 'relational' dimension of social capital (Nahapiet and Ghoshal 1998), which is absent or undeveloped in Burt's theory. From the 'relational' view, social capital exists in a relationship between two people (or two corporate actors) if they develop personal bonds, attachments, and trust. '[A] close working group of [graduate students] working on the same problems constitute social capital for each of them for his [sic] graduate training' (Coleman 1990: 170). This 'relational' view also stresses the restraints on opportunism maintained by social capital that allow cooperation take place (Walker, Kogut, and Shan 1997; see also, Granovetter 1985).


Rather than pitting structural holes theory against a relational definition of social capital, we believe it is more theoreticalll productive to consider these as alternative views of social capital (Obstfeld 1997) and to concentrate instead on delineating and analyzing their characteristic social structures. Burt's (1992) theory emphasizes only one of several structures of social capital. Coleman's (1988: 98) original definition is broad enough to encompass other structures: 'Social capital is defined by its function. It is not a single entity but a variety of different entities, with two elements in common: they all consist of some aspect of social structures, and they facilitate certain actions of actors-whether persons or corporate actors-within the structure. Like other forms of capital, social capital is productive, making possible the achievement of certain ends that in its absence would not be possible.' Similarly, Bourdieu's (Bourdieu and Wacquant 1992) definition is broad enough to include multiple structures of social capital. For Bourdieu, social capital is 'the sum of the resources, actual or virtual, that accrue to an individual or a group by virtue of possessing a durable network of more or less institutionalized relationships of mutual acquaintance and recognition' (Bourdieu and Wacquant 1992: 119).

Social capital 'inheres in the structure of relations between and among actors' Coleman (1988: 98). Structural holes theory emphasizes a structure of social capital characterized by sparse networks and few redundancies. For Burt (1992), social capital resides in the patterned absence of ties. This view is consistent with the argument that social structure is defined more by the patterned absence than presence of ties (White, Boorman, and Breiger 1976). What is the alternative? We argue that social capital also inheres in the structure of social networks as the patterned presence of ties. The alternative to the holes view of social capital is what we call social cohesiveness, where the structure of social capital is characterized by dense networks and mUltiple redundancies. This structure of social capital corresponds to another of Simmel's triad types, the third party who acts as a mediator or 'non-partisan' to create or preserve group unity: The non-partisan either produces the concord of two colliding parties, whereby he withdraws after making the effort of creating direct contact between the unconnected or quarreling elements; or he functions as an arbiter who balances, as it were, their contradictory claims against one another and eliminates what is incompatible in them' (Simmel 1950: 146-147).3 Also see Nooteboom (this volume), who describes six roles of the third party (the so-called 'go-between').

Common examples of this structure of social capital include real estate brokerage, literary agency, and political mediation.4 For example, some Washington lobbyists specialize in the introduction of corporate actors to public officials (such as executive agency officials or congressional representatives) (Coleman 1990: 180-182). Coleman (1990: 180) calls the third parties in these social structures 'intermediaries in trust.' The third-party intermediary is able to bring together the other two parties because each one trusts the intermediary. This structure of social capital is pervasive in society; as Coleman (1990: 184) describes:

This form of intermediary exists in all areas of social life. For example, professors write letters of recommendation to prospective employers about students, and persons seeking a job or a loan list other persons who will recommend them. The acceptance of a recommendation by a prospective employer or creditor is a placement of trust in the


judgment of the intermediary, which allows a placement of trust in the ability of the prospective trustee to perform as expected. If the latter defaults, then the trustor's trust in the intermediary's judgment is reduced.

A social actor can advertise its position in the social cohesiveness structure of social capital as a valuable resource. Consider, for example, the language used by Mayfield, a venture capitalist firm, in its promotional brochure; as quoted by Freeman (this volume): 'Because of our long association with a large number of successful companies and entrepreneurs, a relationship with Mayfield is highly regarded. It can enhance the credibility of a young company with potential customers, vendors and employees, and with other financial institutions.' Similarly, professional service firms, such as advertising agencies and investment banks, 'sell' access to the social capital inherent in their networks. Indeed, advertising agencies that occupy a central position in the market are likely to be kept by their corporate clients, indicating the value clients place on this structure of social capital (Baker, Faulkner, and Fisher 1998).

The level of trust in the social cohesiveness structure is probably higher, on average, than the level of trust in the tertius gaudens arrangement. This is one reason why some critics argue that structural holes theory is not a theory of social capital (see above). Because the tertius gaudens exploits the structural hole between two alters, the level of trust is presumed to be low. This is not necessarily so. Each alter may trust ego, even though the alters are unaware of each other's existence; in other cases, the alters may prefer to remain out of contact with each other, relying instead on their trust in ego. Moreover, the issue of trust in the social cohesiveness structure is not as unambiguous as it might seem. Sometimes an 'intermediary in trust' runs the risk that he or she will be circumvented or 'cut out' by opportunistic alters. For example, the risk that the principals in a real estate transaction might consummate the deal in secret, saving the commission owed to the broker, is so high that the standard legal contract between a real estate broker and seller contains protections against such actions.5

The distinguishing feature between the two structures of social capital is not trust, but the answer to this question: What does the social entrepreneur do with the gap between alters? While one social entrepreneur exploits the structural hole, keeping alters apart, another social entrepreneur may choose to close the gap, bringing together the two alters. These actions represent the two basic entrepreneurial strategies for accessing and mobilizing the social capital inherent in social networks. We next describe and illustrate these strategies, followed by our analysis of the relationship between entrepreneurial strategies and institutional context.

Two Entrepreneurial Strategies Strategy is used here to refer to a social entrepreneur's plan of action for using network structure to access and mobilize social capital. Burt's (1992) conceptualization of social capital as the structured absence of ties favors what we call 'disunion strategies.' In this strategy, the social entrepreneur generates 'profit' by taking advantage of the disconnection of the two parties. The disunion strategy is


Alter I Alter 1

Ego Ego

Alter 2 Alter 2

Disunion Union

Figure 1. lIlustration of disunion and union strategies

illustrated in Figure 1. As shown, ego (the tertius gaudens) is linked to two disconnected alters. The hole is represented by the blank space between alter 1 and alter 2, and enables ego to play the two alters against each other or to secure a valued resource from one and provide it to the other, extracting a profit in the exchange. In such a case, ego benefits from the absence of a connection between the two alters, may act to keep the alters apart, and at the very least, chooses not to introduce the two alters. For disunion strategies, value is created by the exploitation of these structural holes; value, therefore, is a 'private good'-the benefits of social entrepreneurship accrue to the third party (ego).

The view of social capital as social cohesiveness leads to the alternative entrepreneurial strategy we call 'union strategy'-illustrated in Figure 1. In this case, ego (Simmel's non-partisan or arbiter) is linked to two alters who are disconnected or in conflict. Ego 'closes' the gap between alters by bringing them together or by resolving their differences. This suggests a sharing or exchange of resources. The union strategy produces what Coleman (1988: 107) calls 'closure'-a social structure that creates the conditions for the enforcement of norms through 'sanctions that can monitor and guide behavior.' The combination of norms and trust that emerge under these structural conditions facilitates additional use ofthe union strategy.

Disunion Strategies In Action Disunion strategies are common in situations where the formal differentiation of the organization presents the social entrepreneur with many temptations to play one person (or department) against another person (or department). Burt's (1992) original study of managers and structural holes was conducted in one such organization. Consider, for example, an entrepreneur with ties to a person in the sales department who collects current information about customer needs, and to another person in the marketing department who is desperate for customer input for a new product. The entrepreneur can position him or herself to sales as someone with influence over the company's product development process and to marketing as a source of new information (Obstfeld 1997). With this disunion strategy, the entrepreneur seeks to benefit without ever introducing the two alters, a move that would eliminate the advantageous position. A similar case is the classic role of the boundary spanner who links two disjoint groups (Friedman and Podolny 1992).

Competitive markets are principal locations of disunion strategies at the interorganizationallevel. Simmel (1950) notes that the market is a prime example of


the tertius gaudens strategy writ large. Here, disunion strategies are characterized as 'rivalry,' where two or more sellers vie 'for opportunities of exchange' with a buyer (Weber 1978: 63; Swedberg 1994: 271). For example, Coca-Cola Company maintains relationships with six different advertising agencies (Baker, Faulkner, and Fisher 1998: 149), playing one advertising agency against the other in an elaborate disunion strategy. The practice of competitive bidding is based on the disunion strategy, where multiple sellers are pitted against each other. For example, disunion strategies are evident in the garment industry studied by Uzzi (1996a) in cases where dress manufacturers (buyers) select contractors (sellers) on the basis of price alone to effect discrete, nonrecurring exchanges. Of course, sellers in an industry suffering from intense competition caused by the buyers' relentless use of the disunion strategy may become motivated to collude, employing an illegal union strategy to counterbalance the power of buyers (Baker and Faulkner 1993).

Disunion logic drives the avoidance of ties in a competitive market. Competing companies will not use the same supplier because doing so would put the supplier in the structural position of the tertius gaudens. For example, General Motors avoids using the investment bank Goldman Sachs because its chief American rival, Ford Motor Company, uses Goldman as its main bank (Baker 1990). Similarly, corporate competitors avoid using the same advertising agencies, citing 'conflict of interest' as their rationale (Baker, Faulkner, and Fisher 1998).

The social structure of disunion strategies can be fluid, particularly in dynamic markets. When AT&T and China began negotiations to install an undersea cable system to provide a telecommunication link between China and the U.S., the entrepreneurial leverage associated with the disunion strategy shifted from one actor to another. AT&T initially approached the negotiation as the owner and operator of international marine cables, enjoying an advantageous negotiating position (Glain 1997). Deregulation and other changes in the telecommunication industry, however, unleashed a variety of competitors (such as Baby Bell SBC Communications, Nynex, Britain's Cable & Wireless, and Japan' s Kokusai Denshin Denwa Co.) that the Chinese invited into the negotiations. One of China's key negotiators indicated, 'Our general policy is to not engage in projects that exclude other parties. We want to engage as many companies as possible on an equal basis' (quoted in Glain 1997). This approach nullified AT&T's disunion strategy by leveraging confidential information provided by the competing companies and the disconnections between them. Ultimately, the Chinese insisted on a 14-member consortium that included all of AT&T's major rivals.

Finally, direct exploitation of disconnected parties may represent only a fraction of the activities associated with disunion strategies. Some disunion strategies require constant effort and vigilance (perhaps even subterfuge) to keep alters apart and to maintain the structural conditions of disconnection, secrecy, and concealment. Pricefixing conspiracies, for example, require conspirators to deliberately maintain the ignorance of their corporate customers (Baker and Faulkner 1993). Some corporate actors attempt to prevent competitors' actions or the development of new regulations that would reduce or close the structural holes in markets, fighting to maintain the structural conditions that favor disunion strategies. For example, the strategic alliance of IBM and Apple was opposed by many competitors (Baker 1994a) who


viewed the potential union as a threat that would reduce their ability to compete for customers-that is, to operate using the disunion strategy.

Union Strategies In Action The formal differentiation of an organization does not always lead to disunion strategies. For example, Kanter (1983a: 141) describes the union strategy used by a corporate entrepreneur who created an internal alliance between sales, service, and product development:

He first wrote a memo to all of the sales people in his area, copying the district managers for service and products.. .. He then held a series of sales meetings, inviting commercial and service staff too .... [He] explained and reexplained the benefits of cooperation across the sales/service/products boundaries to people from each function. (Ashkenas et al. 1995: 17)

Similarly, Burgelman (1991) depicts how a corporate entrepreneur at Intel collaborated with two other product champions to develop RISC processor technology and to line up a customer base in advance of Intel's entry into the new market. Ashkenas et a!. (1995) describe union strategies used to transform General Electric's Retailer Financial Services into a 'boundaryless' organization. For example, a systems manager employed union logic to streamline a business process: 'Nastasi then brought together a group of systems, marketing, finance, and customer service people and challenged them to complete new customer conversions in a matter of days, not weeks.' In many cases, union strategies are used by senior managers targeting major organizational change. Union strategies are also initiated by relatively low-ranking employees with more modest objectives.

Union strategies exhibit variation in official support. General Electric CEO Jack Welch, for example, officially sponsored GE's Corporate Executive Council to stimulate the exchange of information and collaboration (Baker 1994a,b). In contrast, Hutt, Reingen, and Ronchetto (1988) describe the emergence of a new product through the collaboration of actors from multiple divisions along with key customers-all acting outside prescribed corporate guidelines. Similarly, the successful Intel intrapreneurs described above went against explicit corporate policy and had to disguise the RISC project as just another 'co-processor' until sufficient progress had been made (Burgelman 1991).

Union strategies also occur in markets. Some union strategies involve the introduction of initially unconnected suppliers or customers via common third parties. Starr and MacMillan (1990: 86) describe a case of an entrepreneur who introduces two suppliers:

A corporate entrepreneur saw an opportunity to connect two of his major suppliers while developing a new medical products business. One manufacturer, in the south, had cheap labor costs and good employee morale, but was nonetheless losing money due to lack of business. The other, in the north, had high labor costs and an employee shortage. By introducing these two manufacturers the venture manager reduced the costs of one and increased the sales of the other.

Uzzi (1996a: 679) describes a similar union strategy used in the New York garment industry:


One CEO explained how [a] tie formed between him and a manufacturer named 'Diana.' He said that his contact with Diana began when Norman, a close business friend of his and Diana's, asked him 'to help Diana out' in a time of need (cut her fabric at a special price and time), even though he had no prior contact with her .... [The CEO said] 'So why did I help her out? Because Norman asked, 'Help her out.'

The formation of strategic alliances-joint ventures, technology sharing, marketing arrangements, product development, and others-{;an result from union strategies. For example, Coming introduces alliance partners to each other, fostering creation of ties between them. AT&T Global Information Solutions convenes an annual conference in which its strategic allies meet in 'alliance fests' used to generate new alliances and associations (Baker 1994b). Digital Equipment Corporation (DEC) invites its alliance partners to a four-day conference to share information and foster alliance formation (Gulati 1995a). Similarly, the monthly meetings of the 128 Venture Group were convened by an entrepreneur and venture capitalist to establish a place where venture capitalists, entrepreneurs, consultants, and management team candidates could meet and explore collaborations (Nohria 1992b).

The introduction of disconnected parties is only a fraction of the activities associated with union strategies. Union activities include investing in already established ties, cultivating ongoing collaborations, and maintaining the general structural conditions that facilitate union strategies. Some union strategies are selfsustaining, such as those associated with the trading groups in the industrial districts of north central Italy and southwestern Germany (Powell 1990). In other cases, however, such as real estate brokerage, the repeated cultivation of new alters is necessary to stay in business. Of course, the real estate broker with a good reputation and lots of contacts enjoys the benefits of union strategies in reverse, as satisfied customers refer new alters to the broker.

These examples illustrate the general principle that union logic drives the selection of ties, whereas disunion logic drives the avoidance of ties (as described above). Disunion strategies proscribe the use of common third parties, such as suppliers, while union strategies prescribe the use of common third parties, such as alliance partners. These strategies may exist side by side in the same institutional context, which is a topic we take up in the following section.

INSTITUTIONAL CONTEXT

So far, we have not been very specific about the context in which social entrepreneurs operate. This was intentional, so that we could be clear about the distinctions between the two structures of social capital and their corresponding strategies. However, the relationship of structures and strategies to institutional context is critical. First, we argue that the nature, level, and forms of social capitaland therefore the strategies social entrepreneurs employ~epend on the structure and culture of the institutional context. This relationship holds in the institutional context of an organization as well as that of organizational fields, business sectors, industries, and markets. Second, we argue that the frequency, legitimacy, and success of an entrepreneurial strategy depends on its 'fit' or compatibility with the institutional context in which it is used.


Table 1. Institutional context and entrepreneurial strategies

Institutional Context Disunion Strategy Union Strategy

Structural Conditions

size large size small size

density sparse networks dense networks

connectivity disconnected networks integrated networks

formal differentiation many formal boundaries few formal boundaries

Cultural Conditions

rules of exchange competition cooperation

norms opportunism. distrust reciprocity. trust

orientation individualist orientation collectivist orientation

Nahapiet and Ghoshal (1998) argue that the 'firm' is a better institutional setting than the market for the development of high levels of social capital; because the firm is a 'social community' (Kogut and Zander 1996: 503), it enjoys an 'organizational advantage' over markets. However, some firms are organized and operated as 'markets,' and some markets are organized and operated as 'firms' (Eccles 1981; Eccles and White 1988; Stinchcombe 1985).7 Therefore, we make the bold assumption that the characteristics of 'institutional context' can be specified in such a way that they apply to both firms and markets. An advantage of our approach is the generalization of the concepts of corporate social capital across institutional levelsthat is, within firms and between firms. The relationship between institutional context and strategy is summarized in Table 1.

Structural Conditions By structural conditions, we refer to the 'network configuration' of an institutional context, consistent with Nahapiet and Ghoshal's (1998) definition of the 'structural' dimension of social capital. Standard network measures can be used to represent a configuration. Nahapiet and Ghoshal (1998) propose density, connectivity. and hierarchy. We substitute 'formal differentiation' for hierarchy. because an organization can be differentiated along three dimensions-spatial. horizontal, and vertical. Markets, too, are organized along these three dimensions, as economic geographers have documented. Formal divisions such as these, whether in firms or markets. create structural holes. We add 'size,' since the number of social actors influences the fragmentation of an institutional setting, and along with it, the number and extent of structural holes.

The structural conditions associated with disunion strategies are large size, sparse and disconnected networks, and many formal boundaries. The design of most large-scale, traditional organizations favors disunion strategies because it creates so many structural holes between departments, across levels, and between spatially separated operations. Formal differentiation, for example, hampers collaboration across boundaries (Lawrence and Lorsch 1986). Large size disfavors the integrated 'network' organizational design (Baker 1992a). Similarly, markets with many players are more fragmented than markets with few players, producing many structural holes. The alert tertius gaudens in a fragmented market generates profit by


arbitraging across these gaps in the social structure of trading (Baker 1984), just as the tertius gaudens in a fragmented firm generates profit by exploiting the disconnections inside the organization (Burt 1992).

The structural conditions favoring union strategies are small size, dense and integrated networks, and low differentiation (see Table 1). Union strategies are common in network organizations, for example, because these organizations are characterized by flexibility, lateral ties, and a high degree of integration across low formal boundaries (Baker 1992a). A dense network of customers, producers, and suppliers provides a structural basis for cooperation (Perrow 1992), such as the dense social networks of firms, local universities, community colleges, research institutes, financial institutions, trade associations, and regional governments in Silicon Valley (Saxenian 1991, 1994).

The biotechnology industry features dense and well-connected networks of interfirm cooperation as well (Walker, Kogut, and Shan 1997; in this volume see also Stuart, Smith-Doerr et al. and Omta and Van Rossum). These structural conditions are conducive to the formation of new alliances based on the union strategy. For example, the likelihood that two previously unallied firms will form an alliance increases with the number of third-party ties they have in common (Gulati 1995a). Similarly, the greater the number of research and development alliances and other types of collaborations a biotechnology firm has at a given time, the more diverse its future portfolio of ties will become (Powell, Koput, and Smith-Doerr 1996).

Cultural Conditions Cultural conditions can be defined in many ways. Indeed, the definition of culture itself is a subject of considerable debate and rival interpretations (DiMaggio 1994; Scott 1995). Rather than trying to resolve this debate, we propose a simple definition of cultural conditions-the institutionalized rules of exchange, norms, and social orientation in an institutional context-and focus primarily on the relationship between strategy and cultural context.s

Rules of exchange are shared social understandings about 'who can transact with whom and the conditions under which transactions are carried out' (Fligstein 1996: 658). For example, exclusivity (sole-source) is a rule of exchange governing buyerseller relationships in the advertising industry (Baker, Faulkner, and Fisher 1998: 151). Competitive rules of exchange prohibit union triads. The avoidance of common suppliers is a good example (Baker 1990). Disunion strategies are legitimate and successful in firms operated as 'markets' (Eccles and White 1988). In settings such as markets operated as 'firms' (Stinchcombe 1985), however, cooperative rules of exchange discourage disunion strategies (which would be interpreted as self-serving, opportunistic behaviors). Similarly, organizations in small-firm networks share information, establish long-term relationships, and support each other's efforts, in opposition to the competitive rules of exchange in classic markets (Perrow 1992). The search for greater efficiencies through closer ties between customers and suppliers can displace traditional rules of exchange that pit suppliers against one another. New developments in just-in-time (JIT) inventory control, for example, call for close coordination between customers and their


suppliers. Honeywell orchestrates close cooperation between its internal buyers and five suppliers who would ordinarily compete for business (Bleakley 1995).

Norms are shared expectations that regulate behavior (DiMaggio 1994), including choice of entrepreneurial strategy. Norms of openness, teamwork, trust, and reciprocity favor union over disunion strategies. For example, Putnam (1993a) argues that norms of reciprocity in the industrial districts of northern Italy foster interfirm cooperation and limit opportunistic behavior-that is, the use of disunion strategies. Similarly, norms of reciprocity support the union strategies evident in such business groups as the Japanese kieretsu and Korean chaebol.

Social orientation refers to the distinction between individualism and collectivism (Triandis 1995: 2). Individualism, for example, is 'a social pattern that consists of loosely linked individuals who view themselves as independent of collectives; are primarily motivated by their own preferences, needs, rights, and the contracts they have established with others; and emphasize rational analyses of the advantages and disadvantages of associating with others' (Triandis 1995: 2). Individualism favors disunion strategies; collectivism favors union strategies. For example, collectivist practices such as cross-functional teams, multi-level management networks, group-level reward systems, and team-building programs, facilitate union strategies within organizations (Baker 1994a). Recent research has shown that a collectivist orientation often increases the odds of alliance formation (Dickson and Weaver 1997). Private economic associations and political organizations foster cooperation in the industrial districts of northern Italy (1993a). The robust collaboration found in these industrial districts is supported by technical colleges, vocational training, supportive banks, and extended kinship ties (Powell 1990).

Distribution of Strategies We argue that the frequency, legitimacy, and success of an entrepreneurial strategy depends on its 'fit' with the 'design' of the institutional context in which social entrepreneurs operate. This suggests the following proposition: The ratio of disunion to union strategies varies according to the structure and culture of the institutional context. At one extreme, disunion strategies dominate in settings characterized by sparse, disconnected, and differentiated networks, along with competitive rules of exchange, opportunism, and an individualist orientation; at the other extreme, union strategies dominate in settings characterized by dense, connected, and undifferentiated networks, coupled with cooperative rules of exchange, reciprocity, and a collectivist orientation. A mix of strategies occurs in an institutional context located between these two extremes.

The distribution of strategies in a given context can be determined by a triads census (Wasserman and Faust 1994: 556-602). The pattern of strategic alliances in the global automobile industry9 illustrates one empirical distribution (Baker 1992b), though the triads approach can be applied in any institutional context. For strategic alliances, a triads census includes only four possible triad isomorphism classes: 1) the null triad, composed of three unallied firms; 2) the dyad, composed of three firms of which only two are connected by an alliance; 3) the disunion triad,

Tab

le 2

. Tri

ads

cens

us o

f alli

ance

s in

the

wor

ld a

utom

obile

indu

stry

Nul

l Tri

ad

Dya

d [0

03]"

[1

02]

Typ

e o

f Alli

ance

ob

serv

ed

expe

cted

ob

serv

ed

expe

cted

All

Alli

ance

s 56

16

5510

18

14

2010

jo

int v

entu

re

7045

70

24

681

722

man

ufac

turi

ng/a

ssem

bly

6631

66

07

1057

11

03

tech

nolo

gy s

hari

ng

6877

68

61

844

873

supp

lier r

elat

ions

hip

6529

65

13

1158

11

86

mar

ketin

g/di

stri

butio

n 70

30

7024

71

2 72

2 eq

uity

inv

estm

ent

7257

72

56

502

503

Dis

unio

n T

riad

[2

01]

obse

rved

ex

pect

ed

314

241

43

24

78

60

46

36

77

70

26

24

10

11

g. ~

Uni

on T

riad

[3

00]

obse

rved

ex

pect

ed

26

9 I

0 4

1 3

0 6

1 2

10

I 0

Sign

ific

ance

b

'tD

5.97

2 4.

463

2.76

7 1.

981

.976

.5

55

-.27

8

aNum

bers

in s

quar

e br

acke

ts [

MA

N]

refe

r to

tria

d is

omor

phis

m c

lass

es u

sing

sta

ndar

d la

belin

g, w

here

M =

num

ber o

f mut

ual d

yads

, A =

num

ber o

f as

ymm

etri

c dy

ads,

and

N =

num

ber o

f nul

l dya

ds.

b tau

test

sta

tistic

from

TR

IAD

S (W

alke

r and

Was

serm

an 1

987)

: 't

D =

di

suni

on tr

iads

; 'tu

=

unio

n tr

iads

. So

urce

: B

aker

(199

2b).

Dat

a fr

om W

ards

Int

erna

tion

al,

1985

(N

= 3

7 au

tom

obile

com

pani

es).

'tu

5.54

0

1.53

8 2.

939

3.75

2 4.

079

3.57

3 3.

466

CIl o n E

(')

.§ [ r:T

'< ~

til

~.

::s -o


composed of two alliance dyads; and, 4) the union triad, composed of three alliance dyads. 10 The first class, the null triad, represents a classic competitive situation-the complete absence of strategic alliances among three firms. The second class, the dyad, represents an isolated alliance between two firms. The third class, the disunion triad, represents the tertius gaudens arrangement (as illustrated in Figure 1). The fourth class, the union triad, represents the social cohesiveness structure of social capital (see Figure 1).

The empirical distribution obtained in the triads census of the automobile industry is shown in Table 2. For our purposes, we focus here on the two triad types of particular interest, disunion and union triads. Both triad types occur much more often than expected by chance alone, considering all alliance types combined. 11

However, if we examine the triads census for each type of alliance, we find an interesting pattern: 1) Disunion triads (but not union triads) occur more often than by chance in joint ventures; 2) Disunion and union triads occur more often than by chance in two types of alliances-technology sharing and manufacturing/assembly; 3) Union triads (but not disunion triads) occur more often than by chance in three types of alliance-supplier ties, marketing/distribution, and equity investments.

This pattern suggests that the structure of social capital varies by alliance type. The structure of joint ventures implies that social capital is accessed and mobilized by exploiting structural holes, using the tertius gaudens strategy. The structure of supplier ties, marketing/distribution, and equity investments implies that social capital is accessed and mobilized by closing structural holes. The mixed structure of technology sharing and manufacturing/assembly suggests the presence of both types of social capital. However, the statistically significant use of both disunion and union strategies may indicate a competition of strategies in which neither dominates.

The general relationship between entrepreneurial strategies and institutional context is illustrated in Figure 2. This stylized representation implies the possibility of change or movement, of the transformation of an institutional context and its corresponding strategies. The intersection of the curves in Figure 2 represents a balance of disunion and union strategies. This point is an unstable state in which neither strategy dominates, such as the mixed structure of triadic strategies for technology sharing and manufacturing/assembly in the automobile industry (Table 2). If so, then the automobile industry may fall back on earlier, simpler, and overlearned strategies (arms' -length competition), as people and organizations are prone when faced with uncertainty and ambiguity (Weick 1995: 102).

For an organization, the point of intersection in Figure 2 indicates a particularly risky stage in an organization's transition from one institutional design to another. For example, a firm attempting to improve collaboration and cooperation must foster both the structural and cultural conditions that favor the union strategy (moving from left to right in Figure 2). Some consultants claim that a hierarchy can be converted into a network organization simply by adding links (Lipnack and Stamps 1994: 72; see, also, Mueller 1986), but structural change is not enough. Failure to change both structural and cultural conditions endangers a change effort. For example, the effort to transform Industrial Computer and Control Group failed because change agents altered only organizational structure (Nohria and Berkley

Entrepreneurial

Strategy

Disunion

SO/50

Union

-----

Small, dense, integrated networks; cooperation, trust, collectivism

" " / /

/ I

Institutional Context


.-.-"

-~----

----

-- Disunion

- - - - Union

Large, sparse, disconnected networks; competition, opportunism, individualism

Figure 2. Entrepreneurial strategies by institutional context

1995). By replacing hierarchy with a network design, they may have induced more union strategies, but by itself this structural change was not enough. Without a corresponding cultural change, the effort to foster cross-divisional collaboration was doomed. Employees did not fundamentally change their strategies for action (Nohria and Berkley 1995). The change effort may have collapsed at or near the point of intersection in Figure 2.

CONCLUSION

Our chapter attempts to specify and clarify dimensions and structures of social capital. We offer a set of concepts that encompasses different views of the structural sources of social capital, the basic strategies used to access social capital, and the relationship between institutional contexts and strategies. We argue that social capital 'inheres in the structure of relations between and among actors' (Coleman 1988: 98) in two fundamental ways-one based on the patterned absence of ties, the other on the patterned presence of ties. Structural holes theory (Burt 1992) emphasizes the absence of ties, where social capital resides in a social structure characterized by sparse networks and few redundancies. We maintain that social capital also inheres in social structure as the presence of ties, which we call social


cohesiveness, where social capital is found In dense networks with multiple redundancies.

Each structure of social capital represents opportunities to access and mobilize the resources inherent in a social network. The first structure calls for a 'disunion' strategy that exploits the structural holes between alters by keeping them apart. The second structure calls for a 'union' strategy that creates value by bringing alters together, closing the hole between them. Our qualitative and quantitative examples provide ample evidence of the use of these strategies in a wide range of institutional contexts, both inside and between organizations. The concepts of two types of social capital (structural holes versus social cohesiveness) and two types of entrepreneurial strategies (disunion and union) can be generalized and applied to understand the structure and use of social capital in many different institutional settings.

The actual distribution of strategies in a particular institutional context depends on the 'design' of the institutional context in which social entrepreneurs operate. An entrepreneurial strategy that 'fits' the structure and culture of a given institutional context occurs more frequently, enjoys greater legitimacy, and will be more successful in the long run, compared to a strategy that does not fit. Disunion strategies dominate in organizations and markets characterized by sparse, disconnected, and differentiated networks, coupled with competitive rules of exchange, opportunism, and an individualist orientation; union strategies dominate in organizations and markets characterized by dense, connected, and undifferentiated networks, coupled with cooperative rules of exchange, norms of reciprocity, and a collectivist orientation.

Of course, a mix of triadic strategies falls between the pure disunion and union extremes. Thus, one avenue of additional research is to explore further the precise connection between institutional context and entrepreneurial strategy. This line of work calls for a demography of social entrepreneurial relationships (see, also, Baker, Faulkner, and Fisher 1998: 173). The statistical methods available for taking a triads census (Walker and Wasserman 1987), which we used in our analysis of the distribution of strategies in the global automobile industry (Baker 1992b), can be used to determine the distribution of strategies in any organization, market, or organizational field. In addition, organizational change can be tracked by measuring shifts in the distribution of strategies over time, using methods designed to model longitudinal change in networks (e.g., Leenders 1995a, 1996; Wasserman and Faust 1994). By coupling new and better measurement methods (e.g., Borgatti 1997; Han and Breiger, this volume; Doreian, this volume) with more precise concepts of social capital and their corresponding entrepreneurial strategies, it is possible to take another step forward in the exploration of corporate social capital.

We are grateful to the editors of this volume for their helpful comments and suggestions. Direct correspondence concerning this chapter to Wayne Baker, University of Michigan Business School, 701 Tappan Street, Ann Aibor, MI 48109 ([email protected]).

NOTES

I. The debate takes place in published works (see citations herein) as well as in infonnal discussions on the Internet, such as the recent interchanges on social capital in SOCNET between January and June, 1997. We cite the published literature to support our points, but we also acknowledge the contributions to the debate in SOCNET made by Xavier De Souza Briggs, Robert Putnam, Barry Wellman, and others.


2. This point was also made in SOCNET by Barry Wellman and Robert Putnam (January 1997). 3. Simme1 (1950) describes another triad type, 'divide and conquer,' in addition to the tertius gaudens and 'non-partisan' types. In 'divide and conquer,' the third party deliberately introduces conflict between the other two. This type is not our concern in this chapter. 4. Real estate brokerage and literary agency qualify as this type because the ego is not a principal in a transaction, but instead brings together the two principals (alters) who consummate the transaction. In the first case, the real estate broker (ego) matches a buyer and seller (alters) who exchange property for money; in the second, a literary agent (ego) matches a publisher and author (alters) who exchange 'the property' (manuscript) for royalties. Some types of brokerage do not qualify as this type. For example, in a wholesaler arrangement, the broker is a principal in one transaction (buying goods from a manufacturer) and in a second transaction as well (selling the goods to the consumer). 5. Consider, for example, the protections contained in the standard 'listing agreement' (officially, the 'cooperative selling contract') used in the state of D1inois to define the legal obligations of the seller to the real estate broker: 'SELLER SHALL: Cooperate fully with Broker; refer all inquiries to Broker; conduct all negotiations through Broker; ... and pay a real estate brokerage commission of X % of the sale price; if 1) Broker provides a purchaser ready, willing, and able to purchase in accordance with this Contract; or 2) if the property is sold, exchange, gifted, or optioned by Broker or by or through any other person including the Seller during the period of this contract; or 3) if it is sold directly or indirectly within six (6) months after termination of this contract to a purchaser to whom it was offered during the tenn thereof.' 6. It is possible, of course, that the individual pursuit of entrepreneurial profit as a private good produces value at the collective level. This line of reasoning is consistent with traditional management and economic theories. For example, the tournament model of mobility assumes that competition among managers yields better ideas, more satisfied customers, and greater shareholder value. The theory of the market similarly assumes that individual striving maximizes social welfare. Burt (1992) does not emphasize the public-goods aspect of social entrepreneurship, though he mentions the possibility: An entrepreneur is 'a person who generates profit from being between others. A nonprofit player, pursuing entrepreneurial opportunities just for the pleasure of being the one who brings others together to build value, could choose to reinvest it all' and strengthen existing relationships (Burt 1992: 34-35). Of course, social capital can be considered a public good in and of itself. Putnam argues, for example, that Coleman's (1988) original concept of social capital incorporates such a view. He adds, however, that the views of social capital as a private and public good are complementary (see discussions in SOCNET; see, also, Putnam 1993a, 1995; Knoke this volume). 7. Nahapiet and Ghoshal (1998: 261) briefly mention that some interorganizational networks may develop an institutional context that is conducive to high levels of social capital, and point to this as an avenue of future research. 8. We acknowledge, but do not elaborate here, the importance of the 'constitutive' as opposed to the 'regulatory' view of culture (DiMaggio 1994). In addition, we acknowledge but do not discuss the state as part of the institutional context (Fligstein 1996). 9. We obtained data for the triads analysis from a special issue of Ward's Automotive International (1986). This issue reported the alliances known to exist among 37 major automobile companies, classified by the six alliance types reported in Table 2. We created a square, binary, symmetric matrix for each alliance type, where each company is assigned a row and corresponding column, and an entry indicates the presence or absence (0,1) of an alliance between two companies. Thus, we created six 37 X 37 matrices. We analyzed the triads structure of each matrix using TRIADS (Wassennan and Walker, 1987). These data are treated as symmetric because alliances are naturally mutual ties, and we adjusted the weighting vector in TRIADS accordingly. 10. A triads census includes sixteen isomorphism classes when asymmetric ties are considered (Wassennan and Faust 1994: 566). Since alliances are mutual ties, however, there are only four classes of interest: triad types 003 (null), 102 (dyad), 201 (disunion), and 300 (union ). These three-digit numbers [MAN) refer to triad isomorphism classes using standard labeling, where M = number of mutual dyads, A = number of asymmetric dyads, and N = number of null dyads. II. Structural hypotheses are tested using the TRIADS program (Walker and Wassennan 1987) in which the empirical distribution of triads is compared against a theoretical distribution of triads assuming random assignment of ties. The weighting vector used in these tests reflects the number of disunion configurations contained in the disunion triad, and the number of union configurations contained in the union triad. The test statistic, tau, is interpreted using the standard nonnal distribution (Wassennan and Faust 1994: 595). We use a significance level of .05 in this analysis.

•

Corporate Social Capital and Liability: a Conditional Approach to Three Consequences of Corporate Social Structure

ABSTRACT

5 Dan TaImud

This chapter defines and illuminates three aspects of corporate social capital which are created by different aspects of corporate social structure. The chapter also shows that corporate structure which generates social capital may become later a liability. The chapter briefly reviews the literature regarding the contingent value of corporate structure in creating competitive social capital. Then two other kinds of corporate social capital, political and cognitive, are illustrated via a case study of an Israeli Armament firm. Finally, the chapter briefly discusses the relations between the three kinds of corporate social capital for business policy and strategy as well as for future studies on corporate advantage.

INTRODUCTION: VARIETIES OF CORPORATE SOCIAL CAPITAL

Certain corporate social structures provide comparative advantage to the firm. The term 'corporate social capital' is a heuristic device, assists us in comprehending how corporate social structure benefits certain business organizations at the expense of others. Corporate social capital is any positive, goal-specific outcome, originating from corporate social structure. 1 It includes any means of corporate control or business leverage, embedded in social relations, thus assisting a firm in promoting internal assets and resources.2 Excess profit or economic rent, which is an outcome of the operation of corporate social capital, is depicted as 'any advantage or surplus created by nature or social structure over a certain period of time' (Sorenson 1996: 1344). Corporate advantage may be explained then by the differential capacity of firms to create, promote and take advantage of corporate social capital (Nahapiet and GhoshaI1998).3

Corporate Social Capital and Liability- 107

I argue that one can differentiate analytically between three kinds of corporate social capital, each derived from a particular dimension of corporate social structure (Zukin and DiMaggio 1990) [see Table I]: I) Competitive corporate capital is as a function of a firm's structural location in imperfect competition (Burt 1992). The term expresses the extent to which the social organization of the competition, as measured via transaction networks, systematically benefits certain market players at the expense of others (Burt 1992; Talmud 1992, 1994; Talmud and Mesch 1997). For example, a particular form of competitive corporate capital is the manner by which suppliers or buyers are dependent on a player in an exchange system.4

Another example is the ability of a firm to engage in exclusive relations with subcontractors, thus increasing its capacity for complex adaptation and economy of speed (Uzzi 1997a).

By contrast, 2) political corporate capital indicates the manner in which the political organization of the economy, institutional and legal regulation of corporate behavior, and the collective struggles over economic options extend the strategic possibilities of corporate executives. By virtue of membership in larger political networks, certain firms have more institutional leverage and can buy political protection from state and local institutions. For example, firms with direct institutional ties to the political center may gain more collective resources than such lacking those ties (Talmud 1992; Talmud and Mesch 1997).

In their strategic considerations, firms attempt to influence and take advantage of legal arrangements, political ties, and state policies at various levels (Zukin and DiMaggio 1990: 20-21; Talmud 1992; Talmud and Mesch 1997). Political arrangements have uneven effects on market players and therefore may be defined as political corporate capital. This is measured by strong linkage to the institutionalized political center, as well as by institutional regulation and political arrangements favoring the focal firm or industry. However, political arrangement are dynamic and high reliance on political capital in one period may be detrimental in another time. Continuing benefits from political corporate capital compel corporate players to calculate and rearrange their corporate capital for sudden shifts in the political sphere or to politically-directed changes in the corporate task environment.5

Paradoxically, those corporations rich in political corporate capital which have learned to rely on institutionalized political arrangements are among those firms, highly prone to crisis and mismanagement following transformation of political agencies. These firms often enjoy rewards such as lower tax rates, tariff protection, interest-free loans, business tips, favorable environmental conditions, and direct and indirect state subsidies (Talmud 1992). In economic terms, firms that concentrated their efforts on using political corporate capital are engaged in rent-seeking conduct, or in 'directly unproductive, profit-seeking behavior' (Bhagwati 1982). Therefore, their strategic repertoire is not geared toward efficiency but toward an effective usage of their political leverage.

Finally, only recently the organizational literature has characterized a less known form of corporate social capital: 3) cognitive corporate capital.6 While competitive and political corporate capital are instrumental for the external economy of the firm, cognitive corporate capital is chiefly relevant for the internal economy of business organization. It is defined as the extent to which a firm can facilitate


advantageous business action by using internal networks which reconstruct shared understanding, disciplinary power, mental models, organizational identity, tacit knowledge, corporate norms, the ability for corporate foresight, reflexivity, and selfawareness.7 Cognitive corporate capital then is facilitated by relations, formal and informal, inside the firm (Podolny and Baron 1997; Nahapiet and Ghoshal 1998). It assists the organization in accomplishing goals, to speed production process, to improve quality according to clients' requirements, to 'move things ahead' and to fortify its competitiveness, effectiveness, or efficiency (Uzzi 1996a).

The importance of cognitive corporate capital cannot be overstated. Several perspectives have currently acknowledged this unique form of corporate social capital. In sociology and psychology, organizational self-command has been perceived as an integral part of 'relational management' and 'knowledge management' of the firm as a social community.8 Resource-based business strategy has shifted organizational theory from an emphasis on value appropriation to the exploitation of the firm's social structure for value creation. Similarly, Human Resources Management have recently put more weight on employees' long term relational contracts and 'employability' (Ghoshal and Moran 1996).

These cognitive assets can be constructed and reproduced because any corporate firm is first and foremost a social community, a nexus of social relations (Nahapiet and Ghoshal 1998). Intra-organizational social networks thus yield and convey organizational memory and shared narratives and norms. In fact, social networks can create or modify trust in the organizational system (Kramer, Brewer, and Hanna 1996). Social networks also convey intangible corporate assets such as sociallyproduced tacit knowledge, which is particularly useful when and product complexity and quality are at stake (Polanyi 1958; Dore 1983; Jones, Hesterley, and Borgatti 1997).9 Moreover, cognitive corporate capital, like any other social capital discussed in this chapter, is a firm-level phenomenon. It is endowed in corporate social relations, and not in individual players, and thus the firm can take advantage of these nested relations (Monge, Cozzens, and Contractor 1991). Cognitive corporate capital then may speed a firm's reaction to market cycles, enhancing complex adaptation and reducing learning cost and time-to-market (Uzzi 1997a). It may, furthermore, promotes collaborative production of social knowledge capacity, which is especially relevant for the construction and transfer of socially tacit knowledge (Nahapiet and Ghoshal 1998). It may also increase learning speed and effectiveness, and promote smooth solutions to problem solving (Dore 1983; Uzzi 1996a, 1997a; Nahapiet and Ghoshal 1998). Still, cognitive corporate capital is a firm-level virtual asset which decreases the economic cost of monitoring, learning, adapting, and also diminishes agency problems. Like any other form of corporate social capital, it has diminishing marginal utility, and thus could be accumulated up to a certain point (Uzzi 1997a; Talmud and Mesch 1997). If external conditions significantly change, the very same corporate social structure producing cognitive corporate capital in one circumstance, may in another instance create a corporate liability, thus inhibiting the firm's adjustment to new environmental circumstances.

Because cognitive corporate capital is an intangible corporate, relation-specific asset, it cannot easily be imitated by competitors (Barney 1991). It is transmitted via


Table 1. Three types of corporate social capital

Competitive

Social Structure Producing Structural Embeddedness: Corporate Network Position

Corporate Social Capital

Examples Non-redundant Market Ties (Burt 1992)

- Rewards Profit squeezing capacity (Burt 1992; Talmud 1994); sustained

industry leadership (Talmud and Mesch 1997)

Political

Social Structure Producing Political Economy and Institutional Regulation

Corporate Social Capital

Examples

- Rewards

Cognitive

State regulation; political institutional ties (Talmud and Mesch 1997)

State subsidies (Talmud 1992b)

Social Structure Producing Normative and Cognitive Embeddedness: Shared narratives,

Corporate Social Capital normative role expectations and commitment, corporate identity, vision

and discourse enacted via intra-organizational social networks.

- Examples 'Self-command capital' (Lindenberg 1993); 'Strong culture' (Kotter and

Heskett 1992); tacit knowledge (Polanyi 1958); 'combinational

capability' (Nahapiet and Ghoshal 1998)

- Rewards Corporate innovation (Monge, Cozzens, and Contractor 1991);

complex adaptation, 'economy of speed' (Uzzi 19%a; 1997a);

provision of product quality (Dore 1983)

a complicated web of social relations. The creation, configuration and maintenance of cognitive corporate capital is costly, consuming time and corporate resources.

An organization needs to maintain a careful balance, therefore, between various modes of corporate social capital, but in a similar vein to those business organizations relying upon political corporate capital, firms which have advantageous cognitive corporate capital tend to remain stationary. Corporate networks conveying identities and normative expectations are inclined to become inflexible over time (White 1992). Accordingly, ideal cognitive capital should not prevent a firm from being flexible.

All these three forms of corporate social capital have a few common attributes:

• They are not a property of any individual agent; • They can be accumulated over time; • They are relation-specific and cannot easily be transformed from network to

network; • Therefore they are expensive, bearing investment-specific sunk cost; • If they are not properly maintained, they tend to fade out or become

irrelevant to the extent that a firm is located in a turbulent and uncertain environment, and to the degree that the corporate social structure producing corporate social capital is rigid, they tend over time to turn from a corporate capital into a corporate liability.


There are also some significant differences between the three forms:

a. Competitive corporate structure creating corporate social capital is the most measurable as precisely operationalized by network models of competitive advantage (cf. Burt 1992).

b. Political corporate capital is more vague, composed of many sub-forms, and is molded corporate actors' ties as well as by historical and institutional circumstances.

c. Cognitive corporate capital is the most 'sticky' and vague form. It is a product of both past and present relational composition inside the firm, and of interface with other agents, such as competitors, consultants, suppliers and clients.

The following part has three sections, each of which addresses a particular dimension of the corporate social capital. The first section summarizes the literature on the contingent sources of competitive corporate capital. The next section demonstrates how social structure producing political corporate capital becomes a corporate liability. Finally, the chapter discusses to inability of a manufacturing firm to renovate its cognitive corporate capital.

A CONDITIONAL APPROACH TO CORPORATE SOCIAL STRUCTURE AND CORPORATE SOCIAL CAPITAL

Recent literature on corporate social capital and competitive advantage has stressed the role of weak ties and sparse social networks in determining corporate performance (Burt 1992).\0 Burt (1992) and Talmud (1992, 1994) found that firms using 'structural holes,' spreading their ties with unconnected market segments are more profitable than those connecting to 'redundant market areas.'

In contrast, other studies on corporate social capital have found significant impact of strong ties and dense networks in creating opportunities. New and sensitive business information and opportunities are enhanced through cohesive contacts (Aldrich and Zimmer 1986; Gilad, Kaish, and Ronen 1989). Moreover, Krackhardt (1992), Podolny (1994) and Gabbay (1995, 1997) showed the importance of enclosed networks for managing corporate uncertainty.

It seems, accordingly, that there is no such thing as a universal optimal network structure (dense or sparse); as Coleman phrased it 'social relationships that constitute social capital for one kind of productive activity may be impediments for another' (Coleman 1994: 177).

Furthermore, the effect of strong ties seems to be a conditional one. Burt (1992) and Han (1993) revealed that, contrary to the logic behind the theories of resource dependence and structural holes, those American firms having high recourse dependence on the government survive at higher rates. Also, Israeli firms that are 'locked in' by political ownership are more profitable (Talmud 1992), and are more stable over time (Talmud and Mesch 1997). In uncertain conditions, or where identity and role expectations are important for performance, embedding economic transactions in strong relations, or constructing the economic deal in terms of strong ties, could serve as remedies for survival (Dore 1983; Podolny 1994; Jones, Hesterly, and Borgatti 1997; Podolny and Castellucci, this volume).


Organizations develop indirect corporate leverage via social networks to manage their dependence, while protecting their advantage in strategic positions (Galaskiewicz 1985; Baker 1990). Juggling between contrasting business pressures, corporate firms attempt to maintain their access to reliable and non- standard information, while also striving to reduce trade dependency and uncertainty. Thus, corporations are bound into repeated transactions, while also preserving other opportunities at hand (Powell 1990; Podolny 1994). Uzzi (I996a, 1997a) found in his study of the New York apparel industry that organizational failure increases to the degree that the focal firm's network tends to be composed of either all arm'slength ties (i.e. 'under-embedded,' disconnected, non-redundant contractors), or all embedded ties (i.e. 'over-embedded,' dense, semi-integrated contractors). Uzzi (I997a) showed that organizational survival is positively associated with a 'mixed model' network, which provides a combined advantage. On the one hand, market embeddedness provide benefits such as trust, joint problem-solving arrangements, complex adaptation, reduced bargaining and monitoring costs. On the other hand, arm's-length contacts provide the firms new and novel information outside the immediate ties (Uzzi 1996a, 1997a). Similarly, Talmud and Mesch (1997) have demonstrated that the probability of the leading Israeli firms to sustained their industrial leadership is a function of the 'mix embedded ness' type of relations. Similarly, Uzzi (1996a, 1997a) claims that each type of controlled embedded ness 'yields positive returns up to a certain point,' and 'embedded networks offer a competitive form of organizing, but possess their own pitfalls.' Competitive pressures create complex, dissimilar, and distinct network forms, rather than a singly optimal embeddedness strategy (Uzzi 1997a). As a result, 'a paradox appears: optimal networks are not composed of either all embedded ties or arm's-length ties but integrate the two' (Uzzi 1996a: 694). In sum, the conditional approach to competitive corporate social structure argues that the corporate value behind network forms is contingent upon contextual features of the environment. II Next, I will apply this insight to political and cognitive corporate capital.

THE CASE OF TA'AS: ISRAELI ARMEMENT CORPORATIONI2

Political Corporate Capital: The Demise of Corporate Leverage This section illuminates how political corporate capital diminishes due to a combination of organizational rigidity and changing environmental conditions.

In many cases, competitive corporate capital is not sufficient to understand the construction and maintenance of corporate social capital. In most countries, corporate political embedded ness of the market, institutional regulation, and governmental restrictions on competition augment firms' market positions and elevating barriers to entry (Levacic 1987: 118-138; Talmud 1992). In other words, political corporate capital assists a firm in its capacity for 'price making.'

Ta'as (The Hebrew acronym for the Israeli Military Industry) was incorporated in 1947 as an institutionalized organ of armament. It operated first in the Jewish underground against British rule and later on for the Israeli Defense Forces, which had been its chief client over the years. Ta'as became then a state agency which gradually transformed into a state-owned enterprise with very low autonomy and without financial reserves or structural leverage. State regulation and laws prevented


private investors to compete with Ta'as in the local market, but also to from loaning finances to the company. Ta'as' business policy and strategy was effected much more by political considerations than by sheer economic issues (see also Samuel 1998). This situation has facilitated a monopoly position for many Israeli Defense Industry firms. Yet, the emergence of the newly globalized weapons market, the end of the cold war, and the peace process in the Middle East brought about a radical change in the political embeddedness of the weapon's industry, as demonstrated in the case of Ta'as. If political corporate capital can be measured by a premium paid for (or excessive revenue earned by) a given investment in political ties, thenbecause it is a relation-specific investment-dearly, any drastic modification of the political structure can alter the fate of the firm, from a revenue to a loss.

Until the 1980s, the firm could use the dramatically leaking bucket of the State of Israel's defense budget. Accordingly, its business strategy was typified by the overarching principle of autarky and freedom from functional constraints. Structurally, it had one main buyer (monopsony), but Ta'as was that buyer's main supplier as well (monopoly). These structure of bi-Iateral monopoly had been maintained and fortified by the firm's identification with the heroic Jewish underground under the British Mandate in Palestine, to be superseded later on by its strong cultural affiliation with the dominant security symbols of the State of Israel. Already heavily invested in political corporate capital, the firm rationally preferred a strategy emphasizing effectiveness over efficiency. As a result, the organization forged an ambitious business strategy: it tended to 'make' rather than cheaply 'buy' various components of its product; it stressed, moreover, activities of research and development function and employed a disproportionately large, permanent and expensive workforce (increasing from 1000 employees in 1948 to 14,500 in 1985; Talmud and Yanovitzky 1998). Maintaining its political corporate capital. the organization possessed an excessive production capacity. accompanied by useless vertical integration along many of product lines including logistics, maintenance and support systems (Kleimann 1992; Talmud and Yanovitzky 1998) to overcome prospective shortage or demand fluctuation. Additionally, the firm diversified into about 500 different products in various market areas, at the expense of achieving competitive advantage by focusing on core advantage. Ta'as chose, in fact, to focusing on technological push (imposing scientific innovation and engineering excellence) rather than by market pull (imposing product's fit to customers' preferences) (Samuel 1998; Talmud and Yanovitsky 1998). Additionally, competition in the global market was considered to be only secondary to the focus on the local Israeli market. Many governmental restrictions on military equipment export were compensated by state agencies. Still, the industry was booming until the 1980s, as a result of its political corporate capital. In the 'heroic' period of 1947-1985, Ta'as was socially conceived as strategically important, and thus enjoyed close institutional linkage with the Ministry of Defense, which was formally its owner and main buyer (Talmud and Yanovitzky 1998). The fact that Ta'as lacked competitive corporate capital (its transaction network was concentrated on a single buyer for any single product) was compensated for by the higher value of its political corporate capital. which was deemed instrumental for corporate solvency.


Following 1985, there have been increasingly severe cuts in the Defense budget. The State of Israel thus has become more sensitive to armament prices. In addition, global prices have decreased, as suppliers' competition in the world market has intensified. The sharp modification of Ta'as' political corporate capital created a series of acute structural crises, accompanied by a severe financial toll, layoffs, rapid managerial turnover, and cultural shocks to employees. Moreover, it exposed the structural weakness of the firm's transaction networks and its high dependency on suppliers and customers (Talmud and Yanovitzky 1998). Ta'as' political corporate capital has been eroded because it was based merely on a social ties with political agents which were no longer relevant.

Additionally, it is important to note that political and competitive corporate capital may also be linked. The case of Ta'as shows that the political structure affects the industry competitive structure, and thus provide regulated competitive leverage for the firm. Moreover, state regulation and laws prevented private investors to compete with Ta'as in the local market, but at the same time enforced the company to lean on political backup. Barriers-to-entry and barriers to exit were fixed by the political organization of the industry. The competitive structure was therefore strongly affected by the political web which Ta'as was a part of. Upon state lessening its grasp on Ta'as later on, other kinds of structures could have been nurtured and capital could be drawn from them. Indeed, this was a prerequisite for survival and solvency. The state ceased to be highly dependent on the firm, as the global market emerged, and the sense of closure of politicians, public officials and CEOs has been also damaged because of the eroding symbolic importance of the security sector (Talmud and Yanovitzky 1998). Nonetheless, the firm was unable to embark on developing other forms of compensating corporate capital (neither competitive nor political), partially because of the transformation of its cognitive corporate capital.

Cognitive Corporate Capital: From Corporate Superiority to Corporate Liability In its formative period, Ta'as enjoyed a relatively safe, prestigious and certain environment with a comparatively assured clientele (Evron 1992). The organization was composed of relatively stable intra-organizational networks, producing consensus regarding the corporate mission and working procedures. The firm's managers were production-oriented, lacking market orientation and marketing skills. \3 The firm, leveraged by its political corporate capital, emphasized production capacity and product quality rather than marketability. The formal and informal networks transmitted this shared language throughout the organization, and the kind of novel knowledge developed inside the firm in the 'heroic period' (especially regarding ammunition), reflected this socially accepted value. The fact that the corporate vision was to assist the Israeli Defense Force regardless of cost brought about a shared sense of commitment to a special 'community-organization.'

Ta'as consisted of four basic structural elements deemed necessary for the creation of its cognitive corporate capital: 1) shared symbols and narratives, 2) a sense of closure, being a part of the 'security sector,' collective history, 3) repeated interactions between experts, and 4) interdependence between customers and experts


(Nahapiet and Ghoshal 1998; Talmud and Yanovitzky 1998). Consequently, many risky innovations, devised inside the firm by engineers and technicians-some of them lacking formal training-were made possible by mitigating socially-connected trust within the organization using technical experts' commitment and tacit knowhow (see also Kramer, Brewer, and Hanna 1996).

By contrast, in the post-1985 'competitive era,' Ta'as has attempted remedial measures. Nevertheless, these measures have been ineffective. As I will show, Ta'as social structure~eemed necessary for creating its older cognitive corporate capital-has become inappropriate for the new environment, thus resulting in corporate liability.

Following the radical changes in environmental conditions, the firm undertook a partially successful restructuring program including decentralization, the establishment of relatively autonomous profit centers, implanting export orientation, introduction of (failed) products' modification for the civil markets, layoffs, and cutting administrative costs.

Moreover, the new organizational vision, imposed by new Chairs of the Board, CEOs and General Managers, which emphasized market competition, stood in sharp contrast to the old one, emphasizing priceless product quality, conveyed via existing intra-organizational networks. The management did not trouble to pass on its new vision via social networks since its image of the firm was formal and hierarchical. Yet relations inside the rank and file were persistent, thus assuring a highly shared resistance to change. In other words, the newly introduced uncertainty hampered the 'psychological contract' and the trustworthy social relations inside the firm,14 and was not accompanied by management's proper awareness of informal processes inside the corporate units. Furthermore, uncertainty and a lack of trust were magnified by contradictory messages relayed from the government to the firm, and by high turnover rates of staff and line managers, including the CEO's. This, in turn, diminished social cohesion inside the organization, and restricted the ability of the new managers to set up effective ties with unit managers and other influential players. Consequently, this impaired Ta'as' capacity to implant new-fashioned corporate identity, fresh strategic vision, and more important, trust in the corporate headquarters (Talmud and Yanovitzky 1998).

The literature on corporate social capital underscores the fact that socially embedded exchange may prevent agency problems and opportunistic behavior which resulting in sub-optimal performance (cf. Podolny 1994; Jones, Hesterly, and Borgatti 1997). Yet the lack of social linkage and trust between the new CEO's and the unit managers created a dual agency problem, originated by relational discontinuity inside the firm. This discontinuity, in turn, was created by a lack of durability in identity-making ties (Podolny and Baron 1997), which is required especially in times of crisis (Mishira 1996; Webb 1996). It was also found that both the owner (the State) and the corporate directors were not receiving reliable information on the business and financial operation of Ta'as from their respective subordinates. As a result, they could not make relevant decisions and could not install monitoring devices regarding the fate of various profit centers within the firm (Talmud and Yanovitzky 1998). The corporate ship, consequently, began to drown.


In a highly diversified corporation mitigating a dual-level crisis (at the industry level and at the firm level), suffering from a lack of internal linkage and trust, bureaucratic formal control is not as effective a device as that of corporate social control transmitted via social networks. The new management ignored the veterans' social networks. Layoffs did not solve the problem, as recruitment was virtually non-existence. Thus the old, durable, identity-formation networks inside the firm have created, in fact, corporate paralysis (Crozier 1960). In Ta'as, the 'old guards' of the veteran employees and unit managers gave credence to norms such as 'effectiveness,' 'quality,' 'security,' while the new corporate managers were more attuned to catchwords such as 'efficiency,' 'market testing,' and 'cost-benefit analysis.' Hence, a lack of social linkage accompanied by a deficit in shared narratives and corporate trust inside the firm has diminished the organizational ability to create and foster cognitive corporate capital (Nahapiet and Ghoshal 1998).15 The erosion in the cultural position of Ta'as, marked by the shift from the 'heroic period' into the 'competitive era,' has eroded to a certain extent the employee's self-image and sense of belonging to a close-knit 'special community,' solely dedicated to Israel security's needs; a community inducing special codes of conduct, whose members are trustworthy.16

Moreover, a lack of effective ties, and resulting lack of social trust, have prevented the profit-centers' managers from taking risky decisions. Furthermore, the importance of social networks in blocking corporate vitality is demonstrated by the finding that even though individual managers' financial compensation was moderately associated with their center's profitability, they lack the cognitive corporate capital to facilitate such profitable action. They were unable to take initiatives and risks due to the lack of trust within their corporate constituencies (Talmud and Yanovitzky 1998). Consequently, informal social control prevented managers from initiating business ventures and from excelling in their profit centers' performance. Because the relations inside the firm were constructed in the pre-1985 period and were significantly sustained without corporate ability to modify them, Ta'as lost its ability to create cognitive corporate capital. Moreover, the very social structure which had created Ta'as' cognitive corporate capital and was beneficial to its purposive action, become detrimental to the organization, impeding it from attaining its corporate goals, turning a once-beneficial attribute into a corporate liability.

MANAGING CORPORATE SOCIAL CAPITAL

The case of Ta'as corroborates Gargiulo and Benassi's conclusion (this volume) that strong reliance on cohesive networks, which had been beneficiary to the firm, prevented its executive officers from managing their corporate networks' composition, and therefore compromising Ta'as' adaptability.

To the extent that a firm has structurally equivalent competitors in a market niche (Burt and Talmud 1993), it needs to develop other compensating forms of corporate assets to promote its comparative advantage. Burt et al. (1994) demonstrate that for those corporations weakly situated in the exchange network, 'strong organizational culture' is moderately associated with profitability. One can conclude, then, that where competitive corporate structure cannot be advantageous,


imposing unmanageable constraints on the firm, it is especially useful for its managers to develop other kinds of more manageable corporate social capital, facilitated by other dimensions of social structure.

Nevertheless, investing in corporate social capital is not risk-free. Precisely because corporate social capital is relation-specific, it is highly costly. The keyword in minimizing these costs is balance. A firm must 1) balance its investment in three modes of corporate social capital; and 2) carefully determine how to construct each investment so its sunk cost would be more sensitive and flexible to changing conditions. The case of Ta'as reveals that concentrated, over-excessive investment in one kind of corporate social capital may later impede corporate goal attainment. This inference is consistent with Uzzi's conclusion that 'a firm's structural location ... can significantly blind it to the important effects of the larger network structure' (Uzzi 1997a).

Political corporate capital and cognitive corporate capital may be, in principle, more flexible, indeed even more elastic, and thus may seem to be more changeprone, because they are not merely technologically-driven. In other words, they are more affected by the social construction of the firm (Berger and Luckman 1967; Cicourel 1973; Nahapiet and Ghoshal 1998). Yet the paradox is that the social structure creating political and cognitive corporate capital is one of the most rigid dimensions of the firm. Moreover, political and cognitive corporate capital are costly because they are relation-specific and indivisible. The litrature on corporate social capital clearly indicates that value is created through exchange and by combining assets. This process takes both time and relations-specific investment enabling business organizations to develop, nurture and exploit political and cognitive corporate structures (Nahapiet and Ghoshal 1998: 257-259). Yet the 'sticky' nature of the social structures producing them induces the alternative hypothesis that to the extent that a firm holds crucial investments in corporate capital, its barriers-to-exit are costly as well. The above-mentioned case of Ta'as indicates that sometimes these two rather neglected aspects of corporate social capital can be highly rigid precisely because they are socially situated.

DIRECTION FOR FUTURE RESEARCH

The growing number of managerial perspectives on knowledge-based firms, flexible specialization and the high-technology sector have raised the awareness of all three aspects of corporate social capital among managers, business scholars and practitioners. I would hypothesize that firms investing in cognitive corporate capital (embedded in intra-organizational relations) may be found among: a. Mature technological business, having lengthy learning curve; b. Service industries who focus on a market niche (Porter 1980), wishing to edge

out their competitors (Burt and Talmud 1993); c. Potential entrants developing competence-destroying technology (Tushman and

Anderson 1986).

I also would hypothesize that 1) to the degree that the content of relations revolves around identity formation, shared narratives and tacit norms, the relational management is (after evolution) more rigid, and 2) to the extent that a network

Corporate Social Capital and LiabiJity- 117

circulates strategic options, entrepreneurial opportunities and material resources, its relational management is more flexible. I would put forward the proposition, accordingly, that one can order the degree of managerial flexibility of each form of corporate capital as the follows: competitive corporate capital is the most flexible, then political, and the last in order is cognitive corporate capital. Future research should examine relational management and possible tensions between these three forms of corporate social capital.

I would like to thank Paul Ritterband and the two volume editors for their valuable suggestions.

NOTES

I. This definition is slightly different from those of Coleman's (1994: 170) and Nahapiet and Ghoshal (1998: 243). In this chapter I use the term corporate social capital in general, and for simplicity, I refer to each kind of corporate social capital as 'corporate capital' (i.e. 'cognitive corporate capital' means 'cognitive corporate social capital'). 2. The term 'control' is used here to refer to what Alfred Marshall calls 'the external economy' of the firm, including its relational management, and its 'internal organizational economy,' including governance structure (see also Simon 1991). 3. I limit my discussion here only to kinds of corporate social capital-tangible or virtual-which benefit corporate performance. 4. Competitive social capital is a network-oriented conceptualization of economic action (see Gabbay 1997: 17-20). 5. On the difficulty in shifting social capital see Gargiulo and Benassi (this volume). 6. My use of the term 'cognitive corporate capital' is akin to the 'cognitive dimension' of social capital used by Nahapiet and Ghoshal (1998). It is a generic term, and I do not differentiate here between subforms of cognitive corporate capital (intellectual corporate capital, cultural or symbolic corporate capital), as all of these sub-forms are intangible assets of the firm, embodied by social relations (Barney 1991 ; Nahapiet and Ghoshal 1998). 7. This is a structural form of social capital (Gabbay 1997: chapter I), which is akin to Coleman's closure theory of social capital (1988), and to Lindenberg' s definition of 'self-command capital' (Lindenberg 1993). 8. See, for example, Weick (1979). See Moran and Ghoshal (1996) and Nahapiet and Ghoshal (1998) for its relevance to organizational advantage. A useful manifestation of the self-awareness of firm-based knowledge is the emergence of ERP systems. The sociological version of rational choice terms it 'selfcommand capital' (Lindenberg 1993), while economic theory labels it 'self control,' involving two contradictory forces in the organization: farsighted planner and myopic doer (Thaler and Shefrin 1981). Yet economic theory seldom treats its problem setting in terms of relational management, apart from Williamson (1994). 9. On the relations between networks and shared narratives, see especially White (1992). 10. As I limit my summary here to the firm-level of analysis, I exclude the rich literature relating to individual and market levels of analysis. I specifically treated these levels elsewhere (Talmud 1992, 1994; Talmud and Mesch 1997; Izraeli and Talmud 1997, 1998; Talmud and Izraeli 1998). 11. For more on the conditional nature of social capital, see Han and Brieger (this volume). 12. A complete, in-depth description of the case study is in Talmud and Yanovitzky (1998). I include here only a few examples drawn from the case in order to illuminate my analytical substance. 13. Their marketing was performed solely by an Armament Export Unit at the Ministry of Defense. 14. All Ta'as' employees were tenured and unionized. In 1997. Ta'as outsourced its computation center to a British-owned company which hired Ta'as former employees. 15. On the relationship between personal relations inside the organization and the creation of cognitive and even intellectual corporate capital, see Nahapiet and Ghoshal (1998). 16. On the importance of a sense of community to the creation of corporate capital see Nahapiet and Ghoshal (1998: 258).

•

Dimensions of Corporate Social Capital:

6 Toward Models and Measures

ABSTRACT

Shin-Kap Han Ronald L. Breiger

Despite an emerging consensus on the importance of corporate social capital, little work has been done on the analytical problem of which aspects, precisely, of a corporate network might be identified as manifesting the concept. Where in a specific configuration of network ties is the corporate social capital located? Is network capital a unitary phenomenon or are there various ways to conceptualize it? In addressing these questions, we formulate models for corporate networks that produce counts for the expected number of ties between each pair of actors on the basis of sets of parameters which are themselves measures of network capital. The model we prefer decomposes a network into separable dimensions comprising status, volume, and proximity. We apply the models to a network of 'doing deals' in which billions of dollars of finance capital was raised by syndicates of major U.S. investment banks, data of Eccles and Crane (1988). We show that the model performs well with respect to empirical validity. The modeling framework can be applied and extended to other corporate network settings, and provides measures appropriate for theoretical analyses of markets and corporate relations conceptualized as embedded within social fields.

INTRODUCTION

Analysts as diverse as Coleman and Bourdieu put forward virtually identical definitions of 'social capital' as denoting the resources for social attainment that individuals acquire through networks of mutual acquaintance, obligation, and information channeling. Bourdieu defines social capital as the sum of the resources, actual or virtual, that accrue to an individual or a group by virtue of possessing a

Dimensions of Corporate Social Capital- 119

durable network of more or less institutionalized relationships of mutual acquaintance and recognition (Bourdieu and Wacquant 1992: 119). Coleman (1990: 302) too emphasizes that, unlike physical capital and human capital, social capital 'inheres in the structure of relations between persons and among persons.' Relevant here is the considerable body of cross-national research on the impact of social networks and the transmission of job information on income and occupational attainment (reviewed in Coleman 1990: 302; Lin 1990: 250-52; Burt 1992: 11-13), productivity (Bulder, Leeuw, and Flap 1996), and the organi-zational side of job searches (Marsden and Campbell 1990). Studies such as these illuminate concrete mechanisms by which individuals are linked to larger structures through organizations and labor markets (see the review in Breiger 1995).

Much of the recent excitement generated by research on corporate networks results from a focus on how the structure of their ties both affects and results from the interests, resources, and positions of firms in the network-i.e., their social capital. Galaskiewicz (1985) studies business philanthropy as relations among corporations creating 'a grants economy.' Baker (1990) examines the interface between corporations and investment banks with relation to power-dependence concepts. Leifer (1990) seeks to explain authority and market relations among organizations as embedded within the multiplex relations among key actors. White (1992) emphasizes the mutual relations of corporate identity and intercorporate structures and processes of control. Focusing on profit and a typology of markets, Burt (1992: 82-114) shows how firms in production markets can use gaps in social structure-'structural holes'-to their advantage in negotiating transactions with suppliers and customers. Podolny (1993, 1994) theorizes and studies how status orders arise from market and network relations among firms. Haunschild (1994) investigates the effects of interorganizational relationships on the decision of how much to pay when acquiring another company. Han (1994) shows how the interplay between inclusion and exclusion among firms in a market yields a status dimension affecting networks of imitation leading to isomorphism in the selection of audit services. Many chapters in this volume also deal with the issue in a variety of contexts.

NETWORKS AND DEALS

The new work on corporate social capital, such as the studies cited above, has resulted in an explosion of new thinking and new knowledge about social networks. Nonetheless, little work has been done on the analytical problem of which aspects, precisely, of a corporate network might be identified as manifesting the concept. Where in a specific configuration of network ties is the corporate social capital located? Is network capital a unitary phenomenon or are there various ways to conceptualize it? If multiple meanings exist, how might each of them be measured? These are the questions concerning which we seek to contribute clarification and analysis.

To illustrate the scope of our contribution, including its limitations, consider the network of co-management relations among the major investment banks in the u.s. in the 1984-86 period. Eccles and Crane (1988) show that, in order to do their job effectively, investment banks create a complex network of ties to other banks, in the


form of syndicates containing 'lead manager banks' and 'co-manager banks' with a separate syndicate organized around each specific 'deal' that is successfully put together. Within the syndicates. lead managers and co-managers do the bulk of the distribution. and work most closely together in the underwriting. In the aggregate these 'deals' generated billions of dollars of financial capital in the mid-1980s in the U.S.

The data in our Table 1 are taken from Eccles and Crane's appendix (1988: 230-31). Rows and columns list the major investment banks in the identical order. Rows index each bank in its role as 'lead manager' of a 'deal' in the capital market. that is. as a bank that works with a group of co-managers to form a syndicate to underwrite a security issue. A lead manager normally 'runs the books' (manages the underwriting and determines distribution allocation) and is usually the investment bank that originated the 'deal' (Eccles and Crane 1988: 237). Columns index the same banks in their role of 'co-manager': banks that work with the lead manager and often a group of other co-managers in the syndicate. These are the top investment banks in the country.! Entries off the diagonal in the Table are frequency counts of joint participation in deals. For example. during the study period Salomon Brothers served as the lead bank in 161 deals in which First Boston was a co-manager (see [row 1. column 2] of Table I). whereas First Boston served as lead manager in 118 deals in which Salomon Brothers was a co-manager in a syndicate ([row 2. column 1]). Entries on the diagonal report the number of times that each bank served as lead manager without any other top bank serving as a co-manager (either because it was sole lead manager or because no co-managers were among the nineteen top firms). For example. during the study period Salomon Brothers led 609 deals without participation of other banks listed in the Table (see [row 1. column 1] of Table l)?

THE SIMMELIAN PROBLEM OF ORDER

Table 1 is a network of ties among corporations. These network ties are at once collaborative and adversarial. The ties are adversarial in that a fixed number of securities must be allocated among the members of a syndicate. and also due to the struggle among banks for recognition and position. The ties are collaborative in that the offering must be distributed to desirable investors in a timely manner (Eccles and Crane 1988: 93-94). This world of 'doing deals' is not a Hobbesian war of all against all. which necessitates external control. but rather an instance of what we might term the Simmelian problem of order. Georg Simmel. a theorist of sociology's classical period. identified within certain forms of conflict a 'fight of all for all' entailing the internal social control of an intrinsically ordered interweaving of relations based on 'the possibilities of gaining favor and connection' (1955: 62). The Hobbesian problem of order. as it is described critically by Markovsky and Chaffee (1995: 255; Macy and Flache 1995) portrays the structure as seeking to extract from its components something that they would not otherwise provide. We believe that recent work on solidarity. with a new focus on reachability and relative unity of a structure (Markovsky and Lawler 1994; Markovsky and Chaffee 1995) and on models of macro-structure (Breiger and Roberts 1997). enables contemporary

g 0- -T

able

1. D

ata

of E

ccle

s an

d C

rane

(19

88, p

p. 2

30-3

1) o

n ti

es a

mon

g in

vest

men

t ban

ks,

1984

-198

6

Co-

Man

ager

s:

I 2

3 4

5 6

7 8

9 10

Il

Lea

d M

anag

ers:

1

Sal

omon

Bro

ther

s 60

9 16

1 12

3 23

86

24

3 34

73

36

19

23

2 F

irst

Bos

ton

1I8

33

1 15

9 15

10

0 1

lI

20

122

35

19

6

3 G

oldm

an, S

achs

98

64

44

1 Il

46

57

10

49

30

4

5

4 D

rexe

l 9

12

7 69

9 17

21

7

1 17

5

5

5 S

hear

son

Leh

man

78

99

75

8

309

59

17

28

IS

24

25

6 M

erri

ll L

ynch

12

3 98

47

16

63

24

9 15

32

24

27

17

7 P

aine

Web

ber

28

18

6 8

30

19

420

2 10

8

Il

8 M

orga

n St

anle

y 93

65

57

32

21

56

6

92

28

12

16

9 K

idde

r, P

eabo

dy

22

Il

22

16

13

31

3 7

322

7 17

10 P

ro-B

ache

2

7 2

2 5

3 I

0 3

274

4

II

E.F

. Hut

ton

4 2

4 5

2 I

I 0

I 3

227

12

Sm

ith

Bar

ney

9 2

3 3

8 4

7 4

4 I

9

13

Bea

r, S

tear

ns

2 8

2 4

4 6

5 3

1 1

I 14

D

ean

Wit

ter

3 9

6 3

8 7

3 0

6 5

12

15

Dil

lon,

Rea

d 3

9 4

0 3

2 0

I 2

0 2

16

Ale

x. B

row

n 3

2 2

4 3

3 I

2 5

0 2

17

DU

1

1 0

1 2

0 2

2 0

1 7

18

L.F

. R

oths

chil

d 2

0 0

0 6

I 1

0 I

0 1

19

Laz

ard

Fre

res

1 3

5 0

4 7

1 0

3 3

1

12

13

14

IS

20

17

IS

7

Il

27

2 7

10

1 18

4

12

23

IO

0

33

8 21

5

Il

20

13

2

6 4

2 1

IS

18

8 4

19

2 4

3

4 5

5 0

6 5

II

1

120

6 9

0

1 13

1 1

I 9

5 86

2

2 I

2 86

1 0

0 0

0 0

0 0

2 2

4 0

0 1

0 1

16

17

4 4

2 4

IS

I 5

8

10

8

14

9

I 0

7 9

I 7

0 4

I 2

3 2

1 I

0 4

I 2

80

1

1 83

3 2

0 1

18 7 2 3 2 2 5 3 9 2 4 1 0 I 1 0 3 0 71 1

19

18 3 7 2 2 7 0 3 0 0 2 0 1 5 0 3 0 0 14

I::'

~.

:;, '" o· :;, '" o .....,

(j .8 o ~ til o n ~

(j .g. ~ -tv


researchers to address the alternative, Simmelian, problem of order: that 'the structural membership of the individual in his group always means some mixture of enforced limitation and personal freedom' (Simmel 1959: 47; for further exposition see Breiger 1990). We seek formal models and quantitative measures that allow us to tease apart the subtle intermixture of various kinds of organizing principles (hierarchy, reciprocity, and so forth) that constitute the intercorporate network.

The technical infrastructure for our modeling project is already in place. Within the class of multiplicative and related models for contingency tables of frequency data, we focus on quasi-symmetry and its special cases and generalizations. Maximum likelihood techniques for estimating parameters are well known and widely available (see Sobel, Hout, Duncan 1985; Goodman 1984; Clogg and Shihadeh 1994). Thus freed from the necessity of creating new models, we concentrate instead on the call of Sobel, Hout, and Duncan (1985: 371) for researchers to 'attach new meanings to already existing models of this type and to generate new models for the square table.' Related work of ours includes Breiger and Roberts (1998), Breiger and Ennis (1997), and Han and Breiger (1996); see note 6 below.

MODELS AND MEASURES

To establish notation, we will use fij to refer to an observed count in Table 1. For example, fl2 is the count in Table 1 at the intersection of row 1 and column 2; as we noted earlier, this entry reports that, during the period studied by Eccles and Crane, the first-listed bank, Salomon Brothers, was the lead manager in 161 deals for which the second-listed bank, First Boston, served as a co-manager. The set of fij defines a quantitative network of ties among these investment banks. We seek to model this network so as to bring out aspects of social capital. We will use Fij to refer to the expected count that we derive for cell (i,j) on the basis of a model. We assume the table is square and of size gxg, with the same entities (in our case, a set of investment banks) indexed by rows and columns.

We consider several models, all of which satisfy or are special cases of quasisymmetry, a model which we will explicate. Other kinds of models might also be considered for these data, but we have found quasi-symmetry models to be particularly helpful in studying corporate social capital (cf. Friedkin 1998). These models are distinctive in that they decompose a network of counts into separable aspects comprising status, volume, and proximity, as we will demonstrate. The existing literature on corporate relations is rather vague on whether these aspects are conceptually separate, and their meanings have often been blurred. In fact, they might be strongly related to one another; however, this should be an empirical question. We seek explicit definitions of these concepts as features of networks, thus enabling us to jointly model and measure the concepts in relational terms.

Following the path-breaking work of Sobel, Hout, and Duncan (1985), we parameterize the quasi-symmetry model as follows:

(1)


Terms on the right-hand side are parameters to be estimated. Several stipulations are imposed. An <X is estimated for each of the g actors indexed in the table, and the product of the <Xj is 1. It is understood that ~i = ~j if i = j. The o's are symmetric, Oij = Oji. Maximum-likelihood procedures for estimation of the parameters and expected cell frequencies under this log-linear model of quasisymmetry are well-known; interested readers are referred to Bishop et al. (1975), Goodman and Clogg (1984), Agresti (1990), Clogg and Shihadeh (1994).

It will be useful to consider the pair of cells Fij and Fji. For example, Figure 1 indicates expected frequencies among three pairs of banks.3 We see for example that, according to the model, Goldman Sachs is the lead manager in an estimated 45.618 deals in which Shearson Lehman is a co-manager. In the opposite direction, Shearson Lehman is lead manager in a larger number of deals, estimated at 59.245, in which Goldman Sachs is co-manager.

The ratio between the two, then, indicates and measures the extent to which the pairing between the two actors is asymmetric. In pairing Shearson Lehman (i) and Goldman, Sachs (J), for example, it is the former who is more likely to be the lead manager, as shown in the first panel of Figure 1 (59.245 versus 45.618). The ratio between the two, R ij, then, is 1.299, showing that Shears on Lehman (i) is 1.299 times more likely to be the dominant partner (i.e., lead manager) vis-a-vis Goldman, Sachs (j). In other words, Rij is a measure of the dominance of i over j.

We may form an entire g x g matrix, R, of such ratios, with

R-~ m ij=T. )1

Although the equation above is definitional of dominance, we now turn to a result. The matrix of R is a function of one dimension: the <X parameters in our model. As may be seen by substitution of equation 1 into this definition of R:

F. a. (3) R .. =-2..=_) I) F

ji a i

Each actor in the network thus has a unique value of alpha, and the asymmetry in any pairing between two actors can be described in terms of the ratio between the two alphas. With respect to the tie in network R from Shearson Lehman to Goldman, Sachs, for example, the ratio of the alphas (shown in panel b of Figure 1) is .667/.513, which equals 1.299, the value of Rij given previously in the text. The vector of alphas forms a linear ordering and in this sense captures the dominance hierarchy among all the actors in the network.4

Furthermore, if we define a column vector a to have 1/ <Xj as its j-th element, we can trace the relationship between our alpha and a conventional measure of centrality in the social networks literature (Bonacich 1972, 1987; Wasserman and Faust 1994).

The centrality of an actor is very often operationalized as the sum of an actor's social connections, weighted by the centrality of the others to whom the focal actor is tied. A natural implementation of this concept of centrality is that centrality is an eigenvector of the matrix representation of the social network (see also the applications of eigenvector centrality measures in studies of corporate interlocks


Shearson Lehman

45.618 ) \ 20.665

9.245 5.96~

Goldman, ~ Smith

( sa:hS ~ Ba~ney \

~ 3.860 ~

Panel a: Expected frequencies (Fij) from Model 3 in Table 2.

Shearson Lehman

'1(513)~ Goldman, Sachs

(.667)

____ .~ Smith Barney

2.668 (1.778)

Panel b: Ties between firms (Rij = Fij I Fji). Values in parentheses are Clj.

Shearson Lehman

Goldman, Sachs

Panel c: Oij; see equation 5. .027

Smith Barney

Figure 1. Modeling the network of investment banks: a three bank example


such as those referenced in Mizruchi et al. 1986). With respect to the network R defined above, we have

Ra=A.a (4)

with A. = g, the number of actors in the network. This equation establishes that the ex; parameters of the quasi-symmetry model define an eigenvector of matrix Rand thus, in our modeling context, represent the dimension of networks that is typically captured by the family of centrality measures including prestige, status, and popularity.5

This interpretation of the ex; parameters illustrates the most fundamental feature of our modeling approach: We formulate models for the network that produce expected cell counts for the number of ties between any two actors on the basis of parameters (a., ~, 8) which themselves are measures of network capital.

The models and measures are duals to each other, as roles and positions are duals to the structure. The models themselves (such as quasi-symmetry) are wellknown. Our distinctive contribution is to develop their relevance to the analysis of network data on counts, such as the co-manager ties of Table 1. From a statistical point of view, the fit of the models to observed networks may be assessed by means of standard maximum-likelihood chi-square procedures, making them feasible to apply to square tables of frequency data in many different substantive contexts.6

In models of quasi-symmetry for network data, ~i is the average volume of ties sent and received by actor i, controlling for the other parameters in the model. In the present context, this parameter indexes a bank's total involvement in deals (whether as a lead- or as a co-manager, without distinguishing between these two roles but focusing only on the extensiveness of its ties). In this sense, ~i indexes the (relational) volume of an actor.

The third set of parameters, the 8ij' measures how strongly or closely i and j are related to each other, net of a. and ~. The raw expected count, Fij, is by postulation a function of all three parameters and thus is not a proximity measure. The geometric mean of Fij and Fji could be a proximity measure but it also suffers from a built-in dependence on all three parameters. However, if we norm this product appropriately, then the following is a measure of proximity between the two actors that is net of the asymmetric status (a.) and relational volume (~) effects, as may be seen by substitution of equation 1 (see also Sobel et al. 1985: 364):

(~X~)=5, (5)

Although it is usual to interpret the 8ij parameters with reference to the social process of reciprocity (e.g., Sobel et al. 1985), in the context of our corporate data we prefer the more direct interpretation given by the equation above: 8 is a measure of the social proximity of two actors. The relations themselves may not be symmetric, whereas 0 is an average of their intensities, the degree to which two investment banks are likely to encounter each other in a syndicate formed to put together a deal. In Figure I, panel c reports the estimated 8 parameters (computed as


in the equation above) for three illustrative banks. It is seen that Goldman Sachs and Shearson Lehman are much closer to one another (on the basis of the average net intensity of their relations) than either bank is to Smith Barney. The 0 parameters give us a network of proximity coefficients. The oij might well be related to differences in status (<Xj I Clj) and in volume (/3i I /3j), as they indeed appear to be in the illustrative example of Figure 1. Such relations among the parameters can be investigated on the basis of a model that decomposes the expected cell counts into just these three components.

A variety of models simpler than quasi-symmetry may be postulated; see Table 2. The simplest model we consider consists solely of the /3 parameters for volume; that is to say, the model imposes that all <Xj = 1 and all Oij = 1. This model is taken as the baseline model for analysis of square tables by Hope (1982), who terms it the 'halfway' model, and it is discussed by Goodman (1985) and by Hout, Duncan, and Sobel (1987: 152), who note that the model imposes both independence and marginal homogeneity.

The conventional model of statistical independence for rows and columns of a square contingency table may be obtained from the 'halfway' model by adding to the halfway model estimates of the <Xj parameters measuring dissimilarity among the average counts in column j and in row j; see Table 2. In the independence model no pairwise interactions are allowed (all oij = 1). An alternative generalization of the 'halfway' model is to allow symmetric pairwise interactions (Oij) but to preserve marginal homogeneity (<Xj = 1); this alternative defines the usual model of (full) symmetry. Putting together, so to speak, the model of independence and the model of full symmetry yields the model of quasi-symmetry that we have been discussing in this chapter (see Table 2). In other words, we relax the marginal homogeneity condition, specified in the full symmetry model (<Xj = I), while allowing the pairwise interactions, prevented in the independence model (all Oij = I), with some constraints.

The remaining models of Table 2 provide more parsimonious representation of the symmetric interaction parameters (the oij), portraying these pairwise terms as one or more dimensions of interaction, rather than requiring one parameter for each pair of actors. Notice for example that the 'homogeneous RC(I) model' requires estimation of only (g-l) parameters more than the model of independence (the difference in degrees of freedom is 305 - 287 = 18; see Model 3 in Table 2) in order to represent all the pairwise proximities in terms of a single dimension (Jl). The 'RC(2)' model uses two dimensions in preference to estimating a parameter for each pair of actors. All the models in Table 2 are well known and are discussed in detail by Goodman (1984, 1985) and others (Sobel et al. 1985; Hout et al. 1987; Agresti 1990). As shown by the parameter specification in Table 2, all these models are models of quasi-symmetry.

The fit of any of these models to data on frequency counts may be assessed in the usual way by means of a comparison of the chi-square and the degrees of freedom left by the model (see columns labeled G2 and df in Table 2). Or, one may measure improvement in fit relative to a baseline model. The last column in the

--l ~

0'

IV

Tab

le 2

. Mod

els

of q

uasi

-sym

met

ry *

Para

met

er S

peci

fica

tion

Clj

Pi

I)ij

df

G/

bie

1-{

Gk2

/Gl}

kMod

el:

1 H

alfw

ay

Pi

323

1624

.4

-109

9 0.

0%

2 In

depe

nden

ce

Clj

Pi

305

1248

.0

-132

4 23

.2%

3 H

omog

eneo

us R

C(I

) C

lj P

i exp

{-~¢{u

; -ll

jY}

28

7 77

9.1

-164

1 52

.0%

4 H

omog

eneo

us R

C(2

) C

lj P

i ex

p { -.!

. r.¢

k{u

t; -Il

kjr}

2 k=

l 27

0 64

4.6

-163

2 60

.3%

5 Fu

ll Sy

mm

etry

P

i I)

ij

171

777.

3 -6

64

52.1

%

6 Fu

ll Q

uasi

-Sym

met

ry

Clj

Pi

I)ij

153

330.

9 -9

59

79.6

%

* As

impl

emen

ted

in th

is p

aper

, all

mod

els

fit th

e di

agon

al c

ells

exa

ctly

. M

odel

s ar

e or

dere

d by

thei

r deg

rees

of f

reed

om.

o ~.

::s

en o· ::s

en

o ......

('} .a ~ CD

CIl o ~. a ('

}

{l ~ -IV -.

.l


Table provides such a measure relative to Modell, the 'halfway' model. Alternatively, because the number of deals indexed in Table 1 is large, and to avoid overfitting (that is, the inclusion of terms in a model for which we have little or no substantive interpretation) we may apply the Bayesian information criterion (BIC) of Raftery (1986; see also Raftery 1995); values of this criterion are given in Table 2 for each of our models. In applying BIC one chooses the model with the lowest BIC value.

In fact, all the models of Table 2 are relatively more or less parsimonious efforts to represent quasi-symmetry. For example, the oij values estimated from the onedimensional model of Table 2, RC(l), correlate .82 with the Oij parameters estimated for the full quasi-symmetry model, and the single dimension estimated for the onedimensional model in Table 2 correlates .98 with the first dimension of the twodimensional model, RC(2).

According to the BIC criterion the best balance of parsimony and substance is attained by the one-dimensional model, RC(I), with a single dimension characterizing scores for both the row and (identically) the column categories of our data (see Model 3 in Table 2). This one-dimensional model differs from (full) quasisymmetry only and precisely by modeling the pair-wise Oij parameters by means of a single dimension of proximity, as specified by the l1i parameters in the column of Table 2 labeled 'Oij'. The illustrative example of Figure 1 above is based on expected counts and estimated parameters from this RC(I) model.

NETWORKS AND OUTCOMES

In this section, usil1g the model discussed earlier, we examine the social structure of investment banking industry by analyzing the data on the ties between investment banks. Our preferred model (Model 3 of Table 2) fits the data with three dimensions: status (a), volume (~), and proximity (11). The estimated parameters are reported in Table 3 for each of the nineteen investment banks in the dataset.

The model decomposes the relational structure among the investment banks into three components. The first one, a, captures the disparity between playing the role of the lead manager and the role of the co-manager. In other words, it measures the propensity of an actor to be on one side of this asymmetric relationship versus the other. In particular, the reciprocal of a measures the propensity to be the lead manager in a deal rather than a co-manager, and indicates each actor's position in the linear ordering of the dominance hierarchy among all the actors in the network.

The estimated a's closely reproduce the well-known 'bracket' structure in the industry, the strong and elaborate status hierarchy among the investment banks, typically shown in the 'tombstone' advertisements (Hayes 1979; Eccles and Crane 1988; Podolny 1993). All of the six special (or bulge) bracket firms-First Boston (#2); Goldman, Sachs (#3); Morgan Stanley (#8); Salomon Brothers (#1); Merrill Lynch (#6); and Shearson Lehman Brothers (#5)-are found at the top of the list of estimated a's, with Paine Webber (#7) being an exception in the ordering. Drexel (#4) and Dillon, Read (#15), follow closely after these six.


Table 3. Estimated parameters for Model 3 of Table 2

ID Investment Bank Special Parameters No. Name Bracket In(1/a) In(13) In(~)

I Salomon Brothers Yes 0.9099 3.6608 1.4346 2 First Boston Yes 0.8301 3.3161 1.7183 3 Goldman. Sachs Yes 0.4056 3.2473 1.4390 4 Drexel No 0.3369 3.4433 -0.5509 5 Shearson Lehman Yes 0.6670 3.2002 0.6637 6 Merri11 Lynch Yes 0.4041 2.9608 1.0601 7 Paine Webber No 0.5593 3.2998 0.1743 8 Morgan Stanley Yes 0.8607 2.6913 1.1855 9 Kidder. Peabody No 0.2063 2.9904 -0.0587

10 Pru-Bache Securities No -0.6952 2.4590 -0.4910 11 E.F. Hutton No -0.9994 2.2128 -1.5178 12 Smith Barney No -0.5757 2.1059 -1.0161 13 Bear. Stearns No -0.8767 1.9993 -0.4697 14 Dean Witter No -0.2102 2.1221 -1.2014 15 DiUon. Read No 0.3217 2.3880 0.3272 16 Alex. Brown No -0.3679 2.0071 -0.3828 17 DU No -1.2533 1.5828 -1.4986 18 L. F. Rothschild No -0.4190 1.9218 -0.9267 19 Lazard Freres No -0.1043 1.2674 0.1109

2.v 2.u

» ~ 1 ; 1 , 1.0' i. 1.0' , i

1: ~5 i j 4 a ~5 i ~ " j j :c .. ~ 0.0' Yl 0.0 ~9 '" ~4 ~4 V:

,... ~d.6 111 ~6

" .r. If If c ~o ~o 3-1.0 -1 .0

::: ~3 ~1 ~1 ~3

c ....

-2. 11 _2.n

~7

-2.0 ·1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0

In(Beta). Standardized In(I1). Standardized

Figure 2a. Status (lIa) by Volume (13)· Figure 2b. Status (lIa) by Proximity (p).

• ID numbers in the Figure are keyed to the name of the investment banks in Tables I and 3. Parameters are standardized by taking Z-scores.


The second parameter, p, taps into the volume effect, the extensiveness of a firm's involvement in deals either as the lead manager or the co-manager. Salomon Brothers (#1) is on top, followed by Drexel (#4). These two parameters, a and p, are highly correlated with each other. Eccles and Crane (1988) observed that a firm's hierarchical position is based partly on its volume, and the volume of securities it gets to underwrite and sell is based in turn on its hierarchical position.

Consider, however, the joint distribution of the two. The scatterplot in Figure 2a shows a pattern of deviation from the expected association between the two sets of parameters, lIa and p. Although most of the firms are along or near the regression line describing the expected linear relationship between status and volume, there are two groups of firms that fall far outside the expected range.

1. On the one hand, there is a group of firms-Morgan Stanley (#8); Dillon, Read (#15); and Lazard Freres (#19)-known to be Establishment firms that enjoy high status relative to their volume (Podolny 1994; cf. Stuart in this volume). Morgan Stanley is known to be 'the bluest of the blue-chip investment bankers' (Hayes 1971: 139; Eccles and Crane 1988); Dillon, Read used to belong to the special bracket (Hayes 1979); and Lazard Freres represents the old Wall Street (Stewart 1991). This provides an example of how the inertia of the status hierarchy (a) mitigates the raw influence of the market as indexed by overall volume (P).

2. On the other hand, another group suffers low status despite high volume. Drexel (#4) is the most prominent case among the latter group of outliers, which also includes Prudential-Bache (#10) and E. F. Hutton (#11). Drexel's reputation as an aggressive newcomer and its strong association with Junk bonds' provide partial explanations for this discrepancy,? with Eccles and Crane (1988: 115-16) noting in some detail that 'during our project the firm was described to us .. . in the most unflattering ways' (see also Stewart 1991). This is an example of how the hierarchy, defined and legitimized by the participants themselves, acts as a conservative mobility barrier to firms trying to shift their position.

Volume is important, and it is highly correlated with status, yet the two are separate dimensions, and one does not necessarily translate into the other. The contrast between Lazard Freres (#19) and Drexel (#4) clearly illustrates this point (also see the more recent case of DU-Donaldson, Lufkin and Jenrette, #17-in Doherty 1997). The model we propose is precisely suited to such a setting, for it allows the two conceptually independent components to be separated out from each other. Although the two are rather highly correlated in this case, for instance, the underlying structural dimensions they tap are distinct. As shown in Table 4, it is p that best captures the overall volume effect, as in total market share and number of issues. By contrast, lIa does better at explaining the dimension that is associated


Table 4. Correlations between I/a and ~ and other select variables *

Variable InOla) In(S)

Total Market Share, 1984-1986 .689 .857 Number of Corporate Securities Issued, 1986 .775 .843

Market Share in Investment Grade Bonds .687 .684

Market Share in Non-Investment Grade Bonds .274 .478

Status Score (Podolny 1993) .621 .535

* The sources for the other variables are Eccles and Crane (1988, Tables 5.4 and A.6) and Podolny (1993).

with prestige or status, an excellent example being the status scores computed by Podolny (1993, 1994) applying a widely used measure of centrality to the placement pattern of investment banks in the tombstone advertisements. Also revealing is the way they correlate with the market shares of two different financial products, investment grade bonds and non-investment grade bonds (see note 7). With the former both a and ~ are correlated to the same degree, while for the latter, a high risk product (Podolny 1994), the correlation with a is not statistically significant.

The third dimension, J.l, is about how close or distant the firms are from each other net of a and ~. The more deals a pair of firms do together, the closer they are to each other vis-a-vis the others. The vector of J.l'S thus forms a one-dimensional space along which each firm can be located, and the proximity between any pair of firms can be obtained from the distance between the two. The smaller the difference between J.li and J.lj' the closer firm i and firm j are, and vice versa. For example, Salomon Brothers (#1; J.li = 1.4346) and Merrill Lynch (#6; J.lj = 1.0601) are relatively close to each other (J.l; - J.l:i = 0.3745), and they frequently serve as partners.s In contrast, Morgan Stanley (#8; J.li = 1.1855) and Dean Witter (#14; J.lj = -1.2024) are relatively far apart (J.l; - J.lj = 2.3869), and they rarely do deals together.

Plotting relational proximity (J.l) against status (l/a) reveals a very important social dynamic that occurs among the investment banks, that of status homophily. The firms that are close to each other on J.l are likely to be near to each other on a as well. The smaller (J.l; - J.lj), the smaller (<Xi - CX;). Or, to put it otherwise, the investment banks that put together the deals tend to be status equals, which is the central finding in Podolny (1993) based on somewhat different data for these same firms.

This result also precisely matches what Eccles and Crane observed. The top six firms, those in the special bracket, are clustered to the right in Figure 2b. They are close to one another as a result of the security issues they do together. Yet they can be broken down into two groups. The first one consists of Goldman, Sachs (#3), Salomon Brothers (#1), and First Boston (#2). These three rely most heavily on other special bracket firms as co-managers. The second group-Shearson Lehman (#5), Merrill Lynch (#6), and Morgan Stanley (#8)-<10 so less often, bridging instead to banks further down the status ordering. The relative location of the two groups on J.l, i.e., the second group being closer to the rest of the banks to the left, bears out Eccles and Crane's account. In particular, these authors report that 'a partial explanation of the difference between the two groups is that a larger share of


the deals led by the second group, particularly Merrill Lynch and Shearson Lehman, were security issues in which regional firms were used as co-managers to get more retail distribution' (Eccles and Crane 1988: 94).9

IMPLICATIONS FOR RESEARCH AND THEORY

An emerging consensus on the importance of corporate social capital spans a broad spectrum of theorists and researchers, as aptly illustrated in this volume. Although the core ideas seem to be shared by many and have been shown to be operative in a variety of settings, little work has been done to parse out analytically the mUltiple and conceptually distinct aspects of social capital. In examining a network of 'doing deals' constructed among the major U.S. investment banks (Eccles and Crane 1988) from the analytic perspective of log-linear models, we rely on a set of wellestablished techniques. Yet at the same time we heed the call to 'attach new meanings' (Sobel et al. 1985: 371) to the models, and we expand their interpretive horizon by means of our novel application.

We formulate models for the networks that produce counts for the expected number of ties between each ordered pair of actors on the basis of a set of parameters which themselves are measures of network capital. More specifically, the model we prefer decomposes a network into separable dimensions comprising status, volume, and proximity. We show that the model and its associated estimated parameters, which situate the actors within a social space in relation to one another, performs well with respect to empirical validity.

'Something about the structure of the player's network and the location of the player's contacts in the social structure of the arena,' Burt argues, 'create a competitive advantage in getting higher rates of return on investment' (1992: 57). That 'something' is social capital, upon which economic constraints and opportunities that confront a producer are very much contingent (Podolny 1993). The approach proposed in this chapter makes it possible to formulate this intuitive notion in a more explicit and structural manner. Providing measures appropriate to analyze the intercorporate relations, competitive or otherwise, and a way to map the configuration of the market as a whole, the model can easily be applied and extended to other corporate network settings, such as joint ventures, strategic alliances, or R&D collaboration (e.g., Freeman, Pennings and Lee, Smith-Doerr et aI., and Stuart in this volume). The model thus should allow better informed strategic deliberations in a variety of management settings (e.g., Porter 1980).

Furthermore, the model and its parameters offer a way to portray a social field in general, beyond this specific framework. Consider, for instance, different configurations of actors in <X-~ space, as in Figure 2a. For the investment banking industry, there is a high correlation between a firm's volume of underwriting business and its place in the industry'S hierarchy. The two, however, are conceptually distinct. The model we propose is precisely suited to such a setting, for it allows each dimension to be separated out from the other. It should be possible, nonetheless, to find a case completely opposite to ours, one in which the two are correlated negatively-as in luxury goods. Or, one might be able to find a setting in which the mid-sized firms are the most prestigious (cf. White 1988, 1992). In <X-Il space, shown in Figure 2b, we showed precisely a pattern of status homophily. And


yet, other configurations are plausible as well. There are settings that exhibit social proximity between status unequals, as for example between managers and secretaries (Kanter 1977). Curvilinear relationships are also possible, as for example among U.S. Supreme Court justices, where we have argued that those justices 'in between' the coalitions of liberal and conservative ideologies are highest in the status hierarchy (Han and Breiger 1996). Such an inverse-U shape is itself in contrast to the U-shape for plotting status against social proximity that is often postulated in the cases of the middleman minority (Bonacich 1973).

In sum, the theoretical and research work that is necessary to further develop the concept of corporate social capital requires structural and systemlltic measures on networks. In this chapter we have introduced relationally based measures which appear to be most promising for future work.

For comments on an earlier draft we are grateful to the editors and to Robert Faulkner, Noah Friedkin, Joseph Galaskiewicz, Philippa Pattison, and John M. Roberts, Jr.

NOTES

1. That these nineteen investment banks are the major players seems to be a robust finding. Eccles and Crane (1988: 228) report that, 'since nineteen is not a round number, we attempted to chose a twentieth but could not. All the candidates were firms strong in narrow product categories, purely regional firms, or varied in the strength of their performance through the three-year period' of 1984-86. 2. Of these, 488 were deals in which Salomon Brothers was the sole lead manager, and 121 others involved deals with banks other than those listed in Table I. 3. The expected frequenCies in Figure 1 are derived from Model 3 in Table 2 of this chapter. This model is a special case of quasi-symmetry, to be discussed below. 4. This discussion of R;Jis somewhat analogous to related formulations, including those of R. D. Luce (who developed implications of the choice axiom for the scaling of preferences) and of S. E. Fienberg and K. Lamtz (who showed that the Luce model implies a linear preference model in the logit scale, and who formulated aspects of Luce's model as well as other models for paired comparison experiments in terms of the quasi-symmetry model). See Agresti (1990: 370-74) and the brief review that appears in Breiger and Roberts (1998). 5. Equation 4 may be obtained by substituting equation I into the definition of R;J. 6. For example, Breiger and Roberts (1998) and Han and Breiger (1996) examine networks of joining in one another's opinions on the part of members of the U.S. Supreme Court. Breiger and Ennis (1997) examine as a social network a cultural field of relations among writers in KOln, Germany, data of Anheier et al. (1995). 7. A 'junk bond' or noninvestment-grade bond is a certificate of debt promising a high rate of return on investment but carrying a high risk; these securities are often used to finance corporate takeovers. 8. Eccles and Crane (1988: 96) note that 'the tie between Merrill Lynch and Salomon Brothers was particularly strong.' 9. See Doherty (1997) for a detailed account of DU' s (#17) relative location vis-a-vis the others and its recent movement in the parameter space.

•

Organizational Standing as Corporate Social Capital

ABSTRACT

7 Patrick Doreian

Organizations in social service delivery networks interact in order to provide many services to a wide variety of client populations. In the course of these interactions, staff and directors of these agencies form assessments of the utility of working with other agencies. These assessments, as social network information, can be used to operationalize a measure of network generated corporate social capital. Organizations well regarded by other organizations have higher social capital. Further, organizations well regarded by well regarded other organizations have higher social capital. Input-output methods are used to generate measures of standing. These assessments are also disaggregated by sector in a way that permits a comparison of the relative contributions to social capital by sector. Data from the SSDURC project are used to illustrate these methods.

INTRODUCTION

Social service organizations are distributed across a variety of sectors. Scott and Meyer (1991: 117) define a sector as '1) a collection of organizations operating in the same domain, as identified by the similarity of their services, products or functions, 2) together with those organizations that critically influence the performance of the focal organization.' This seems too inclusive and it is useful to modify the first characteristic to focus exclusively on the primary functions served by organizations in their sectors. Two organizations in the same sector need not have the same services and products: a residential facility providing community living for the mentally ill will differ from a multifunction agency and a psychiatric ward in a hospital will differ from both.

Organizational Standing as Corporate Social Capital - 135

The sectors considered here are defined in simple functional terms. Units in the education sector educate students (primarily in schools); units in the health sector provide services to promote good health and to treat ill-health; units in the judicial sector process perpetrators and victims; mental health agencies provide services to deal with people with mental illnesses andlor those who are developmentally disabled while units in the poverty sector attempt to reduce poverty and lessen the adverse consequences of being in poverty.

Scott and Meyer (1991: 120) argue that 'the concept of a societal sector suggests the presence of organizational systems that are, to some degree, functionally differentiated.' The five sectors listed above can be viewed as differentiated systems. Yet there are many overlaps and, with them, potential conflicts. If social service agencies, having differing charters, mandates, philosophies and technologies, are distributed across a variety sectors, there is limited consensus concerning the core technologies used by these agencies. The presence of 'multi-problem clients' makes this even more acute. This is particularly the case when there are poor clients with health andlor mental health problems and who break the law or are victims of others who break the law. Distinct agencies will have claims, with differing degrees of legitimacy across sectors, concerning their organizational domains. Further, the conditions are created where there can be both duplication of services and clients who 'fall through the cracks.'

Social service organizations have to interact under ambiguous conditions in order to provide services and coordinate their activities. It is clear that organizations working together with, or fighting over, specific clients have impacts on each other. Or, if a sequence of services is provided for clients in different agencies, there have to be clear referral pathways linking the organizations. The second defining feature of sectors, as proposed by Scott and Meyer, has limited utility. On one hand, if agencies are linked in networks, they cannot help but have impacts on each other and the condition is satisfied trivially. On the other hand, even if they belong to different sectors they do have impacts on each other and the second feature is contradicted. In the following, Scott and Meyer's second criterion for defining a sector is discarded here.

Regardless of these distinctions, organizations still have to work with each other. This is not a straightforward process and takes place in a web of interorganizational relations (IORs). In part, these relations are generated by funding streams and, in part, they are shaped by local interactions between organizations in some geographical location.

CORPORATE SOCIAL CAPITAL

Social service organizations are embedded in many lOR networks. It is a truism that these networks provide access to needed resources for some organizations and that they place constraints on the actions of other organizations in the network. Network ties are created, changed, maintained or dissolved. In short, the lOR network (ION) changes, and with this change, the opportunities and constraints shift also. Organizations negotiate their way through their changing environments. As these agencies interact, histories of interactions are generated across the network. Some working arrangements are successful while others fail. Organizations build histories


of which ties work for them and which ties do not work. As personnel in these agencies come to form assessments of other organizations, these other organizations get reputations. Shrum and Wuthnow (1988) argue that the reputational status, in part, is driven by past performance.! Such performances-and the shared histories they create-form foundations for the reputations that agencies acquire. Agencies acquiring positive reputations acquire also social capital while those acquiring less positive, or even negative, reputations lose social capital or build up social liability.

Given sector memberships and uses of different treatment procedures, the worth of different technologies becomes ambiguous. There is limited agreement on what constitutes an 'effective' treatment modality and assessing the merits of technologies, together with the organizations employing them, becomes subjective. Knowing the reputation of another organization makes this evaluation simpler. Consider an organization and suppose, for whatever reason(s), there is a need to interact with one or more other organizations. Some criteria are needed to judge if future joint action with another unit is worth pursuing.

One is suggested by institutional arguments where procedures and forms come to be accepted as 'legitimate' in some sense (Meyer and Rowan 1991). Part of this stems from which agencies and programs are funded. Getting funded is a profound legitimating force, assuming the funder is reputable. These funds have three primary sources: 1) extra-local governmental (Federal, State and County level) funding sources; 2) legitimate charities, and 3) agencies purchasing services from one another.

In the mobilization of relations for the delivery of services, agencies build social capital through the quality of the relations that are established and used. Substituting 'agency' for 'person,' Coleman's (1990) insight that social capital is generated through the 'structure of relations between agencies and among agencies' is pertinent. Bourdieu and Wacquant define social capital as 'the sum of the resources, actual or virtual, that accrue to an individual or a group by virtue of possessing a durable network of more or less institutionalized relations .... . (emphasis added)' (Bourdieu and Wacquant 1992: 119). Han and Breiger (this volume) also use this idea as a point of departure for establishing measures of social capital. Burt (1992: 8) views social capital of an actor as the relations the actor has with (relevant) others. Lin (1982) regards an actor's relationships as resources for instrumental action. Coupling these ideas, an organization's network generated social capital comes from the number of ties with other organizations and the evaluated quality of those ties (where the evaluations are made by the relational partners). Further, the network generated social capital will be higher if the favorable evaluations come from highly regarded organizations. The proposal here is to use the reputational standing of an organization as an indicator of its (network generated) social capital.

A MEASURE OF SOCIAL CAPITAL

The foundations for doing this in a social network context can be traced back to Hubbell (1965) who used input-output models to generate a measure of standing for the nodes in a network.2 Doreian (1985, 1987) applied these ideas to actors in small groups and journals in citation networks. For this application, a more important


development is the method provided by Salancic (1986) to disaggregate such standing measures, a procedure of particular use in studying IONs.

Let C be a (gxg) matrix containing the evaluations of agencies (by directors and/or staft) in a square matrix form. In C, cij is the evaluation of agency j by (representatives) of i. Let Tbe a (gxl) vector containing all of the evaluation activity where T=C'u+Cu and u is a unit (gxl) vector. Let diag(1) be a diagonal matrix where T is in the main diagonal. With v a (gxg) matrix of 1 's and D=diag(1)v, a matrix of weights, W, is created by the elementwise division of C by D. Next, let Sj

be the standing of the jth agency, then with ei an exogenous status measure:

(1)

which in matrix form is

s=Ws+e (2)

leading to

(3)

where I is a (gxg) identity matrix. This measure is akin to the measure of status of Podolny and Castellucci (this volume) and, as noted by Han and Breiger (this volume), to eigenvector based measures of centrality.

The operation in equation (3) is used to generate a set of measures of standing for each organization in the network. However, these organizations also belong to sectors and it is useful to disaggregate these measures into contributions to standing, as social capital, from sectors. In the context of a journal citation network, Salancik (1986) has provided a method for doing this. Let M be a matrix with a column for each sector and mik is defined as:

mir 1 if organization i is in sector k o otherwise

The intrinsic value of an organization in the ION can be viewed as stemming from the sector to which the organization belongs. Let ek be this contribution for sector k. and let E=diag(ek) be a diagonal matrix dimensioned by the number of sectors. Equation (3) is then replaced by:

(4)

and if each sector has the same intrinsic value, E is the identity matrix. The measures of standing generated, through the use of equations (3) and (4) operationalize the idea of network based social capital.


METHODS AND DATA

The data used here come from the Social Service Delivery Under Resource Constraints (SSDURC) study. For details see Doreian and Woodard (1992), Woodard and Doreian (1994) and Woodard (1995). An expanding selection procedure was used to identify all of the relevant agencies in three Pennsylvanian rural counties. We use the Fall County network from the first wave of the study. An expanding selection procedure, using both director and staff respondents, was used to identify all organizations in the network (Doreian and Woodard 1992). The data used here come from the director responses to the following question 'To what extent do you feel your relationship with this organization is productive?' with response characteristics: 1, no extent; 2, little extent; 3, some extent; 4, considerable extent and 5, great extent.

A network with sixty five (65) agencies in a social service delivery network is considered here. These agencies are located in a variety of sectors. The education sector is represented by the Intermediate Unit (lUI) providing special educational services and five schools (AGSCH, BSCH, CNSCH, LHSCH, USCH). From the health sector there are two hospitals (BVILL, UHOSP), a provider of human services (GALL), Easter Seals (ES), Family Planning (FP), the local office of the State Department of Health (PH) and the Association for the blind (BLIND).

The judicial sector has 20 agencies in the network. The core agency is Children and Youth Services (CYS). There are four local police departments (CPOLC, GPOLC, MPOLC, UPOLC) and the State Police (SPOLC). Two agencies, the Women's Resource Center (WRC) and the Family Abuse Council (FAC), provide services, usually for women and children, to deal with domestic and marital problems. Present also are many units of the court system-the court office (COURT), Domestic Relations (DR), the District Attorney Office (DA), JUDGEs, Legal Aid (LAID), magistrates (MAGRT), the Mental Health Review Officer (MHRO), the Public Defender's Office (PDEF) and the Victim Witness Program (VICTW). Included also is the Juvenile Probation Office (JPO), the Drug and Alcohol Office (DnA) and a residential treatment home (ARTH).

The mental health sector is dominated by the Mental HealthlMental Retardation Office (MHlMR) and the Community Mental Health Center (CMHC). The CMHC is the largest mental health agency, with a large array of programs. The MHlMR Office does not provide direct services but is the largest conduit of mental health funds into the local network of mental health organizations. There a second (and geographically peripheral on the county boundary) community mental health center (DHSI) and a psychiatric hospital (HPSYC). The Base Service Unit (BSU) provides intake and case management services for the MHIMR. Two units providing community living arrangements for the developmentally disabled (CLRI and CLR2) are present. CMHC runs two partial hospitalization programs: STEP is for 8-13 year olds and PHASE is for teenagers. SRRMI provides social and residential rehabilitation for mentally ill patients. THOME provides domestic counseling, foster placement and adolescent pregnancy foster placement services. The Fall County Association for the Developmentally Disabled (CDC) provides a variety of services and the Child Development Care Center (ECDC) provides daycare and preschool services.3 Finally, the Child and Adolescent Social Service Program (CASSP) was


established to create coordination and referral services for agencies in different sectors.

Fourteen units are included from the poverty/social welfare sector. The dominant agencies are the County Assistance Office (CAO) and the Community Action Agency (COACT). Catholic Charities (CA TH) and Community Ministries (CMIN) are religion-based service agencies for the poor. The Housing Authority (HAUTH) provides the poor with information and resources to obtain subsidized housing they otherwise could not afford and the Food Bank (FOODB) provides food for the poor. The Red Cross (RC) and the Salvation Army (SARM) provide resources for people in poverty. The Women, Infants and Child (WIC) program provides nutrition resources and screening to identify people who are developmentally disabled. Social Security (SS) runs financial support programs and provides Supplemental Security Income. Head Start (HEADS) provides educational opportunities for the poor. Both the Private Industry Council (PIC) and the Jobs Training Partnership Act (JTPA) provide job training, remedial counseling and drop-out prevention programs. A transportation service unit (TRANS) provides transportation for the poor.

6 CYS

5 CMHC DnA

4 MHMR

lUI

3 CAO CAOCT WRC

2

o~ ______ ~====~ ________ _ Figure 1. Distribution of overall organizational standing

6

CYS 5

4 -r-CMHC MHMR

3 lUI CAO COACT

2

o bJ = e h m p Figure 2. Side-by-side boxplot of standing by sector


Table 1. Measures of standing for all agencies in network

Rank Agency Sector Standing Rank Agency Sector Standing

1 CYC j 5.010 34 UWAY f 0.501

2 CMHC m 4.040 35 HAUTH P 0.489

3 DnA j 3.896 36 CMIN p 0.477

4 MHMR m 3.651 37 PIC P 0.475

5 lUI e 2.771 38 PDEF j 0.462

6 CAO p 2.510 39 VICTW j 0.455

7 COACT p 2.376 40 CLRI m 0.453

8 WRC j 2.343 41 BLIND h 0.424

9 JPO j 2.214 42 JUDGE j 0.420

10 COURT j 2.189 43 CASSP m 0.395

11 FAC j 2.059 44 FP h 0.381

12 SPOLC j 1.739 45 ARTH j 0.380

13 CDC m 1.253 46 PHASE m 0.339

14 HPSYC m 1.253 47 DREL j 0.306

15 DA j 1.190 48 RC P 0.272

16 MAGRT j 1.142 49 THOME m 0.268

17 UHOSP h 1.078 50 WIC P 0.250

18 CNSCH e 0 .936 51 CATH P 0.242

19 CCOM f 0.832 52 BSU m 0.227

20 PH h 0.825 53 JSERV P 0.227

21 GALL h 0.780 54 MPOLC j 0.213

22 BVILL h 0.776 55 FOODB P 0.200

23 BSCH e 0 .762 56 UPOLC j 0.191

24 LAID j 0.734 57 ECOC m 0.182

25 YMCA 0 0.724 58 CLR2 m 0.166

26 USCH e 0.717 59 DHSI m 0.160

27 ES h 0.716 60 GPOLC j 0.152

28 AGSCH e 0.670 61 MHRO j 0.139

29 SRRMI m 0.650 62 CPOLC j 0.130

30 HEADS P 0.644 63 TRANS P 0.117

31 SARM P 0.569 64 STEP m 0.116

32 SS P 0.540 65 CHCTR 0 0 .047

33 LHSCH e 0 .519

MEASURES OF CORPORATE SOCIAL CAPITAL

As argued above, the measure of standing can be used to operationalize corporate social capital. These measures are given in Table 1 where the organizations are ranked in terms of their stocks of social capital. These measures are displayed in Figure 1 in the form of a boxplot. Figure 2 contains a side-by-side boxplot of these measures by sector.

Organizational Standing as Corporate Social Capital- 141

Table 2. Contributions to standing in the education sector

Percentage Contribution by Sector

Agency Education Health Judicial Mental Health Poverty

lUI 17.2 11.7 8.4 37.0 15.9

AGSCH 2.5 17.0 7.5 Sl.8 19.2

BSCH 3.4 16.0 12.8 48.2 16.7

CONSH 3.0 13.8 32.5 25.9 15.9

LHSCH 12.6 4.0 17.1 38.5 24.1

USCH 4.7 3.3 40.0 28.0 20.3

There are eight agencies whose social capital makes them stand out as high outliers in Figure 1. The overall dominant agency is CYS, a large multi-purpose agency, which, despite the high variability of social capital among judicial sector agencies, remains a high outlier in that sector. The Drug and Alcohol Office (DnA), another agency from the judicial sector, is prominent also in Figure 1. The Rape Crisis Center (WRC) is the third prominent agency from the judicial sector, one that was particularly active in Fall County with a high profile director who sat on many boards. Neither DnA nor WRC stands out when attention is confined social capital measures of agencies in the judicial sector.

Both the CMHC and the MHlMR are prominent overall-consistent with their domination of the mental health sector. Similarly, the County Assistance Office (CAO) dominates the poverty-social welfare sector (with one of the largest budgets state wide due to the high incidence of poverty in Fall County). The Community Action Agency (COACT) is large also with many programs. The Intermediate Unit (lUI) is prominent in the education sector. However, as argued below, its prominence has less to do with dominating the education sector and more to do with its linking role between sectors. There are no such prominent agencies from the health sector. The two primary funding agencies (CCOM and UW A Y) are unremarkable with regard to social capital generated through the provision of services.4 Having CCOM well above UW AY reflects it being the larger funding agency. From the view of service provision, it is surprising that the Fall County Health Center is mentioned as it is the facility within which many agencies are located. Having it rank last seems veridical as it provides no direct services.

Disaggregation of Corporate Social Capital by Sectors The social capital measures are more interesting when they are disaggregated into contributions coming from sectors. These are shown in Tables 2 through 6 and were generated by the use of equation (4). To aid the interpretation of these contributions the following labels are used. A contribution (from a sector) is labeled primary when it contributes 50% or more of the total social capital of an organization. A contribution is secondary if it is between 25% and 50% and tertiary if it is above 10% and below 25%. The primary contributions to social capital are bolded and the secondary contributions are emphasized in the following tables. Attention is confined to contributions from the education, health, judicial, mental health and poverty/social welfare sectors.5


Table 3. Contributions to standing in the health sector

Percentage Contribution by Sector Agency Education Health Judicial Mental Health Poverty

BLIND 24.3 5.5 5.5 37.7 21.9

BVILL 8.9 24.8 22.1 26.8 12.4

ES 10.2 8.0 6.8 32.7 36.0

FP 1.8 21.9 8.1 42.4 19.4

GALL 2.5 17.2 7.0 34.2 22.9

PH 10.5 29.5 5.9 13.4 24.4

UHOSP 4.3 12.7 18.5 35.8 24.7

The mental health sector, as a secondary contributor (37%), is the largest contributor to the social capital of the lUI . The education, health and poverty sectors are tertiary contributors. See Table 2. Mental health agencies were seen as useful by the Intermediate Unit for providing their services. When the school districts are considered, the mental health sector is a primary contributor (52%) to one district (AGSCH) and a secondary contributor to all of the rest with contributions ranging from 26% to 48%. This makes a lot of sense. Given the nature of this ION, the schools are linked to other agencies through the treatment of 'unusual' students (highly gifted, developmentally challenged and disruptive).6 The gifted and developmentally disabled children are referred often to the Intermediate Unit7 or to mental health agencies . Hence the high contribution from the mental health sector as agencies in that sector saw lUI as effective. Often, these children come from poor households and poverty agencies are mobilized. The poverty sector is a tertiary contributor to the social capital of all agencies in the education sector. While there are referrals to judicial sector agencies, not all of these are seen as effective. However, the judicial sector is a secondary contributor to the two largest schools (USeH at 40% and CONSH at 32%). The location of the largest schools is the location also of the largest police departments. The judicial sector is only a tertiary contributor to another two schools. As judicial agencies process children and youth as either victims or perpetrators they are seen as more useful when the disruptive students are removed from the school system. That the education system does not contribute much to the social capital of schools is not surprising. Their primary interactions are with the lUI and agencies in other sectors and not with each other.

As shown in Table 3, no sector is a primary contributor to the social capital of health sector agencies. The mental health sector is a secondary contributor for all but one of the health agencies with contributions ranging from 27% (to BVILL) to 42% (FP). It is a tertiary contributor to the remaining health unit (PH). The poverty sector is a secondary contributor to one health unit ES at 36% reflecting the perception of ES as an agency whose' members believed in charitable sevice. In this network, ES worked with poverty ageffcies and was seen as having value in the provision of subsidized service. This sector is a tertiary contributor for all of the remaining health sector units. The health sector is a secondary contributor to the social capital of PH and a tertiary contributor to another four units in its sector. The education sector is a high tertiary contributor (25%) to BLIND which is consistent with the low


Table 4. Contributions to standing in the judicial sector


ARTH 6.3 3.1 46.2 11.8 29.4

COURT 6.3 1.3 69.6 10.0 11.0

CPOLC 1.7 0.8 85.2 3.2 8.1

CYS 6.5 8.4 34.5 22.9 23.1

DnA 11.4 6.4 26.5 22.6 24.8

DREL 1.6 3.8 58.4 6.1 26.4

DA 2.4 2.0 82.8 5.8 5.6

FAC 2.0 2.6 55.6 11.4 20.4

GPOLC 0.0 0.0 100.0 0.0 0.0

JPO 14.5 2.6 52.6 16.7 11.4

JUDGE 5.8 3.5 57.4 17.2 13.6

LAID 3.2 6.5 53.9 5.8 28.4 MAGRT 15.9 1.8 62.0 10.1 8.4

MHRO 4.3 0.9 79.2 68 7.6

MPOLC 0.0 0.0 100.0 0.0 0.0

PDEF 4.0 1.0 84.0 5.1 4.9

SPOLC 12.2 1.7 67.1 11.8 5.8

UPOLC 7.7 3.6 63.2 12.4 10.3

Vlcrw 4.4 3.4 70.2 10.2 9.6

WRC 2.4 12.0 32.8 20.8 23.4

contribution of the education sector to the social capital of this agency. The relations of education sector agencies to BLIND are educational services for the blind and relations to the health sector are less consequential for BLIND. With the exception of the hospitals (BVILL and UHOSP), the judicial sector does not contribute much to the social capital of agencies in the health sector-a veridical result.

The primary feature of the social capital of agencies in the judicial sector is that they are generated primarily in that sector. There are sixteen judicial sector units for which the judicial sector is a primary contributor. See Table 4. These contributions range from 53% to 100% (with the latter figure holding for two local police departments). Many of these units can be viewed as being in the court system of the sector: Public Defender Office (84%), District Attorney (83%), Legal Aid (53%), Mental Health Review Officer (79%), the Court (70%), Magistrates (62%), Domestic Relations (58%), Judges (57%), Legal Aid (56%) and the Juvenile Probation Office (53%) all have most of their social capital generated within the judicial sector. For the remaining four units of the sector, it is a secondary contributor to their social capital with the contributions ranging from 26% to 46%. This is an inward looking subsystem within the overall social service system. When children and youth enter as perpetrators they tend to remain and judicial agencies are useful for each other within the subsystem boundaries with most judicial agencies lacking strong ties elsewhere.


Table 5. Contributions to standing in the mental health sector


BSU 0.5 1.4 2.7 85.2 5.5 CASSP 16.1 4.8 12.1 54.7 8.6

CDC 7.6 10.6 3.9 49.4 24.2

CLRI 1.7 4.4 8.6 52.5 17.5 CLR2 2.5 6.4 12.4 17.1 29.6 CMHC 8.8 6.8 15.5 30.1 32.9 DHSI 6.6 4.9 55.5 19.0 9.6

ECOC 0.5 \.3 12.0 2.5 92.5 HPSYC 5.9 16.0 37.2 24.7 11.4

MHMR 5.2 12.9 31.8 18.6 20.5

PHASE 0.9 0.7 34.2 57.2 4.1

SRRMI 2.0 3.1 12.7 56.4 6.2

STEP O. 0.3 0.5 97.2 1.0

THOME 0.9 1.5 7.7 13.6 9.6

The poverty sector is a secondary contributor to the social capital of four judicial agencies-ARTH (29%), DnA (25%), with CYS and the Rape Crisis Center (WRC) at 23%. These are exactly the four agencies for which the judicial is 'only' a secondary contributor to their social capital with contributions at 46% (ARTH), 34% (CYS ), 33% (WRC) and 26% (DnA). All four of these agencies have significant ties outside the judicial sector. The mental health sector is a tertiary contributor for CYS, WRC and DnA. These are exactly the three high social capital agencies in the judicial sector and while the judicial system overall is an inward looking system, these three agencies have social capital generated from three sectors. The poverty sector is a secondary source of social capital for Legal Aid (28%) and Domestic Relations (26%) both units that provide legal services for the poor. The mental health sector is a tertiary contributor for nine judicial agencies. With the education sector being a tertiary contributor to only four judicial agencies and the health sector being a tertiary contributor to only one unit in the judicial sector, it is clear that these sectors are not a source of social capital for units in the judicial sector.

Just as there are agencies in the judicial for which social capital is generated largely within their sector, there are corresponding (but fewer) agencies in the mental health sector as shown in Table 5. They are STEP (97%), the BSU (85%), PHASE (57%), SRRMI (54%), CASSP (55%) and CLRI (52%). The first three are internal to the CMHC. Both SRRMI (mental illness) and CLRI (developmentally disabled) provide residential treatment and are internal to the mental health sector. It is not surprising, therefore, that most of their social capital is generated within the sector. CASSP was a unit set up (with funds through the mental health sector) to promote coordination of services between sectors. Having most of its social capital generated within the mental health sector, it is not surprising that it had little success in establishing between sector coordination. The mental health sector is a secondary


Table 6. Contributions to standing in the poverty sector

Percentage Contribution by Sector Agency Education Health Judicial Mental Health Poverty CAO 2.3 9.5 26.0 14.1 36.4

CATH 0.5 0.9 7.5 6.9 38.4

CMIN 1.5 2.6 29.8 11.4 37.4

COACT 4.3 14.0 15.6 12.2 39.2

FOODB 1.8 6.7 7.0 6.0 72.0

HAUTH 3.0 20.3 7.2 5.4 60.8

HEADS 2.0 30.4 1.7 19.6 25.8

JSERV 1.6 5.2 13.8 11.0 62.5

PIC 2.8 4.9 10.4 15.8 61.0

RC 2.1 26.0 3.2 4.6 59.4

SARM 1.7 10.8 9.5 10.7 57.8

SS 2.3 6.4 17.9 39.8 28.7

TRANS 3.7 33.3 7.6 7.6 38.7

WIC 0.9 47.6 5.1 8.1 34.5

source of social capital for the CMHC (30%) and CDC (49%). It is a tertiary source of social capital for another five agencies. This includes MHMR for which only 19% of its capital is generated within the sector. However, the judicial sector is a secondary source of 32% of its social capital and the poverty sector is a source of another 33%. As with CYS in the judicial sector, MHMR has a significant amount of its social capital generated in multiple sectors. The CMHC has 33% of its social capital generated in the poverty sector. The only mental health agency for which social capital is not generated within the mental health sector is ECDC. Virtually all of its social capital (92%) was generated by its ties to the poverty sector. As noted above, a strong case can be made for viewing this agency as part of the poverty sector.

Six of the agencies in the poverty sector have most of their social capital generated in that sector. See Table 6. The Food Bank at (72%) heads this list followed, in order, by JSERV (62%), HAUTH (61%), PIC (61%), RC (59%) and SARM (58%). The poverty sector is a secondary contributor to the social capital of the remaining agencies in the sector. This suggests, rightly, that this is another inward looking sector. However, only the education sector is not a contributor to the social capital of poverty sector agencies. The health sector is a secondary contributor to the social capital of four poverty agencies-WIC (48%), TRANS (33%), HEADS (30%) and RC (26%) underscoring the interdependence of poverty and health problems. This is particularly the case for the nutritional program (WIC) for women, infants and children. TRANS provides transportation services for health provision services for the poor. The health sector is a tertiary contributor for another three poverty agencies. Mental health agencies are a secondary contributor for the social capital of two agencies and a tertiary contributor to another seven, with contributions ranging from 40% (SS) to 11%. Finally, the judicial sector is a


secondary contributor to two poverty agencies (CMIN with 30% and CAO at 26%) and a tertiary contributor for another four.

DISCUSSION

With organizations occupying niches located in inter-organizational networks, they generate ties in order to coordinate their actions and deliver social services. The services are located in different sectors and are directed at a variety of client pools. As organizations have scarce resources, issues of how to allocate them arise. While it is clear that some agencies cannot and do not work together, the range of collaborative efforts is quite remarkable. Over time, collaborative (and, on occasion, adversarial) relations are explored.8 Organizations come to acquire reputations. (Indeed, we came to have a working hypothesis that agencies giving us trouble also gave trouble to other agencies in these networks. It was confirmed.) The evaluations they form of each other can be used to provide a measure of network generated corporate social capital.

While this measure is useful and a veridical image of standing as social capital, there are limitations to the materials presented here. One is the use of a unit vector e for the overall measure of standing and E for the disaggregated measure. With mUltiple waves for these networks, it is possible to use standing, s(t-l) at a prior year as e for the estimation of social capital s(t) in the following year. In a similar fashion, E(t-l) can be used as E to generate E(t). While it is not surprising that the dominant agencies of sectors have high social capital, a next step will be to seek predictors of high social capital other than size.9 Clearly, funding from extra-local sources and the nature of the working relations will be relevant. A secondary gain from using these measures of social capital is measurement of the relative contributions to social capital from different sectors. The overall dominant agencies have their social capital generated from multiple sectors. Examining the relative contributions by sector permits images of where cracks form in the overall system. Of particular importance is the idea that the judicial and poverty sectors are the most inward looking sectors. It is clear that collaboration must involve agencies from multiple sectors and, given the nature of these counties, even if the focus is on health and mental health issues, both judicial and poverty agencies will be involved. This will not be easy and a reasonable speCUlation is that these service delivery systems will work best when the social capital of enough key agencies is generated from multiple sectors. Exploring the generation of social capital in multiple sector networks remains another important next task.

This work was supported by NIMH Research Award #ROI-MH44-1948. Comments by the editors and Katherine L. Woodard on an earlier draft are much appreciated.

NOTES

I. They argue that another two factors-organizational structure and network location-operate. In their framework, all three factors can be used to 'explain' the variation reputational standing. As my focus here is measuring standing, as social capital, such explanations are not the primary focus. Also, if I am using the network structure, within which organizations are located, to define standing, network location cannot be a non-redundant predictor of standing. Of course, this is my problem and not theirs.

Organizational Standing as Corporate Social Capital- 147

2. Viewing the structure of an economy as a social network, this lineage can be traced back further to Leontief (1951) 3. As this agency is primarily funded by the County Assistance Office, a strong case can be made for including it in the poverty sector. This receives support when the contributions to social capital by sector is examined. 4. Again, there is an emphasis on the provision of services which tends to make those agencies providing resources to fade into the background 5. While the funding agencies (CCOM and UW A Y) and the other agencies (CHCfR and YMCA) were used in the estimation of social capital, they are not used in these comparisons. Thus the row sum for lUI in Table 2 is 90.2% leaving slightly less than 10% coming from the 'non-sectoral' agencies. 6. These categories are not all mutually exclusive 7. This accounts for the 17% as the highest contribution to the social capital from the education sector to an agency in that sector. 8. Han and Breiger's (this volume) empirical example considers ties that are always both cooperative and adversarial. 9. Podolny and Castellucci (this volume) propose a status measure that seems similar to the one proposed here. Also, Han and Breiger (this volume) are persuasive in their decomposition of social capital into (related) components.

•

Customer Service Dyads: Diagnosing Empirical Buyer-Seller Interactions along Gaming Profiles in a Dyadic Parametric Space

ABSTRACT

8 Dawn Iacobucci

This chapter proposes a methodological approach to studying buyer-seller interactions to understand corporate social capital in consumer services. The method combines the theoretically strong game theory tradition with newer dyadic modeling that should provide fruitful means of characterizing buyer-seller relations.

INTRODUCTION

Managers and corporate leaders acknowledge that their satisfied customers are an essential asset to their firms. Those customers can be other firms, in the roles of supplier, distributor, partner, etc., or the customers can be the end-users, the consumers. While many of the chapters in this book (in sections 3 and 4) focus on the corporate social capital that resides in inter-firm relations, this particular chapter focuses on the social capital inherent in the relationship of a firm with its consumers.

Within the class of purchases consumers can make, this chapter further focuses on services industries (e.g., hotels, airlines, health maintenance, legal advice) rather than manufactured goods (e.g., shampoo, toothpaste, blue jeans, etc.). One of the primary distinctions between services and goods is that the former is comprised of a greater interpersonal interaction between the buyer and seller (Iacobucci 1998): for services, the customer engages in a transaction with the service provider, a frontline representative of the service firm. By comparison, for goods, a customer selects the desired packages from shelves, and deals rather minimally with a retail representative, and not at all with a manufacturer representative. Thus, services would seem to offer a better environment (than goods) in which to study corporate

Customer Service Dyads - 149

social capital at the consumer level (e.g., also see the chapters by Ferlie and Doreian, this volume, who study networks of relationships in the health services setting).

Indeed, Reichheld and Sasser (1990) describe the importance of consumer loyalty, drawing the analogy between the manufacturer's goal of 'zero defects' and a service provider's goal of 'zero defections.' Heskett et al. (1994) take the implications of loyalty further, demonstrating the long-term profitability of satisfied customers; their life-time customer value. The long-term phenomenon of relationship management (Baker and Faulkner 1991) is enhanced by positive employee-customer relations. Strong positive interactions add value and contribute to customer satisfaction (Crosby and Stephens 1987; Koelemeijer 1995; Price, Arnould, and Tierney 1995) and future sales opportunities (Crosby, Evans, and Cowles 1990). Thus it is clear that the consumer quite directly provides corporate social capital, and that the process by which this occurs is at least in part social or interpersonal, involving the interactions between the consumer and the provider.

Schlesinger and Heskett (1991) emphasize the interdependent nature of a firm's employees and its customers. They illustrate the importance of a firm keeping its employees happy and competent (i.e., well-paid and well-trained) so that they may in turn keep the customers happy. For example, Marriott is known for relatively satisfied employees (high retention rates) and customers (high room occupancies), and the argument is that each drives the other. Given that customer and employee satisfaction are positively correlated, the converse is also true; i.e., customer dissatisfaction and poor customer service can also be attributed in part to dysfunctional customer-employee interactions. The service provider's inability to anticipate customer needs or recover from service failure (i.e., resolve the customer complaint to the customer's satisfaction) is frequently seen as the source of customer disgruntlement (Bitner et al. 1990; Brown and Swartz 1989; Solomon et al. 1985; Surprenant and Solomon 1987). There are, of course, also other classes of explanation for poor customer service, e.g., systemic industry problems, poor organizational cultures, etc., but the dyadic exchange is of primary importance to the present investigation.

In addition, we know that services purchases are typically characterized as being more intangible, and accordingly the quality of the purchase is more difficult for the customer to assess. Hence, customers rely upon cues to quality, including the interpersonal nature of the service provider. Customers place credence in those providers whom they trust, so the relational aspects that exist in the customerprovider dyad are critical. Positive relationships, conveying trust, thus bring the service provider the corporate social capital of drawing and keeping customers.

Researchers are coming to recognize that repeat purchases do not necessarily imply loyalty. However, if the service organization provides quality and value and satisfaction, perhaps with additional perks such as customization and friendly frontline staffs, customers will be more inclined to return to these providers, given that there are always some search and start-up costs in switching to alternatives. These provisions, e.g., quality and value, are not monumental requests on the part of the consumer-which is not to say that many service organizations provide such. And certainly there are customers who are over-demanding, constituting social liability, and firms are also coming to recognize that they do not necessarily wish to support


the efforts of building relationships with all consumers, rather presumably those segments projected to be more profitable.

We might also offer the caution that while firm-to-firm and consumer-toconsumer relations are those between matched entities, there is a peculiar asymmetry inherent to a firm-consumer relation (Iacobucci and Ostrom 1996), and such imbalances in power tend not to favor (Hibbard and Iacobucci 1997) long-term relational development. A dissatisfied customer may vow to never return to a provider, but there are masses ready to take his or her place, thus usually voting with one's purchases going elsewhere rarely feels like a satisfactory solution. On the other hand, many firms are now talking about the relationships they wish to develop with customers, e.g., through interactive marketing, data-base and direct marketing, etc., yet inquiries to customers suggest that their frequent user accounts do not make relationships. Clearly the relationship with a customer is precious, and firms that manage customer interactions can enhance these ties, strengthening the social bonding capitals with their target segments.

In sum, provider-customer relationships do not necessarily yield social capital. Overdemanding customers create social liability, and service providers are better off without them. In addition, providers can not take the maintenance of profitable customer-relationships for granted and are starting to actively intensify these ties.

If we have established thus far that consumers contribute to corporate social capital, e.g., in the form of their relational purchasing behavior, and that services sectors may be an ideal business environment in which to study buyer-seller interactions, we might also query as to strategies for studying these transactions. As in studies of inter-firm relations, most researchers survey one side of the dyadic relationship, primarily for ease of data collection. Critics of one-sided perspectives would say the approach is suboptimal for studying the entire relational phenomenon, and that dyadic methods are inherently superior: both points of view in a dyad must be integrated to understand the individual actors in the relationship, as well as the gestalt effect of the relationship itself. For example, Swartz and Brown (1989) studied the interaction between medical professionals and their clients as dyadic: they obtained both parties' perspectives on perceived quality and service encounter expectations. Researchers studying household purchase decisions must also integrate sometimes conflicting dyadic perspectives (e.g., Corfman and Lehmann 1987; Davis, Hoch and Ragsdale 1986). Perhaps not surprisingly, Menon et al. (1995) found that accuracy of proxy-reports improved with greater partner communication.

This chapter will provide a dyadic analysis of buyer-seller interaction. Two methods will be used. First, we will draw from elementary game theory, a highly evolved theoretical framework that is fundamentally dyadic in nature. Second, we will explore the utility of dyadic interaction models that have been created within the social network paradigm to model data that would arise in a customer-provider engagement. We begin with a brief description of game theory. This introduction may be skipped for the reader familiar with the area.

ELEMENTAL GAME THEORY

We should begin by acknowledging that the area of game theory comprises a huge


Table 1. Diagram of outcomes for the prisoners' dilemma game

Suspect B:

Suspect A Do Not Confess C Confess 0

Do Not Confess C 3,3

4,1

Confess 0

1,4

2,2 Note: The first number in each pair corresponds to the outcome for suspect A, the second number corresponds to the outcome for suspect B. Higher numbers represent better outcomes.

literature, and developments are far more complex than we need for the current research purposes. Many good references exist for additional information. Recent, lucid introductions include Zagare (1984) and Van Lange et al. (1992).

This class of modeling refers to 'games' because two or more parties engage in interactions, each player has goals to achieve, and they play by given rulestogether these properties characterize gaming. The simplest games require only two players. Certainly researchers have extended games to more than two players, as in studies of social dilemmas and coalition formation, but we can focus on the twoparty game, given that it most closely resembles the simple customer-employee dyad. Even with only two players, Rapoport and Guyer (1966)' presented a taxonomy of 78 classes of games, the most familiar of which is the 'prisoners' dilemma game.'

The intuitive motivations of the players in that game follow (Dawes 1980: 182):

two prisoners have jointly committed a felony and have been apprehended by a District Attorney who cannot prove their guilt. The D.A. holds them incommunicado and offers each the chance to confess. If one confesses and the other doesn't, the one who confesses will go free while the other will receive a maximum sentence. If both confess they will both receive a moderate sentence, while if neither confesses both will receive a minimum sentence.

These outcomes are diagrammed in Table 1. One suspect's choices (i.e., to confess or not) form the rows of a 2x2 matrix, the other's comprise the columns. The Table contains the valuation of the outcomes for each party. For example, note that if A confesses, and B does not, A receives the best outcome (a '4') and B the worst a ('I'); and vice versa if B confesses and A does not. If both confess, the outcomes for both are worse than if both resist confessing.

As this game illustrates, each player chooses how to behave. For simplicity, games may be restricted to decisions between two choices. Depending on the research context, the choices take on different guises, but one choice can generally be referred to as 'cooperative' (commonly labeled 'C') and the other as 'competitive' (labeled 'D' for defect from cooperation). In this game, the prisoners who choose to not confess are said to be cooperating with their partners.

Both parties obtain their outcomes as a function of the joint dyadic choices, thus the players are interdependent because the actions taken by each affects the outcomes of the other. Each player is expected to maximize his or her outcomes, but note that in this game, there is a conflict of interest between maximizing one's own outcomes versus maximizing the joint dyadic outcome; i.e., individualistic versus collectivistic rationality. For both A and B, confessing yields better outcomes for


themselves than not confessing, when considered independently of what the other chooses to do. Thus, confession is said to be a dominant strategy for the individuals involved. Unfortunately, if both players select their dominant strategy, a nonoptimal collective outcome results, because both players could have done better as individuals under a different choice (i.e., had they not confessed). Individual rationality prescribes noncooperation, whereas collective rationality prescribes cooperation (Van Lange et al. 1992).

DYADIC INTERACTION STRATEGIES

In our context, cooperating may be defined as the customer and provider interacting pleasantly and helpfully, whereas defection may represent a break-down in smooth interpersonal transactions. More extensively, cooperation may be the provisions of positive elements in the marketing exchange, e.g., providing good service at reasonable prices perhaps with some customization on the part of the provider, and repeat purchases, positive word-of-mouth, and creative input on the part of the customer. Defection may be more broadly defined as well, e.g., providers who overcharge, act impudent and uncaring, are unresponsive to customer requests, or tempermental customers who take much of the providers' time and effort without repeat purchasing, and so forth.

Given the superiority of the collective outcomes for mutual cooperation, and the implications for social interactions and welfare more generally, much research has been devoted to determining how cooperative behavior might be enhanced (Pruitt and Kimmel 1977). Three strategies seem especially effective. First, the payoff structure might be modified; i.e., the incentive to cooperate can be increased or the incentive to not cooperate can be decreased. Second, if the parties are allowed to communicate, they can promise to cooperate with each other (Misumi 1989). Third, for games with multiple trials, each player can adopt the so-called 'tit-for-tat' retaliation strategy whereby a player chooses 'C' or 'D' at time t to mimic the other's choice at time t-I (Axelrod 1980). Thus, a cooperative choice is answered by subsequent cooperation, and a competitive choice is reciprocated with competition. This responsive strategy is thought to encourage cooperation because neither party is taken advantage of, and neither party makes a first strike; accordingly, each player shapes the other's behavior. The retaliation strategy requires that the dyadic interactions iterate over multiple trials.

Table 2 contains several patterns of dyadic interactions that will illustrate different gaming structures. The leftmost column depicts the 10 trials or time periods over which the pairs of actors interact with each other, learning how best to respond to the dyadic partner to optimize individual or collective outcomes. The next three pairs of columns of numbers depict three patterns of dyadic interactions that may be described as exemplar, in that they may not be likely in real data, but they depict particular dyadic structures clearly, at the extreme. The first of these patterns is the 'mutual cooperation' pattern in which both parties, actor i and partner j, make the cooperative choice, defined in these data as a '0.' In contrast, the second pattern illustrates 'mutual competition' in that both parties choose to continually strike the other competitively, defined by the 'l's. The third dyadic pattern is also a


Table 2. Dyadic interactions in gaming data

trial 3 exemplar strategies ~timal data interactions mutual mutual chump retaliate random rand-comp cooperation competition

ij ij ij ij ij ij

00 1 1 01 00 00 10 2 00 1 1 01 00 00 10 3 00 1 1 01 01 10 10 4 00 1 1 01 1 1 00 01 5 00 1 1 01 10 01 10 6 00 1 1 01 01 00 10 7 00 1 1 01 10 00 10 8 00 1 1 01 01 10 10 9 00 1 1 01 1 1 00 00

10 00 1 1 01 10 00 10

clear, but probabilistically unlikely extreme, that we may tenn, 'chump,' due to the fact that one actor, i continues to choose to cooperate even though the other actor, j, always responds with competition;j always takes advantage of i.

The fourth dyadic pattern is labeled 'optimal,' because it depicts the aforementioned tit-for-tat retaliation strategy. For purposes of illustration, the choices by j were chosen randomly (literally by the flip of a fair coin), and whatever choice j selected at time t, actor i chose at time t+ 1.

The last two data patterns are termed 'data interactions.' The data in the pattern designated 'random' were selected by subsequent coin tosses, resulting in a 50-50 chance for the 0-1 datum for each of the actor i and partner j at each of the ten trials. The data in the pattern designated 'rand-comp' were selected similarly, except that for actor i, there was a 25% chance of obtaining a '0' and a 75% chance of obtaining a '1 '; i.e., it was a random pattern that favored the competitive choice.

The proposal in this chapter is as follows. We will model each of the three exemplar patterns, and the optimal retaliation pattern, and thereby obtain their parametric description via dyadic modeling. The two data interaction patterns will then also be analyzed using the same modeling, and the resulting parameters will be used to classify the data streams as resembling more or less one of the previous four theoretical patterns. While the first four dyadic patterns are indeed exemplars and therefore highly unlikely in real data, we can use them to anchor the parametric space that can be used to characterize dyadic interactions in real data. As the parameters describing the real dyadic interactions tend toward the ideal bounds of the exemplar space, it will become spatially clear what sorts of structures underly the real dyadic interactions. We turn now to the dyadic models.


DYADIC INTERACTION MODELS

The dyadic models that will be useful in this chapter have been described elsewhere in detail, so we are brief here (Iacobucci and Hopkins 1992; Iacobucci and Wasserman 1988; Iacobucci and Wasserman 1987). Typically, the dyadic model:

(1)

is fit via the hierarchical log linear model:

U +Ul(i) +U2(j) +U3(k) +llt(1) +UI2(ij) +UI3(ik) +U24(jI) +U23(jk) +UI4(il) +U34(k1) (2)

to the four-dimensional y-array defined as Yijkl=1 for actor i sending relational ties of strength k (i.e., 0 or I) to partner j and receiving at strength I (also 0 or 1 for these data)? The a, ~, and p parameters are usually of particular interest to network researchers. The first two parameters reflect actor-level behavior called 'expansiveness' and 'popularity,' tendencies for actors to send and receive relational ties at strengths k and I, respectively. The p parameter is dyadic in nature, reflecting mutuality or the extent to which relational ties are reciprocated. The a parameters would represent an actor's tendency to act cooperatively, ~ would represent the likelihood that a party typically elicits cooperation, and p would represent tendencies for mutually cooperative interactions. We say more shortly about these parameters in terms of which ones would expected to be statistically significant in the presence of different gaming strategies.

For the strategies described by the data in Table 2, however, a modification of the dyadic model (1) may be of greater interest. In real data, we would presumably have replicate dyads, or even groups of dyads that differed on theoretically interesting properties (e.g., games occuring with and without communication between parties, say). For the purposes of our illustration, Table 2 contains only a single dyad, for each of six patterns. Furthermore, the data in Table 2 occur over ten trials, so model (1) must be modified to allow for longitudinal structural effects. In particular, we may begin by examining adjacent time points, a time lag of two; i.e., collapse over the ten trials to obtain pairs of data for times t and t+ 1. We may aggregate Table 2 over the individuals comprising the dyad and focus on the relational structure (note that in the very act of aggregating, the model examines relational structure with no regard to the particular identity of the actors, but rather the typical interaction stream, e.g., for a class or segment of customers):

Vkl ,II,k2,12 = :Ei:Ej Y i,j,kl,II,k2,12.

We may now fit the model:

In P{YkI ,II ,k2,12=Ykl,l1,k2,12}= 91c1+911+9k2+912+Pkl,l1+Pk2,12

+'¥kI,12+fu,l1 +$kl,k2+$II,12

(3)

(4)


via the log linear terms:

u +Ul(kl) +U2(11) +U3(k2) +U4(12) +UI2(kl .lI) +U34(k2.12) +UI4(kl.12) +U23(k2.II) +UI3(kI.k2) +U24(1I.12)' (5)

The model (4) includes the main effects for the relational ties, the S's, which essentially reflect the volume of 'l's (vs. 'O's). The p's are the reciprocal effects, as per model (2), but now depicted for time 1 (or t) and time 2 (or 1+1).

The terms new to model (5) are the y's, which express a multivariate exchange: what one actor does at time one (kl) determines with greater than random likelihood what they receive from their partner at time two (12). Similarly, what the other actor does at time one (11) determines what their partner does at time two (k2).

The cj>'s are also new terms. These parameters reflect multiplexity, the constancy of behavior on the part of the actors. The term reflects an actor's autocorrelative behavior; what an actor does at one point in time (e.g., kl) is related to what they do at the following point in time (e.g., k2)'

INTEGRATING THE GAME THEORY STRUCTURES AND PREDICTED DYADIC PARAMETER PATTERNS

Cooperation and Competition. We can now reexamine the six dyadic interactive patterns in Table 2 with an eye toward these model parameters. The data patterns for mutual cooperation and mutual competition appear similar in structure, in that both actors' streams of relations to their partners are constant, either O's in the case of cooperation or 1 's in the case of competition. Accordingly, the parametric structure for these two cases should also appear similar in the reciprocal structures. Both patterns depict mutuality, albeit (1,I),s for competition and (0,0) for cooperation, so we may test the proposition that for the cooperative and competitive interactions, the reciprocity parameters should be statistically significant:

PI: both Pkl .1I and Pk2.12;tO. On the other hand, the pattern for competition clearly depicts mutual strikes (1,1) and that for cooperation is its mirror image (0,0), so the baseline volume parameters should appear in the opposite direction (i.e., positive for competition and negative for cooperation, given the data coding):

P1: Skit SII, Sk2, and SI2 will be negative for cooperation, and positive for competition.

Chump. The chump pattern should exhibit high degrees of autocorrelation because both parties continue to do what they had previously done without having been affected by their partner's past behavior:

P3: cj>kI.k2 and cj>1I.I2 will be statistically significant and positive. In addition, the reciprocity parameters should be statistically significant, but negative in direction to express the asymmetry of these dyadic interactions-all dyads are of the class (0,1):

P4: Pkl.1I and Pk2.12:;t0 and are negative.


y

• Retaliation

• Random • Rand-Ccmp

.Chu np

• Competitive • Cooperative

p

• Chump

• Rand-Comp

Figure 1. Gaming strategies in 3-dimensional parametric space

• Competitive • Cooperative

• Random p

• Retaliation

Retaliation. The retaliation dyadic exchange should not result in statistically significant reciprocity parameters, given thatj's behavior was determined randomly, without consideration for i's behavior:

Ps: Pkl,lI and PU.12 = 0; i.e., not statistically significant. And yet i's behavior at time t is perfectly predicted by knowingj' s behavior at time t+ 1. Thus, we should expect to see statistically significant multivariate exchange parameters:

P ,: {'Yk1.12} = {fu.lI} ;t O.

Random Interactive Behaviors. The final two dyadic interactions were the randomly generated pattern, and that which was a cross between random and competitive action patterns. For the wholly random data, we would expect all parameters to be statistically non-significant, unless the modeling approach were hypersensitive to erratic structures, or unless the particular random stream were odd. For the randomcompetitive data, we would expect to see parameters between the near-zero random parameters and those describing the competitive interactions. Notice too, that the random-competitive data look fairly similar to a mirror image of the data for the 'chump' stream.

P,: All parameters should be near-zero for the random data. Ps: Parameters for the random-competition should moderately resemble those for the competitive data.

Let us examine the results of the dyadic modeling and evaluate these propositions.

RESULTS

Figure 1 contains the results of the dyadic models applied to the data in Table 2, presented in two plots. It represents a p-dimensional space, where p=3 is the number of parameters being used simultaneously to create diagnostic profiles to characterize interaction data. The first plot in Figure 1 represents the estimates Pkl ,I2 and 'YkI ,12,

and in the second, Pld.l2 is plotted against chl .u (1 and 2 represent times t and HI)? The values plotted in Figure 1 represent the parameter estimates, though plots of the fit statistics (i.e., essentially representing statistical significance values) confirm the predictions regarding statistical significance. (Points far from the origin are


statistically significant; e.g., competition and cooperation strategies did not yield statistically significant yor $ values.)

Let us first focus on reciprocity parameter, Pld.lI, comprising the first dimension in these plots. These parameters were predicted to be statistically significant for the cooperative, competitive, and chump patterns (PI and P4), and indeed these are all large. The cooperative and competitive estimates are positive indicating symmetry (O,O's and 1,1 's), and the chump reciprocity is negative indicating the asymmetry of the (O,I),s. The retaliation dyadic interaction was expected to exhibit a statistically insignificant reciprocal pattern (Ps), and indeed, it is near zero.

One prediction was made about the multivariate exchange parameter, the 'YkI,12'S

and one about the mUltiplex autocorrelation, $t1.U' P6 predicted that for the retaliation data, 'Yk1.12 should be statistically significant and positive, and indeed it is. P3 predicted that for the chump data, $11.12 would be statistically significant and positive, and it is.

For the random data patterns, P7 had predicted that all parameters would be near-zero and indeed none of them are statistically significant-the point representing the purely random data maintains a position near the origin in both plots. The prediction for the random-competition data, Pg, stated that parameter estimates should appear somewhere between the random and competitive data patterns, and that they may resemble the opposite of the chump parameter results. The latter characteristic appears to have dominated; evidently the asymmetry of the random-competition dyadic interactions drove the model parameters more than the symmetric-appearing competition element. Finally, the main effect parameters, 9ki>

911 , 9u , and 912 are not included in the three-dimensional plot in Figure 1, but P2 had stated that these effects should be negative for cooperation, and positive for competition, and indeed they were.

In terms of a brief summary, we might offer a profile of the gaming strategies in terms of the main parameters investigated: Pk\,\i> 'Yk1 .12, and $11.12. That is, rather than examining the plots for distances between empirical profiles and the exemplar gaming profiles, we might alternatively find just as diagnostic an exercise by which we compare the parameters' profiles more discretely. For example, cooperation, as per the plots in Figure 1, might be characterized qualitatively along the threedimensional parameter vector as: high, low, low. Competition would be characterized similarly, except that a fourth parameter, the 9's would be required to differentiate these patterns. The remaining profiles are also included in Table 3.

Table 3. Profile of Kaming strategies

Ptl,l1 "(kI,12 ~1.t2 Btl

Cooperation high (+) low low Competition high (+) low low +

Chump high (-) high (-) high (+)

Retaliation low high (+) high (-) Random low low low Rand-Camp high (-) medium(-) low


Table 3 also shows how the data pattern labeled 'random competition' fits none of the exemplar row-profiles perfectly, presumably because it is more errorful, as would be real data. Nevertheless, it is fairly straightforward to identify the closest candidate case as the 'chump' profile for these random competition data.

CONCLUSION

This methodological approach, the combination of gaming strategies and dyadic modeling, could be used somewhat like a discriminant analysis in that any number of dyads may be modeled, and their resulting p-dimensional vector of parameter estimates used to plot the dyad in space. The distance between the empirical dyads and the exemplar gaming strategies may be interpretable like ideal point characterizations, in that dyads lying nearer the retaliation extreme would presumably be predicted to be more optimal in outcomes than dyads lying nearer any of the other points. Dyads lying nearer cooperation than, say, chump or randomness could be predicted to yield more social capital, extended longevity, even greater profitability depending on the research context, etc. The pure data patterns have been simulated and serve as functional anchors to the p-dimensional space, and are therefore useful in describing the presumably more errorful data interactions.

It should certainly be noted that this demonstration is only a beginning; both game theory and dyadic modeling are further developed than what is suggested by the fundamental tools used in this chapter. However, this chapter is an initial combination of a theoretically strong gaming tradition with a newer sophisticated dyadic modeling heritage that should provide fruitful means of studying buyer-seller interactions.

From the models we presented, it turns out that some buyer-seller interaction patterns lead to more intensive and long-lived cooperation than others. With the use of empirical data, we can describe (and perhaps predict) which patterns are likely to bring sufficient social capital to make the interaction worthwhile to both parties.

Finally, we might also suggest that while we presented this approach as ideal to study buyer-seller interactions in a consumer services framework, the approach is purely methodological and could certainly be applied to inter-firm relations as well. The requisite data need only be dyadic and temporal, though surely these qualities in data will be more easily obtained on persons rather than firms.

NOTES

1. Cited in Zagare 1984. 2. If multiple dyads were to contain common buyers or sellers, the dyads would not be wholly independent units, but researchers have demonstrated fair robustness of this class of models to such circumstances (e.g., Frank and Strauss 1986; Strauss and Ikeda 1990). 3. For these simple data, the results for the reciprocity parameters at times 1 and 2 were similar in sign and magnitude, so we use only the first set for these plots. This finding is not a general result, and the plots could be extended to four-dimensions, or any p parameters the researcher wishes to model and examine simultaneously, e.g., including higher order terms as say, 0%1.11.>2.

• SECTION II

Structure at the Individual Level social capital in jobs and careers

•

The Sidekick Effect: Mentoring Relationships and the Development of Social Capital

ABSTRACT

9 Monica Higgins

Nitin Nohria

This chapter examines the benefits and pitfalls of mentoring relationships with respect to a protege's ability to develop social capital, measured here as ties across multinational subsidiary boundaries that might produce access to information and resources. The results indicate that early mentoring relationships are negatively related to a protege's stock of social capital and that later mentoring relationships are positively related to a protege's stock of social capital. We call for a contingency approach to studying how mentoring relationships affect career outcomes and discuss implications for future research.

THE SIDEKICK EFFECT

Mentoring is not a new concept. The word 'mentor' may be traced back to Homer's Odyssey, in which a guardian named Mentor took on the role of adviser and teacher to Odysseus' son Telemachus. Building on this ancient tradition, research on mentoring has flourished of late. According to the International Center for Mentoring in Vancouver, British Columbia, academic research on mentoring includes more than 500 published articles, 225 conference papers, 150 doctoral dissertations, 65 books and 150 mentoring program descriptions (Caruso 1992).

Most of this literature concludes that the effects of mentoring relationships are quite positive. Mentoring has been shown to enhance career development (Kram 1985; Phillips-Jones 1982), career progress (Zey 1984), rates of promotion and total compensation (Whitely, Dougherty, and Dreher 1991), and career satisfaction (Fagenson 1989; Riley and Wrench 1985; Roche 1979). Given these generally positive findings, much of the recent research has focused on two important


questions. First, what are the major factors affecting the amount of mentoring a protege actually receives and/or initiates? And, second, what types of help do these mentoring relationships actually provide, or, put differently, what functions do mentoring relationships serve?

In addressing the first question, scholars have uncovered important debates around individual difference factors such as proteges' gender (Scandura and Ragins 1993), socioeconomic status (Ragins and Scandura 1994), and personality characteristics (Turban and Dougherty 1994). In addressing the second question, scholars have demonstrated that mentors provide career and psychosocial help and other functions, such as protection or role-modeling (Turban and Dougherty 1994; Scandura 1992). These questions and answers have brought us much closer to understanding the antecedents and consequences of mentoring relationships. Yet we still lack a good understanding of how these relationships actually work. As Turban and Dougherty (1994: 699) put it, 'we now need to more closely examine how mentoring influences career success.' For example, how do career mentoring functions such as providing exposure and visibility enable proteges to obtain promotions?

We suggest and adopt two means of better understanding how mentoring relationships have an impact on career progress. First, we suggest that, in order to understand what makes mentoring relationships work, it is also important to consider the conditions under which mentoring relationships might not work. To do this, we consider both early and late mentoring experiences and generate hypotheses regarding both when mentoring should work and when it should not work with respect to individual career variables. Second, we suggest that, in order to develop theory in this area, it is important to examine specific mechanisms through which mentoring might facilitate career progress. To do this, we focus our research on one career factor that has been empirically linked to upward mobility-the development of beneficial social ties (social capital), or ties with others in an organization. PuIling these two approaches together, then, we examine the benefits and the pitfalls of career mentoring with respect to a protege's ability to develop social capital.

Researchers have already cautioned that there may be liabilities associated with having a mentor (also see Brass and Labianca, this volume). Scholars have warned that having a mentor creates the possibility for overdependence (Fagenson 1988; Zey 1984), smothering (Kram 1985), anxiety that can turn to dependence (Clawson 1980), and clinging (Clutterbuck 1985). Still, most of the empirical work on mentoring has focused on positive career outcomes. And, when negative outcomes have been explored, the research has tended to focus on detrimental effects to the mentor-protege dyad itself and not to relationships that extend beyond the mentoring alliance (for an exception, see Fagenson 1994).

The career variable we study here, social capital, measures an individual's potentially beneficial relationships beyond a particular dyad. We define social capital here as the number of ties an individual has across multinational subsidiary boundaries. Research on networking and on multinational organizations has demonstrated that the volume of social relationships is related to upward progress in individuals' jobs, higher paying positions, and career performance (Burt 1992; Lin 1982; Nohria and Ghoshal 1997). The volume of a protege's cross-subsidiary ties

The Sidekick Effect - 163

thus reflects social capital by providing the protege with access to information and resources which are beneficial to his or her individual performance. Although scholars of the mentoring literature have not directly tested the effect of mentoring on this career variable, many have suggested that mentoring relationships may enhance career progress because these relationships enhance the protege's ability to develop ties with others. Specifically, researchers have espoused that career mentoring enables a protege to overcome barriers of access to elite groups, to bypass hierarchies (Fagenson 1994), and to have a 'special form of entry into important social networks' (Dreher and Ash 1990: 540).

In this article, we empirically test the link between mentoring and social capital by testing hypotheses regarding when mentoring should and should not work to increase a protege' s stock of social capital. We do not take the blanket view that mentoring is either generally beneficial or harmful, but rather argue that mentoring may have its own time and place in helping an individual develop social capital.

THEORY AND HYPOTHESES

Early Career Mentoring In general, scholars concur that mentoring relationships provide at least two major types of assistance: career help and psychosocial help. Career help includes sponsorship, coaching, protection, exposure, and challenge, while psychosocial functions include role modeling, counseling, acceptance, confirmation, and friendship. During the 'initiation' stage of a mentoring relationship, usually during the period from six months to a year on the job, mentors are said to perform career helping functions for the protege, such as providing exposure, visibility, and highprofile work assignments (Hill and Kamprath 1991; Krarn 1985). At the beginning of the second (or 'cultivation') stage, a period of two to five years, both the mentor's career development functions and psychosocial functions emerge-with the career development functions emerging first. Thus, at the initial stages, the mentor is said to provide the protege with a maximum level of 'challenging work, coaching, exposure-and-visibility, protection, and/or sponsorship' (Krarn 1985: 53).

What are the short term benefits for proteges receiving early career help from their mentors? Research and practical experience provide us with at least three possible answers. First, career mentoring signals the value of the protege to the organization and, hence, facilitates the newcomer's socialization process. Second, career mentoring provides the protege with quick access to important information and resources by lending the mentor' s social capital to the protege. Third, career mentoring not only brings the protege 'to the racetrack' by providing access, but actually places him or her 'on the starting blocks' by providing concrete opportunities to display valuable skills and talents. We review each of these benefits in turn, paying particular attention to how these factors affect the protege's stock of social capital over the course of his or her career in an organization.

Mentors as Signaling Agents Having a mentor early in a protege's organizational career I can signal to others that the protege is a valuable player in the organization and, hence, can smooth his or her socialization into the organization (Van Maanen and Schein 1979). Because


organizational members can't fully assess the qualities of newcomers, they rely upon information that reflects the protege's prior work experience, such as a resume or recommendation, and they rely upon available signaling devices, such as who hired the individual or who the individual's mentor is. Both actions taken on behalf of the protege by the mentor and the protege's mere association with the mentor can facilitate this socialization process. Specifically, providing exposure and visibility by introducing the protege to high status others in the organization grants the protege 'reflected power,' (Fagenson 1994; Kanter 1977), allowing the protege to reap positive short term rewards in two ways. First, the protege's reputation may get a boost as he or she rides on the mentor's coattails of success (Whitely, Dougherty, and Dreher 1991). Second, people's first impressions of the protege may be enhanced since others may assume that the mentor has, to some degree, selected this person above others to be his or her protege.

However, beyond these immediate rewards, there are also liabilities associated with developing one's reputation in an organization through an early association with a mentor. Kram (1985: 26) describes such liabilities as 'the risk that others will question the extent to which the younger individual can thrive on his or her own without a particular senior's support.' The protege's reputation can become inextricably linked to that of the mentor such that his or her competence, apart from that touted by and associated with the mentor, is called into question. Therefore, while short-term benefits may exist, the protege may, over the long run, be viewed as the mentor's sidekick rather than as a worthy individual contributor with whom to develop an independent relationship. Consequently, we expected that having a mentor early in a protege's organizational career would be negatively associated with the amount of social capital the protege was able to develop later on.

Mentors as Access-Providers Having a mentor who provides career help early can also assist a protege by giving him or her quick access to important information and resources. In effect, the mentor lends the protege his or her social capital (Burt 1997) by introducing the protege to others in the organization. This enables the protege to bypass traditional hierarchies (Fagenson 1994) and barriers to information and enables the protege to tap more directly needed resources. Therefore, not only do career functions such as exposure and visibility grant the protege reflected power in the organization, they grant the protege 'legitimacy' as well (Burt 1997)-Iegitimacy that is needed to get things done or to get others to get things done for the protege.

In the short run, borrowing a mentor's social capital provides the protege with the resources he or she needs to get specific tasks accomplished. It may also increase the protege's returns of human capital (Burt 1997), since it positions the protege in spots in which his or her knowledge can increase. However, there are long term costs associated with borrowing someone else's social capital. Relying upon a mentor for access to social networks may, over time, reduce one's own motivation to develop instrumental relationships with others in the organization. As Fagenson (1994: 56) warns, mentoring functions may 'prompt [proteges] to be less attentive to and concerned about relationships external to the mentor-protege union ... [suggesting] that proteges will have less favorable peer, superior, and department


relationships than nonproteges.' Thus, we again expect that having a mentor early will be negatively associated with a protege's social capital over the longer run.

Mentors as Opportunity-Providers Having an early mentoring relationship can also benefit a protege in the short term by giving a protege opportunities to prove him- or herself to others in the organization. Providing high visibility work assignments is beneficial beyond merely introducing a protege to important others; it gives a protege the chance to demonstrate unique skills and competencies to those who might not yet know him or her. The risk of providing such opportunities early is that, despite the mentor's best intentions, the protege might not be ready for such assignments. Noe (1988a: 475) has implored scholars to develop a 'readiness for mentoring' scale to ensure that proteges who are selected for mentoring programs in organizations actually benefit from their mentoring experiences.

We agree with Noe that advising is not necessarily advisable. There are indeed long term risks to providing a protege with high-exposure work opportunities before he or she has built sufficient individual-level competence and confidence at work. From an outside-in perspective, organizational members might view this early exposure as 'jumping the gun' or providing the protege with too much, too soon. Jumping the gun could result in early career failure or, at a minimum, could result in the perceived risk of failure, such that the protege has neither the capability nor the referent power (French and Raven 1968) needed to develop substantial social capital. From an inside-out perspective, providing career opportunities early, when the protege has not yet developed a sense of self-efficacy or confidence in his or her ability to perform (Bandura 1977) might backfire in building alliances with others. The social support research suggests that people with high self-efficacy are better able to obtain effective support from their established social networks than people with low self-efficacy (Wills 1991). Without such confidence and competence, we expected that, over time, proteges would be unable to maintain the social capital they had borrowed from their mentors and/or that they would be unable to develop substantial social capital on their own: HI: Having a mentor early in a protege's career in an organization will be

negatively associated with his or her long term stock of social capital.2

Later Career Mentoring The aforementioned liabilities associated with receiving early career help from a mentor do not apply to the protege who has spent more than a couple of years in an organization-i.e., once the socialization process is over. By then, he or she should have established some sort of reputation and identity in the organization so that there should be substantially less risk that the mentor would smother the protege and, hence, prevent him or her from developing ties with others. And, with his or her own professional identity established, there should be less risk that others would view the protege as his or her mentor' s sidekick. Rather, we expected that the mentor's role as signaling agent would shift from socializing the protege to confirming the protege's value to the organization. Now, more than before, the mentor's affiliation with the protege should signal to others that the protege is a valuable organizational member-that the protege has actually demonstrated potential. Therefore, the


mentor should accentuate, rather than overshadow, the protege's accomplishments. Hence, we expected later career mentoring to be positively associated with a protege's stock of social capital.

Regarding access to important resources and information, earlier concerns should subside once again with respect to later mentoring relationships. After several years in the organization, the protege is no longer an organizational member seeking legitimacy who needs to borrow another person's social capital in order to gain organizational legitimacy. The protege should have his or her own ties with others in the organization and, hence, should be attentive and motivated to develop ties with several organizational members. Moreover, at this later stage in the protege's career in the organization, the mentor may serve as a linchpin to facilitate the connection of clusters of subnetworks of individuals in the organization. In this way, the mentor's introductions should build upon or enhance the protege's own stock of social capital. Therefore, we again expected a positive relationship between having a mentor later in a protege's organizational career and having a large number of intraorganizational ties.

Finally, we do not believe that later mentoring relationships that provide high visibility work opportunities run the same sort of risk of jeopardizing the protege's ability to develop social capital as is the case for the new protege. At this point, there should be less risk that the protege would fall on his or her face by jumping the gun, since he or she had already joined the race, so to speak. Although the stakes may still be high for individuals later in their organizational careers, we presume that the protege has amassed sufficient self-efficacy to benefit from potential helping situations and to view them as challenges rather than as threats. Therefore, we expected that the opportunities provided by a mentor to a protege later in his or her organizational career would facilitate the development of social capital and, hence, would be positively associated with a protege's stock of social capital: H2: Having a mentor later in a protege's career will be positively associated with

his or her long term stock of social capital.

METHODS

Respondents and Setting Data were collected from 323 managers in the consumer electronics businesses of three major multinational corporations (MNCs), Philips of Holland, Matsushita of Japan, and Nippon Electronics Corporation (NEC) of Japan. Managers were either department heads or general managers of subsidiaries in 24 countries in which all three organizations had national subsidiaries. The subsidiaries in our sample were selected in consultation with corporate managers of the MNCs and constituted, in their view, a representative sample of all their subsidiaries. Questionnaires were mailed to all departmental managers in each of these subsidiaries. The response rate to the survey was 87% in Philips, 93% in Matsushita, and 75% in NEC, representing an overall response rate of 83%. In no subsidiary did we get a response rate of less than 60%.

The population from which our sample was drawn was a well-defined group of upper middle-level managers in all the firms. Respondents had worked for an average of 13.6 years in their current organization and had careers spanning more


than 20 years across different organizations. Although we did not collect data on their ages, we estimate that the median age of the managers in our sample was 40+ years. All the managers were men, reflecting the gender bias of the population surveyed. There were no female senior subsidiary managers in the particular MNCs we surveyed.

Because of the uneven distribution of responses across countries (there were several small countries in which there were only a few subsidiary managers at the level we had selected), we narrowed the sample to include respondents from only the top ten countries (on the basis of the total number of respondents in all three firms) for this analysis. After cases were eliminated because of missing data, the final number of managers included in our analysis was 177. The composition of our final sample was not statistically significantly3 different than the population surveyed (based on t-tests comparing the means of all the variables included in our survey).

Measures Social Capital. An individual's social capital is generally determined by the

breadth or range of his or her social networks as well as the position that the individual occupies in them. People are presumed to have greater social capital if they have greater range (i.e., they have contacts that span the different clusters within the network) and they occupy a more central position or are positioned to take advantage of the structural holes in the network (Burt 1992; Nohria 1992b).

In this study, we operationalized social capital as the range or the number of national subsidiaries of the firm in which the respondent had contacts. Range was calculated as the number of different subsidiaries in which the respondent reported having a contact with whom he communicated, regardless of the frequency.

Given that each national subsidiary represents a relatively separate network, these contacts can be viewed as 'bridging' ties as they allow the individual to tap into otherwise disconnected social clusters (Granovetter 1973). Indeed, Burt (1992) has suggested that the number of bridging ties possessed by an individual may be one of the best measures of social capital. Our focus on ties across subsidiaries was also influenced by previous research on multinationals that shows that such ties are crucial conduits for information and resource flows and have a positive impact on an individual's performance (Nohria and Ghoshal 1997). Research by Brass (1984) and Blau and Alba (1982) provides additional support to the notion that ties across organizational subunits are important to individual career outcomes.

We recognize that-although it is an important measure of an individual's social capital-range (as we have operationalized it here) is, nevertheless, a limited measure of social capital. It does not fully capture the strength or weakness of a person's network position. Ideally, we would have liked to measure the individual's centrality as well as autonomy in various networks. However, that would have required data on the ties among all the actors in the multinationals we studied-a data collection effort that was impractical because of the very large number of individuals who would have to be surveyed.

Mentoring Relationships. To measure whether respondents had early and later mentoring relationships, we asked them if they had developed a close personal


relationship with any senior manager in the company during their first two years of employment with the organization and also if they had such an informal advisor or sponsor at present (again, after working an average of 13.6 years at the company). These questions were based upon Kram's (1985) work on mentoring relationships and are consistent with more recent definitions of mentors as high-ranking, influential members of an organization who are supportive of their proteges' career development (Ragins and McFarlin 1990). These questions allowed us to examine the existence of an informal mentor at two distinct points in time to see whether, and when, such mentors affect an individual's formation of social capital.

Controls We included several control variables in our analysis in order to account for factors, other than the individuals ' mentoring relationships, that might influence their range of intersubsidiary contacts. A brief description and rationale for including these controls follows.

Firm. We included dummy variables to control for the possibility of firmspecific variations across the three firms. One might expect that being a part of a firm that placed greater emphasis on career development might have an impact on the effectiveness of early and/or later mentoring relationships, as well as a protege's stock of social capital. Since we lacked information regarding the relative emphasis placed on career development for these three firms, we had no expectations that one firm would be more important with respect to a protege's ability to develop social capital than another.

Subsidiary. Within any multinational system, some subsidiaries may be more important and central than others in the internal resource-flow network (Bartlett and Ghoshal 1989; Gupta and Govindrajan 1991). In such subsidiaries, managers may have more opportunities to form intersubsidiary contacts. To control for this source of internal variation within any MNC, we included a variable that measured the relative importance of the respondent's subsidiary in the multinational system. This measure was based on ratings of each subsidiary' s relative strategic importance and resource capabilities provided by five headquarters-level respondents in each multinational. The final score was the average rating of the subsidiary on an additive scale comprising both dimensions. The scale ranged from subsidiaries that scored 1 (low importance) to 5 (very important). Given that our final sample was composed of responses from the top ten countries across these multinationals, we ended up with subsidiaries that were all quite important in each multinational, as reflected by the high mean and low variance of the subsidiary measure for our sample of respondents.

Function. Because our respondents were either functional department heads or general managers in their respective subsidiaries, we controlled for their job responsibility. Several dummies were included to control for the possibility that being a general manager as opposed to a functional head of marketing-or manufacturing, research and development, finance, purchasing, legal affairs, or administration--could have an influence on an individual's social capital. We expected general managers to have a broader range of contacts than the heads of functions by virtue of their more integrative organizational role.


Career History. We also controlled for variations in the career histories of our respondents that might influence their social capital. We included variables that measured their tenure in the firm, the number of years they had worked in other organizations, the number of other functions they had performed within the organization, the number of other subsidiaries in which they had worked, the speed of their promotion to their current position, and whether they were expatriates. We expected each of these measures to have a positive influence, if any, on an individual's range of intersubsidiary contacts.

Networking Opportunities. We controlled for differences in social capital that might arise as a result of activities in which the respondent participated that could provide him with unique opportunities to network or make contacts. In keeping with work done by Ghoshal, Korine, and Szulanski (1994), we controlled for the time (days per year) the respondent spent attending meetings and conferences; participating in committees, teams, or task forces; attending training programs; and whether or not the manager had received any formal training upon joining the company. We expected all of these networking opportunities to have a positive influence, if any, on the range of contacts developed by the managers in our study.

Perceived Constraints. Finally, we tried to control for the extent to which the respondents perceived they had autonomy in their organizations and the extent to which they perceived that they were not restricted by formal organizational rules and procedures. Both measures were based on scales of perceived autonomy and informality proposed by Ghoshal and Nohria (1989). We expected that managers who perceived their jobs as more autonomous and/or informal would be more likely to develop more extensive social networks than those who felt more constrained.

Analyses We used multiple linear regression (MLR) initiaIly to explore the effects of the various independent variables on range. The output of this first regression suggested that there are two clear groups of respondents: those with many ties, or high range, and those with some ties, or low range. We defined 'many ties' as having 11 to 17 ties with other subsidiaries, while those 'some ties' as having from 1 to 10 ties with other subsidiaries. No individual responded that he had no ties at all with other subsidiaries. Given the results, it was appropriate to recode the subjects into these two groups and to perform subsequent analyses based on this new dichotomous dependent variable.3

Following correlational analyses, two different levels of analyses were conducted. First, using logistic regression, we tested the initial hypotheses: having a mentor early hurts one's development of social capital, while having a mentor later helps a manager develop a broad range of intersubsidiary contacts. Second, using analyses of variance, we examined more closely the differences among individuals falling into one of four categories: 1) those who had a mentor neither early nor late; 2) those who had a mentor late but not early; 3) those who had a mentor both early and late; and 4) those who had a mentor early but not late.


Table 1. Means, standard deviations, and correlationsa

Variables Mean s.d. 2 3 4 5 6 7 8

1. Mentor early .46 .50

2. Mentor late .65 .48 .18*

3. Employee of .60 .49 -.03 .20** Matsushita

4. Employee of Philips .17 .38 -.02 -.14 -.55 **

5. Relative importance of 3.71 .76 -.13 .06 -.09 .10 subsidiary

6. General manager .14 .34 .17* -.12 -.21 ** .22** .05

7. Marketing function .22 .42 -.13 -.01 -.04 .01 .06 -.21 **

8. Manufacturing function .19 .39 -.00 .05 -.16* -.10 -.19* -.19* -.25**

9. Purchasing function .07 .25 .02 .01 -.01 -.00 -.03 -.11 -.14 -.13

10. Research & .06 .23 -.03 -.03 .10 -.11 -.07 -.10 -.13 -.12 development function

II. Finance function .15 .36 -.03 .07 .01 .03 .05 -.16* -.22** -.20**

12. LegaVother .10 .30 .09 -.00 .03 -.10 .09 -.13 -.17* -.16* administrative functions

13. Tenure at current finn 13.58 7.69 -.09 .09 .18* .21** -.04 -.03 -.07 .18*

14. Years at other 7.10 8.58 .06 -.17* -.17* -.02 .12 .08 .03 -.12 employers

15. Number of other 2.20 1.98 .18* .01 -.06 .05 .02 .22** -.05 .08

functions worked in

16. Number of other .98 1.56 .02 -.00 -.03 .13 .05 .03 .03 .13 subsidiaries worked in

17. Speed of promotion .23 .12 .11 -.11 -.11 -.11 -.11 .12 -.05 -.12

18. Expatriate status .41 .49 -.24** .23** .12 .05 .06 -.10 -.03 .13

19. Time spent in cross- 13.69 15.04 .09 .04 .03 -.02 .25** .13 -.22** -.10

subsidiary meetings

20. Time spent in cross- 7.99 12.01 .07 .12 .00 -.15* .17* .07 .11 -.01

subsidiary teams

21. Time spent in cross- 4.39 10.22 .15* -.04 -.04 .22** -.06 -.01 .03 -.01

subsidiary training

22. Received initial training .47 .50 -.12 .32** .41** -.22** .01 -.11 -.12 .16*

23. Perception of 3.58 .74 -.09 .14 .21 ** -.10 .07 -.18* -.05 .04

organizational

24. Perception of 2.29 .57 -.19** -.13 -.22** -.17* .09 -.04 .12 -.12

organizational informality

25. Social capitalb 9.95 6.91 -.23** .15* -.14 .05 .18* -.07 .14 -.17*

*p<.05, ** p<.OI, aN = 177, Social capital is defined as the range or the number of national subsidiaries of the firm in which therespondent has contacts.


9 lD 11 12 13 14 15 16 17 18 19 20 21 22 23 24

-.07

-.11 -.lD

-.09 -.08 -.14

-.04 -.02 .07 -.14

-.01 -.11 -.09 .12 -.55**

-.04 -.05 -.24** .08 .18* -.00

-.05 -.06 -.03 -.06 .18* -.08 .49**

.18* .06 -.05 .04 -.42** -.34** -.16* -.07

.04 -.06 .11 -.20** .46** -.46** -.07 .17* -.11

-.lD -.01 -.08 -.12 .09 -.11 .08 -.04 .02 .15

-.02 .02 -.11 -.01 .00 -.02 .06 .02 .00 .06 .21 **

.12 .14 -.09 -.03 .04 -.07 .02 -.01 .08-.15* .05 .08

.06 .01 -.04 -.08 .33 ** -.36 ** .02 .13 -.lD .51 * .07 .10 -.01

.06 .13 .lD -.01 .24** -.20" -.17* -.14 -.12 .20** .lD .03 .17* .25**

.05 .02 -.lD .11 -.16* .lD -.02 -.01 .07 -.02 -lD .03 -.04 -.06 .03

.04 -.00 .12 -.02 -.06 .04 -.13 -.00 -.03 .18* .04 -.lD -.16* -.04 .03.18*


RESULTS

Correlational Results Table 1 presents the means, standard deviations, and correlations of the variables. Intercorrelations between the control variables were of low to moderate strength. The correlations between the control variables and the two mentoring variables were all low, ranging from 0.00 to 0.24, indicating no problems of multicollinearity for our regression analyses.

Logistic Regression Results The results from the logistic regression are shown in Table 2. The logistic model fits the data nicely. Overall, 72.32% of the respondents were correctly classified by our model. The goodness-of-fit statistic (chi-square of 171.00 with 151 degrees of freedom) further confirms that the model fits the observed data. We observed a statistical significance level of p=0.13, indicating that our model does not differ statistically significantly from a 'perfect' model in which the observed probabilities exactly match those predicted by the model. Finally, similar to the overall F-test for multiple linear regression, the logistic model chi-square (chi-square = 50.7, 25 degrees of freedom, p=O.OO 17) allows us to reject the null hypothesis that all of the coefficients in our model are 0, except the constant. In sum, our model fits the observed data quite well.

Our results show that the only variables that have a statistically significant effect (at the p<0.05 level) on the respondent's range of contacts are having a mentor late, having a mentor early, and the amount of time respondents spent in teams. These results confirm hypotheses 1 and 2. Specifically, having a mentor early decreases the odds-whereas having a mentor later increases the odds-of an individual being in the high-range group. Counter to our expectations, the time a manager spent in meetings actually had a negative impact on their range of contacts. With respect to the other controls, the only other variable that had any statistically significant influence was expatriate status, which had a positive impact on range if we relaxed the confidence level (p<O.1 0).

The values shown in the last column of Table 2 allow us to gauge the magnitude of these effects. They indicate the amount by which the odds are increased that a respondent would fall in the category of those with a broad range of contacts, given a unit increase in the explanatory variable. Our results indicate that if a respondent had a mentor early in his career (the 'mentor early' variable changes from 0 to 1), the odds of the respondent's being in the high-range group falls statistically significantly (by a factor of .41). In short, an individual with an early mentor is only 40% as likely as someone who does not have an early mentor to fall into the high-range group. In contrast, we find that if a respondent has a mentor later in his career (the 'mentor late' variable changes from 0 to 1), the odds of the respondent being in the high-range group increase dramatically (by a factor of 3.81). These odds ratios confirm our main hypotheses.

In addition, expatriate status had a sizable effect on range. If the respondent was an expatriate, he was much more likely (by a factor of 2.67) to be in the high-range group. This result makes sense because expatriates are likely to have a cosmopolitan


Table 2. Results of logistic regression analysis for social capitala,b

Variables Regression Standard Significance Odds Coefficients Errors Levels Ratios

Mentor early -0.90 0.41 0.03 0.41

Mentor late 1.34 0.46 0.00 3.81

Employee of Matsushita -0.61 0.53 0.25 0.54

Employee of Philips 0.06 0.73 0.94 1.06

Relative importance of subsidiary 0.30 0.26 0.25 1.35

General management 0.52 0.95 0.59 1.68

Marketing function 1.13 0.93 0.22 0.31

Manufacturing function 0.10 0.93 0.91 1.11

Purchasing function 1.16 1.11 0.30 3.18

Research & development function 1.19 1.16 0.30 3.29

Finance function 1.02 0.97 0.30 2.77

LegaVother administrative functions 0.74 0.99 0.45 2.09

Tenure at current fmn -0.01 0.05 0.92 1.00

Years at other employers 0.03 O.OS 0.48 1.03

Number of other functions worked in -0.09 0.12 0.44 0.91

Number of other subsidiaries worked in 0.04 0.16 0.79 1.04

Speed of promotion 0.53 3.13 0.86 1.71

Expatriate status 0.98 0.52 0.06 2.67

Time spent in cross-subsidiary meetings 0.01 0.01 0.45 1.01

Time spent in cross-subsidiary teams -0.03 0.02 O.OS 0.97

Time spent in cross-subsidiary training -0.03 0.02 0.27 0.97

Received initial training -0.48 O.SI 0.3S 0.62

Perception of organizational autonomy 0.01 0.28 0.98 1.01

Perception of organizational informality 0.S4 0.38 O.1S 1.72

a Social capital is defined as the range or the number of national subsidiaries of the fmn in which the respondent has contacts. For this analysis, a manager may have 'many ties' (contacts in 11-17 subsidiaries) or 'some ties' (contacts in 1-10 subsidiaries).

b Goodness of fit:

Model Chi-Square

Goodness of Fit

Chi-Square

50.71

171.00

df

2S

lSI

Significance

0.00

0.13

orientation (Ritti)5 as well as contacts in their home country. Research has also shown that expatriates often serve as gatekeepers to the outside world in many subsidiaries, which also partially explains the results we observe (Edstrom and Galbraith 1977).

Finally, although participation in teams was a statistically significant variable in the overall model, the size of the effect is negligible; the odds of being in the highrange group versus the low-range group remain largely unchanged as participation increases.


Analyses of Variance Results Having confirmed our initial hypotheses, we performed a second level of analysis to explore further differences among respondents who fell into one of four groups: 1) those who had a mentor neither early nor late; 2) those who had a mentor late but not early; 3) those who had a mentor both early and late; and 4) those who had a mentor early but not late. In this way, we examined if having a mentor both early and late or neither early nor late had an equivalent effect on an individual's ability to develop social capital. In addition, we strove to understand individual differences between respondents that may have influenced the mentoring relationships they developed.

We used one-way analyses of variance with a dichotomous dependent variable to assess differences between the four groups. In accordance with our regression analysis, the results (see Table 3) show that respondents in each of these groups differed statistically significantly in their range of intersubsidiary contacts. More specifically, the range of contacts for respondents having a mentor late but not early (Group 2) was significantly greater than each of the other groups (i.e., having a mentor early but not late (Group 4), having a mentor both early and late (Group 3), and not having a mentor at all (Group I)). Interestingly, although not statistically significantly so, having no mentors at all was better than having an early mentor, in line with recent research that has suggested that those who take personal responsibility for building their social capital do better than those who rely on others to build contacts (Gabbay 1995, 1997). Also, having a mentor only early was clearly the worst scenario because those who had mentors both early and late (Group 3) appeared to have had somewhat higher range than those who had only an early mentor. All of these results are consistent with our hypotheses: having a mentor early creates a sidekick effect, whereas having a mentor later in one's organizational career helps a manager develop social capital.

Beyond differences in their range of contacts, there were a few other noteworthy differences among these groups. For instance, expatriates were more likely to have formed a mentoring relationship later in their career than earlier. We speculate that one reason might be that these individuals moved to a subsidiary late in their careers-thus becoming expatriates-and, at that stage of their careers, sought out mentors to help them perform better in the context of the broader multinational network.

Another difference to note is that initial training can be a substitute for initial mentoring. Those who received some formal training on joining the organization were most likely not to have an early mentor and most likely to have one later on (Group 2). In sharp contrast, 90% of those reporting having a mentor early but not late (Group 4) did not receive initial training. This finding has interesting implications for the design of career development programs. Rather than focusing on forming early mentoring relations, as many career development programs are increasingly attempting to do (Caruso 1992), firms may do better by providing initial training as a way of socializing new employees into an organization and letting individuals develop mentoring relationships on their own later in their careers. Indeed, research has suggested that some of the greatest benefits of the

~ 0- VJ

§. ~

g-;

s. ~

Des

crip

tion

of V

aria

ble

Gro

ups

5: ~

@

8. !l

g'

Dif

fere

nt·

_. E¥

-l9

e?..

;' '

0

~.

(1)

g n

{:,

§.

n 10

1 (l

..

C»

• CI5

~ ;

s "s

. 8:

5'

~ ....

. [

c:: "'I!

:S

Men

tor e

arly

M

ento

r lat

e ::r

_.

'"

!!!

Ef ~ ~ ~.

g D

epen

den

t Var

iab

le

~ g ~

>-j

g-So

cial

cap

itala

g.

g a

. ~ j

!i C

on

tro

l V

aria

ble

s (f

,I

U

'<

(f,I

.....:::

!Z el

",-a

Em

ploy

ee o

f Mat

sush

ita

g 61

g-8 ~

Em

ploy

ee o

f Phi

lips

~ g.

§ .€I

.. R

elat

ive

impo

rtan

ce o

f sub

sidi

ary

;.

iii'

<Ill

a· ~

Gen

eral

man

agem

ent f

unct

ion

_ ~

Sl g ~

Mar

ketin

g fu

nctio

n -

-"'"

3

0 !..

~

if (

1)

j!i

Man

ufac

turi

ng f

unct

ion

-; p

;' .. ~ ~

Purc

hasi

ng f

unct

ion

g. g-~

':-' §

Res

earc

h &

Dev

elop

men

t fun

ctio

n e;

ii" .

., ~

Fina

nce

func

tion

~. g

~ ~

Leg

al lo

ther

adm

inis

trat

ive

func

tions

~ g~.

!!!

Ten

ure

at c

urre

nt F

irm

o

_ 0

~

:-<

i

~.

Yea

rs s

pent

at o

ther

em

ploy

ers

~ ~.

~ ~.

N

umbe

r of o

ther

func

tions

wor

ked

in

:! if

go

~

Num

ber o

f oth

er s

ubsi

diar

ies

wor

ked

in

~~ iii

' ~

~ Sp

eed

of p

rom

otio

n to

cur

rent

pos

ition

8'

~ i

i' ci.

Exp

atri

ate

stat

us

S-~ ~

~

Tim

e sp

ent i

n cr

oss-

subs

idia

ry m

eetin

gs

~

3 e-

g T

ime

spen

t in

cros

s-su

bsid

iary

team

s s·

~ ~

;;;, T

ime

spen

t in

cros

s-su

bsid

iary

trai

ning

!

;,; 8

~ R

ecei

ved

initi

al tr

aini

ng

o §

S·

~ Pe

rcep

tion

of o

rgan

izat

iona

l aut

onom

y

J1; ~

::r

Perc

eptio

n o

f org

aniz

atio

nal i

nfor

mal

ity

!l ::r

No

No

9.73

0.41

0.

27

3.72

0.

17

0.29

0.

20

0.02

0.

07

0.10

0.

07

13.2

0 8.

68

2.27

0.

98

0.23

0.

32

11.8

3 5.

54

4.49

0.

34

3.42

2.46

No

Yes

12.6

9

0.76

0.

11

3.86

0.

02

0.25

0.

18

0.09

0.

05

0.20

0.

07

14.9

8 5.

03

1.60

0.

95

0.22

0.

67

13.0

2 8.

45

1.89

0.

67

3.81

2.35

Yes

Y

es

8.57

0.58

0.

15

3.64

0.

18

0.18

0.

22

0.0

5 0.

05

0.13

0.

12

13.2

2 6

.88

2.80

1.

02

0.23

0.

33

15.0

7 9.

58

6.12

0.

52

3.51

2.14

Yes

N

o

7.19

0.57

0.

19

3.52

0.

24

0.10

0.

10

0.14

0.

05

0.14

0.

14

11.7

1 10

.00

1.95

1.

00

0.29

0.

14

15.1

4 7.

05

5.67

0.

10

3.50

2.28

(1,2

) (2

,3)

(2,4

)

(1,2

)

(2,3

)

(2,3

)

(1,2

) (2

,3)

(2,4

)

(1,2

) (2

,3)

(3,4

) (1

,2)

(1,3

)

;'

S!: .. !'" ~

~ iii'

o ..., § f g, < ~.

§ ~ ~

0. r ~ '" f ~ .g '"

~

(1)

C/.l

0..

: (1

) t::

n ?I"

trl ~

~ .....

-.l

U.


mentoring relationships are derived through informal means and pertain to informal aspects of the company (Collin 1979). Alternatively, early training may serve as one of many mechanisms through which individuals can begin to develop helping relationships that, in the longer run, could blossom into mentoring relationships. Future research is needed that investigates the relationship between formal organizational initiatives, such as training programs, and the development of valuable informal relationships, such as mentoring.

Finally, the results suggest that those who form no mentoring relationships (Group 1) may be 'prisoners of their own perceptions.' These individuals scored lowest on perceptions of both organizational autonomy and informality. Because these people view the organization as being constrained, they may undervalue the benefits of mentoring relationships that can serve as pathways to accessing the informal side of the organization. This finding supports Turban and Dougherty's (1994) view that we need to examine more closely how an individual's personality and attitudes might affect the mentoring relationships he or she seeks and initiates and the benefits derived from them.

CONCLUSIONS AND DISCUSSION

Our objective in this chapter was to better understand how mentoring relationships actually work in organizational settings to affect specific career outcomes. We pursued this goal by examining both when mentoring relationships do work and when mentoring relationships do not work with respect to one career variable that scholars have linked to career progress-social capital, or beneficial ties with others in an organization. Our results suggest that the timing of having a mentor affects a protege's ability to develop social capital over the course of his or her organizational career. Specifically, having a mentor early in one's organizational career can produce a sidekick effect in which individuals are overshadowed by their sponsors such that their development of independent social capital is difficult. However, this sidekick effect disappears once the protege has developed a reputation as an individual contributor within the organization. At this later time, a mentor may be quite helpful in facilitating the formation of new ties since the protege has already developed some form of professional identity and reputation and, therefore, is not viewed as walking in anyone's shadow.

In addition, two other findings from this research were interesting and consistent across our analyses: expatriate status had a positive effect, and integrating mechanisms (such as participation in team work), had a negative effect upon the development of social capital. First, it appears that being an expatriate increases the likelihood that a protege will have many ties across multinational subsidiary boundaries. This seems plausible, given that an individual who has lived outside of his or her home country may have a better understanding of the cultures, history, language, and customs of other nations. Overall, expatriates may develop a more cosmopolitan orientation (Ritti)5 that later serves to facilitate the development and maintenance of international ties.

Second, integrating mechanisms, such as the amount of time spent engaged in team work or participation in training activities seems to have a negative impact on the ability to develop ties with others across subsidiaries. Although the specific type


of integrating mechanism varied somewhat across our analyses, this general finding seems worthy of further investigation. Drawing on the work of Louis (1990), we feel that these types of interactions may lead to densely clustered subnetworks which impede one's ability or interest in fostering a broad range of connections with others outside the subnetwork. As Louis' work suggests, those who develop strong 'buddy relationships' are less apt to negotiate relationships beyond their local work environments. Thus, peer group interaction can lead to fraternizing that impairs newcomer acculturation into the organization. In contrast, 'isolates' are more likely to gain a better understanding of their work environment simply because they have been forced to rely on multiple and varied sources of information. Thus, while interaction with others may provide social support, it does not necessarily lead to a broader network of contacts across intraorganizational boundaries.

Our study has several limitations that would be important to address in future research. First, and most importantly, our work is based on a limited amount of data regarding the nature of the mentoring relationships we studied. All we asked our respondents was whether they had a mentoring relationship or not. Factors not known to us-such as the identity of the mentor, the mentor's structural position in the firm, the process by which the mentor was chosen, and the nature and strength of the mentoring relationship--should all be studied to see whether and how they mediate our main findings. For example, we did not know whether the early mentor was also the protege's direct boss, and so could not compare supervisory and nonsupervisory mentoring relationships-a topic Scandura and Schriesheim (1994) have suggested for future research. Indeed, as Burt's (1992) work showed, hierarchical networks are most effective when they revolve around sponsors who are distant from the individual's immediate work group--that is, when the sponsor is not the manager's direct boss. In short, we expect that some mentors may be better than others at preventing their proteges from being seen as sidekicks.

Second, we studied just one issue of importance to an individual's career-the formation of social capital. Although social capital plays an important role in shaping such career outcomes as advancement, mobility, and performance, the amount of human capital (educational qualifications and skills) and physical capital (money) possessed by an individual can play an equally important role. Thus, even if we find that early mentoring relationships have a negative effect on the development of an individual's social capital, we cannot conclude that such relationships necessarily harm the individual's career because they may have a countervailing positive effect on the development of the individual's human capital.

Third, although we attempted to control for several factors that might influence the range of intersubsidiary contacts possessed by our managers, we recognize that there are other factors for which we did not control. For instance, recent research suggests that an individual's personality (on which we collected no data) can influence the extent to which they seek out and make use of mentors (Turban and Dougherty 1994). It has also been argued that gender and race can playa significant role in mentoring relationships (Ibarra 1993c; Thomas 1993). We were not able to assess gender effects in this study since all of our respondents were male; neither did we have data on race and, hence, could not control for such effects.


Finally, the group of managers we studied are only a subset of the population of all managers. They all worked in subsidiaries of foreign multinationals and were all male. Thus, our findings have limited generalizability. It would be useful to explore these issues in other organizational settings, especially in U.S. firms, and with managers of more diverse demographic characteristics.

Despite several limitations, this work has many important implications for those conducting research on mentoring relationships and on helping relationships more generally. First, by examining the impact of mentoring relationships on a protege's stock of social capital, we began to unravel possible explanations of how mentoring functions actually work to affect proteges' career outcomes. We proposed that there are several roles that mentors can take in providing career help to proteges; mentors may serve as signaling agents, providers of access, and providers of opportunities. Although we expect that such roles may lead to near term benefits for the protege, our findings suggest that there are also long term costs associated with having such relationships. Specifically, having a mentor early in one's career can produce a sidekick effect such that the development of ties with others in the organization becomes difficult.

Second, our findings and the rationale that supports them challenge assumptions in the mentoring literature. For instance, much mentoring work asserts that different mentoring functions work together to influence career outcomes, and that the more functions provided by a mentor, the more beneficial the mentoring relationship (Kram 1985; Noe 1988a). The research presented here, however, points to a contingency model in which we consider how different types of mentoring assistance provided at different times during a protege's career can affect career outcomes. This raises important questions regarding the appropriate sequence and ordering of specific career mentoring functions over the span of a protege's career. For example, is it beneficial for a mentor to provide access before the protege has established confidence and competence in his or her ability to excel in an organization? On a related note, this research calls for future work to examine whether there is a trade-off between which sets of mentoring functions best serve the protege early versus later in his or her organizational career. For example, is psychosocial help early and career help later the best combination of help for a protege who is trying to advance in an organization?

Third, this work suggests a host of opportunities for future research on both the timing and types of interventions designed to help in general (Higgins 1997a). For example, since our questions tapped whether or not a respondent had a mentor at different points in time, regardless of whether this was the same person doing the mentoring, our research opens up the possibility for future work on the optimal mix of help-providers over the course of one's career. Stretching this notion of timing even further, one could also consider how a protege's career is affected by the mix of mentors he or she has across the stages of adult development (Levinson et al. 1978) or during the transitions between them (Daloz 1986; Kegan 1982). Thus, rather than explore the timing of different types of help that are given by one helpprovider over the course of one mentoring relationship, researchers might also consider the combination of different types of help-providers an individual has over the course of his or her career and life.


In short, we need to consider not just knowing how mentoring relationships work, but also what effects the who and when of such relationships have on a protege's ability to progress in his or her career (Higgins and Thomas 1997). This approach would capture the inherently more contingent nature of careers and helping relationships. Most importantly, it would allow our thinking to move beyond the 'myth of the perfect mentor' (Hill and Kamprath 1991) to be more in line with recent research that has suggested that career developmental relationships are not monodyadic, hierarchical, and intraorganizational, but rather exist in constellations (Kram 1985) or portfolios composed of both superiors and subordinates and can exist both inside and outside the firm (Higgins 1997b; Thomas and Higgins 1996). Indeed, the sidekick effect itself is likely to be affected under such a perspective, for a person may be less likely to be overshadowed if he or she has several people from whom to draw career support rather than just one prominent superior.

In sum, the sidekick effect highlights the potential negative implications of early mentoring relationships with respect to one important career factor, the development of social capital, within a literature that has characterized such relationships as primarily positive. From a practical perspective, when initiating informal mentoring relationships, this work suggests that it is important to consider the possibility of a sidekick effect which could hinder a protege's ability to develop ties with others. And, from a theoretical perspective, when trying to understand how mentoring relationships actually work in organizational settings, this research suggests that it is important to recognize the inherently contingent and complex nature of giving and receiving help--in particular, to consider how the effectiveness of the type and timing of mentoring relationships can affect individual career outcomes.

We would like to thank Chris Bartlett and Sumantra Ghoshal for providing us with the data used in this study and Jack Gabarro, Richard Hackman, Linda Hill, Herminia Ibarra, David Thomas, and Douwe Yntema for their helpful comments on this research.

NOTES

1. We use the term 'organizational career,' rather than 'career,' because this research focuses on the impact of mentoring on an individual's career in one organization, as opposed to an individual's career in more than one organization. 2. 'Early' is defined as within the first two years of joining an organization. 'Later' or 'long term' is defined as anytime thereafter. 3. Statements on statistical significance refer to the .05 level. 4. We reached this conclusion after examining the frequencies table of the dependent variable and after reviewing the plot of residuals versus predicted values, which showed two parallel lines of scattered points. Note that this procedure is not equivalent to simply dichotomizing the range variable using a median split. Instead, the logistic regression analysis is conducted because the residuals violate the random distribution assumption in ordinary least squares models. 5. As referenced by W. Humphrey (1987).

•

Social Capital in Internal Staffing Practices

ABSTRACT

10 Peter V. Marsden

Elizabeth H. Gorman

This chapter examines the information sources that U.S. employers use in the course of internal staffing, that is, when promoting or transferring employees. We focus on the use of methods involving informal ties: referrals and direct approaches to candidates for promotion or transfer. Such ties may produce 'social capital' by providing employers with information about the qualifications and abilities of personnel; at the same time, they provide employees with information about opportunities for mobility within the workplace or firm. Data from a representative sample of work establishments indicate that informal methods are widely used in filling vacancies with internal candidates, often in combination with more formalized procedures such as job posting and seniority systems. Differences in internal recruitment procedures across types of employers and jobs suggest that they are selected in light of both efficiency concerns and pressures for equity and procedural rationality in the treatment of employees.

INTRODUCTION

Organizational employment policies and practices playa central role in structuring the inequality of rewards among individuals (Baron 1984). Careers consist of sequences of positions, which may involve moves-promotions, lateral transfers, or even demotions-within as well as between organizations. Promotion practices, in particular, shape upward career mobility for individuals (Baker, Gibbs, and Holmstrom 1994; Rosenbaum 1979a, 1979b; Stewman 1986; Stewman and Konda 1983) and influence earnings (Cappelli and Cascio 1991; KaUeberg and Lincoln 1988; Le Grand, Szulkin, and Tahlin 1994). One strand of research on

Social Capital in Internal Staffing Practices - 181

organizational promotion practices has examined structural, institutional, and market correlates of the prevalence of job ladders and other structured mobility channels within organizations (Baron, Davis-Blake, and Bielby 1986; Bridges and Villemez 1991; Kalleberg, Marsden, Knoke, and Spaeth 1996; Pfeffer and Cohen 1984). Other studies have explored the shape of formal job ladders and their interplay with informal career trajectories (DiPrete 1987; Miner and Estler 1985; Kanter 1983b). Still others have examined the criteria applied when rewarding individuals with promotions (Baker, Gibbs, and Holmstrom 1994; Bills 1988; Kanter 1977; Nicholson 1993).

In light of the centrality of intraorganizational mobility to careers, it is surprising that researchers have given little attention to the processes through which it occurs, that is, the manner in which organizations identify and select candidates for promotion. Both formal and informal methods exist for accomplishing these tasks. Seniority systems and job posting are the principal formal procedures. They identify and select from candidate pools using abstract, universalistic criteria. As such, they can be viewed as part of a larger process of bureaucratization within the employment relationship (Baron, Dobbin, and Jennings 1986; Baron, Jennings, and Dobbin 1988; Bridges and Villemez 1991), along with other features of internal labor markets. Indeed, the dearth of research on internal recruitment procedures may reflect an assumption on the part of researchers that promotion systems necessarily rely on formal selection procedures.

Yet even relatively bureaucratic organizations can and do make use of informal channels when identifying and selecting candidates for promotion (Pinfield 1995). Such methods include referrals as well as direct contacts between the selecting official(s) and candidates for promotion and transfer; they limit the pool of candidates to those with direct or indirect social ties to the selecting official(s). Studies examining individual differences in mobility within organizations have shown that particular configurations of social ties are associated with promotions (Burt 1992; Podolny and Baron 1997), and it is therefore of interest to examine them from the organization'S side.

In related research on methods of hiring from external markets, researchers have demonstrated that organizations frequently rely on recruitment and selection procedures that make use of candidates' social ties (Fernandez and Weinberg 1997; Fevre 1989; Jenkins, Bryman, Ford, Keil, and Beardsworth 1983; Kalleberg et al. 1996: chapter 7; Marsden and Campbell 1990). A cognate body of work has demonstrated that the form and content of social ties is linked to individual success in finding jobs (e.g., Boxman, De Graaf, and Flap 1991; Flap and Boxman, this volume; Granovetter 1995; Lin, Ensel, and Vaughn 1981; Marsden and Hurlbert 1988; Wegener 1991). This chapter focuses on similar phenomena in the intraorganizational setting.

In the next section, we discuss the manner in which we view organizational information channels as avenues for the operation of social capital. We then describe the various methods that organizations use in disseminating information about opportunities for promotion and transfer. Our theoretical account of variation in the use of informal methods is based on efficiency and strategic concerns, together with external pressures for equity in the allocation of opportunities. After introducing the


data base for our analysis, the 1991 National Organizations Study (NOS), we present findings, and conclude by discussing their implications for organizations and employees.

INTERNAL STAFFING PROCEDURES AND SOCIAL CAPITAL

We view recruitment procedures that draw on preexisting social ties as settings in which the operation of social capital within organizations may be observed. Because the concept of social capital has been used in diverse ways, however, as illustrated by discussions in Foley and Edwards (1997), Greeley (1997), and Newton (1997), analyses invoking it must be explicit about its denotation. Most or all variants on the concept posit that social ties may serve as resources for individual or collective action; we take this as the essential core of social capital, as distinct from related or additional features such as norms of trust that may be associated with the presence of social ties. The ties in question need not be channels deliberately developed to facilitate purposive action; indeed, much is made of the unplanned or adventitious use of social ties originating outside of a given context. Thus, Granovetter (1985) contends that economic transactions are often 'embedded' in pre-existing social relations, while Coleman (1988) stresses the 'appropriability' of social organization-i.e., its use for purposes other than those for which it was established-as a fundamental characteristic of social capital.

Different conceptualizations of social capital highlight its role as either a property of collectivities and communities or an individual resource. Both conceptual strands are evident in Coleman's (1988) influential early statement. As a collective property, social capital inheres in the web of social relations within a collectivity, and the set of obligations, information channels, and norms that those relations support and enforce (Coleman 1988). Like clean air or municipal policing, social capital at this level is a collective good, one that can facilitate both collective and individual actions (Coleman 1988; Putnam 1993b, 1995a). As an individual resource, social capital resides within the direct and indirect social ties giving the social location of a focal individual. Like human capital, social capital at this level provides individuals with differential competitive advantages (Burt 1992).

In this study, we emphasize the second conceptual strand; we conceive of informal channels within organizations as vessels for social capital that enable employers to gather information of superior quality about candidates for promotion or transfer. Individuals, of course, benefit from being in social locations permitting access to such networks. As well, the staffing decisions reached by making use of the social capital contained within informal channels may result in collective benefits such as strengthened organizational performance, but we do not stress this here. We direct our attention to the nature and determinants of organizational staffing procedures that draw on individual social capital and thereby enhance its influence on individual mobility outcomes.

INTERNAL RECRUITMENT AND SELECTION METHODS

The principal contrast we draw in studying internal staffing methods parallels the formaVinformal distinction used in existing literature on job matching in external


markets. We begin by discussing informal methods involving referrals and direct approaches to employees. We then mention two formal methods, seniority and job posting systems, more briefly.

Informal recruitment involves information transfers in social networks, and may take several distinct forms. Within a work unit, supervisors usually monitor the performance of employees on an ongoing basis. Using this information, they may directly approach employees they deem qualified when higher-level vacancies occur or new positions are created. Information about employees who might be promoted or transferred may also be shared through word-of-mouth referrals from other supervisors, other employees, or even people outside the firm such as clients or customers. Such channels contain social capital because they can convey relatively rich, qualitative information about the capabilities of potential candidates for a new position; reciprocally, they may also provide candidates with 'realistic job previews.'

A prototypical case in which one would observe informal staffing is that of 'drive systems' involving direct, personal control (Edwards 1979; Jacoby 1985). Such systems delegate substantial discretion to supervisors who may be capricious, even despotic. Pinfield's (1995) description of staffing as an 'administered process' in his case study of a forestry firm involves important elements of informal recruitment, though it occurred in a context where formal procedures also were present. He observes that managers preferred the administered appointment process because it took less time, and permitted them to 'manage lines of progression' for their subordinates. With administered appointment processes, managers could use the promise of promotion as a means of motivating lower-level employees within their units; they were also able to reduce their risks because informal channels yielded trustworthy information about the personal reputations of candidates.

Of course, as drive systems and personal control remind us, by using informal channels to disseminate and gather labor market information, organizations provide an opening for not just supervisory discretion and good jUdgment, but also prejudice and favoritism, to enter promotion and transfer decisions. There may, then, be liabilities as well as benefits to the use of social ties. This highlights the omnipresent distinction between form and content in the study of social networks; it is problematic to infer content from form. The fact that personal ties may be used not only to transmit subtle, local 'impacted' information, but to exclude and/or unduly advantage some candidates at the expense of others, raises concerns about equity in the treatment of employees and calls for limitations on the discretion of supervisors.

Formal methods of internal recruitment have arisen in response to these concerns. Seniority-the allocation of promotion opportunities on the basis of length of service-represents one approach to limiting managerial caprice and thereby increasing universalism within the promotion process. Kochan, Katz and McKersie (1994) include seniority among the central elements of the Job control' industrial relations system that was widespread in the U.S. in mid-century, particularly in unionized settings. Freeman and Medoff (1984) identify several ways in which the seniority principle may enter into promotion decisions, ranging from seniority as the dominant factor in promotions (irrespective of ability) to seniority as a basis for setting priorities among candidates whose qualifications are otherwise equivalent.


Job posting constitutes a distinct approach to formalizing procedures for promotions and transfers within internal labor markets. A posting system publicizes internal job opportunities through written notices on bulletin boards or employee newsletters; see Pinfield (1995) for a generic description of such a system. Announcements of vacancies include job descriptions and specifications of qualifications; current employees are encouraged to put themselves forward as candidates to fill vacancies. Posting systems introduce open competition into the allocation of job opportunities to current employees.

Pinfield (1995) observes that posting systems provide a high degree of procedural rationality and legitimacy to promotion and transfer decisions. Jacoby (1985) notes that unions view posting as an alternative or complement to seniority as a mode of developing job ladders and providing advancement opportunities for their members. Additionally, many seeking policy levers with which to reduce sex segregation in employment (e.g., Kanter 1977) advocate job posting as one approach to increasing equality of opportunity within organizations by publicizing vacancies beyond the reach of informal ties that are often confined within work units.

THEORETICAL CONSIDERATIONS IN THE SELECTION OF INFORMAL STAFFING METHODS

Our concern in this chapter lies in the factors that make it more or less likely that an organization will use informal channels in internal staffing decisions. Some factors of interest refer to the organization per se, while others pertain to differences in the nature of work done in positions within organizations. We argue in this section that these factors operate through several distinct intervening mechanisms. From a rationalist standpoint, some factors are linked to the benefits to be gained by using informal channels or the costs (including opportunity costs) of doing so. Others instead reflect institutional pressures toward equity or procedural rationality in the management of staffing, which may emanate both from sources internal to an organization and from the environment in which it is situated.

Benefits: Information QUality From the standpoint of rationality and efficiency, one would expect organizations to adopt staffing methods that rely on interpersonal connections under circumstances in which the benefits of such methods are greater. The principal benefit of informal methods drawing on social capital is that they may provide richer, more extensive, and more situated information concerning candidates than can formal, impersonal methods.

Informal methods involving social ties should be especially likely to yield higher-quality information when interpersonal skills and the ability to form networks of contacts are central to effective performance. Such skills are important in jobs that involve frequent communication with clients or customers; for example, investment banks rely on their employees' networks of external ties to bring in new business and maintain client relationships (Eccles and Crane 1988). We therefore expect to find greater use of informal internal staffing methods in professional, sales, and service occupations.


Interpersonal skills and social networks are also important in flexible organizations with staffing strategies that adapt goals and objectives to the capabilities of their personnel, rather than seeking people who fit the requirements of standardized jobs. In such organizations, staffing can be seen as equivalent to strategy formation (Snow and Snell 1993). They make less use of formal job descriptions and rules to clarify and coordinate tasks, and instead expect the contours of jobs to be altered to suit the talents and initiative of their incumbents. In such situations, networks of internal ties serve as a coordination mechanism (Eccles and Crane 1988) and convey clear normative expectations associated with each employee's role in the organization (Podolny and Baron 1997). Given the need for rather nuanced, local knowledge of someone's capabilities in such systems, we anticipate that organizations without formal job descriptions will tend to make greater use of informal staffing procedures.

Informal selection methods should also provide better-quality information when the skills involved in the position to be filled are difficult to measure objectively. If the responsibilities of the position to be filled are nonroutine and involve the exercise of discretion, objective standards are difficult to apply. Subjective evaluations obtained through social networks may provide the best available information concerning a candidate's potential (Pfeffer 1977). This argument suggests that organizations will rely more heavily on informal selection methods for managerial, professional, and skilled-craft positions. For managers, this claim is supported by a study of supervisory selection in plants within eight manufacturing industries (Northrup et al. 1978) which found that 'most procedure is highly informal.'

Finally, informal selection methods should provide better-quality information in circumstances where the costs of a selection error are relatively high. In such cases, employers are more likely to be interested in obtaining the fine-grained information concerning candidates' personal characteristics that only informal procedures can provide. The costs of a selection error are higher when an employer invests in the training of promoted employees, because such investments will be lost if the subsequent performance of the employees is poor. Selection errors can also be costly when promotions occur within multiple-level job ladders, because incumbents in such positions are likely to have long tenures in the organization. Due process guarantees may make it difficult to discharge them, and selection for a position on a job ladder is likely to enhance an employee's eligibility for further promotions. We therefore expect to find greater use of informal methods in occupations that receive formal training and that have mUltiple levels (i.e., that form part of job ladders) .

Administrative and Opportunity Costs The intrinsic costs of administering informal procedures for identifying and selecting promotion candidates are generally low. Because substantial informal information is acquired passively in the course of undertaking other tasks, it can be viewed as essentially costless. Even when invoked actively, the use of such methods may involve nothing more than picking up the telephone or wandering down the hall to discuss a candidate; of course, intensive pursuit of closely-held information through informal ties may be costly, especially for confidential matters.


One reason for relying on relatively low-cost informal methods is that alternatives are too expensive. Small, single-site organizations may not find it costeffective to administer formal staffing procedures, for example. Small establishments that are part of multi-site organizational systems may, however, be required to use centrally prescribed formal procedures for making promotion and transfer decisions, and hence be less apt to use informal methods of doing so.

Because the range of candidates that can be identified using informal ties is limited to those who can be tapped through the interpersonal networks of selecting officials, a principal cost (or social liability) of informal methods is the foregone opportunity of missing talented candidates who are not identified through the social ties used for an internal personnel decision. In small, single-site establishments, this opportunity cost should be low, because selecting officials are apt to have direct or indirect social ties to most or all employees. In larger establishments and establishments within multi-site organizations, on the other hand, the pools of candidates that can be defined on the basis of social ties will usually consist of much smaller fractions of the potentially eligible workforce, creating a greater opportunity cost. Accordingly, we expect to find less use of informal procedures in large or multiple-site establishments.

Constraints: Equity and Rationality Pressures As noted above in the discussion of seniority systems and job posting, constituencies both internal and external to organizations exert pressures on them to allocate rewards, including promotion and transfer opportunities, in an equitable and procedurally rational manner. Informal methods of identifying and selecting promotion candidates are more vulnerable to charges of prejudice and favoritism than are formal procedures that rely on objective, universalistic criteria. Informal procedures are also less consistent with norms of rationality and bureaucratization, which many see as institutionalized myths or societal values that infuse organizations (Meyer and Rowan 1991; Bridges and Villemez 1991; Baron, Dobbin, and Jennings 1986).

External pressures emanate from unions and government regulators. Differences in personnel practices between unionized and nonunion workplaces have been stressed by a number ofresearchers (e.g., Jacoby 1985; Baron, Dobbin, and Jennings 1986; Dobbin et al. 1988). Cohen and Pfeffer (1986), for example, argue that unions advocate formalized staffing procedures because such procedures can prevent employers from penalizing employees with pro-union attitudes. Institutional arguments hold that exposure to the public sphere places organizations under special pressure to conform to evolving norms about legitimate employment practices (Dobbin et al. 1988). Larger establishments and establishments that are part of multi-site organizations are more visible to regulators, and consequently such establishments should be more reluctant to make use of informal promotion procedures. Public sector establishments, in particular, must demonstrate high levels of fairness, objectivity, and openness in their employment practices; they are also subject to civil service laws and regulations that mandate the use of certain formal procedures (DiPrete 1989; Tolbert and Zucker 1983). Thus public sector


establishments should be especially likely to avoid the selection of personnel through informal methods.

Internally, personnel departments are likely to favor formal methods of identifying and selecting promotion candidates. One reason is that personnel professionals are especially aware of the constraints and sensitivities of external constituencies (Jacoby 1985); a second is that they enhance their intraorganizational power through the possession of specialized knowledge about the conduct of personnel actions (Pfeffer and Cohen 1984). We therefore expect that establishments having personnel departments will make less extensive use of informal staffing procedures.

THE NATIONAL ORGANIZATIONS STUDY

The data presented in this chapter are drawn from the National Organizations Study (NOS), conducted during 1991. Telephone interviews were completed with informants for a multiplicity sample (see Parcel, Kaufman and Jolly 1991) representative of U.S. work establishments. l When contacting establishments, NOS interviewers were instructed to speak with 'the head of the personnel department or the person responsible for hiring.' Overall, the NOS attempted to contact informants for 1,067 establishments; it successfully conducted interviews with 688 of them, a completion rate of 64.5 percent.2 For additional details about field procedures used in the NOS, see Spaeth and O'Rourke (1994) or Kalleberg et al. (1996: chapter 2).

To take into account possible between-occupation, within-establishment variation, the design of the NOS interview schedule specified that several question sequences, including that having to do with internal staffing, were to be repeated for up to three different occupations in each establishment. One of these was the job title of the employees 'most directly involved' with the main product or service provided by the establishment; below we refer to this as the 'core' occupation. A second was the occupation given by the General Social Survey (GSS) respondent who had given the name of the establishment. Finally, questions were also posed about 'managers or other administrators.' This multiple-occupation design permits the separation of establishment- and occupation-level influences on several NOS outcome variables.

MEASURING INTERNAL STAFFING METHODS

To measure the procedures used when filling a job internally, the NOS used this question sequence:

When you fill this job with a person already in the organization, how often do you 1. Consult a seniority list? 2. Inform current employees by posting or circulating a vacancy notice? 3. Ask the person leaving the job to recommend other current employees? 4. Ask others at your workplace for recommendations? 5. Go directly to specific employees and encourage them to apply?


Table 1. 'Frequent' use of internal staffing methods by NOS establishments

Percentages of Establishments

Referrals Any

Seniority Job from Referrals Direct Informal Occupation Lists Posting Incumbent from Others Approach Method

'Core' employees

Unweighted 38.5% 67.8% 8.1% 24.1% 19.4% 34.8%

Weighted 26.9 42.2 12.7 25.0 24.1 38.1

(N) (387) (388) (382) (386) (386) (388)

'GSS' employees

Unweighted 34.4% 64.7% 8.3% 16.9% 16.4% 29.0%

Weighted 39.0 45.3 12.8 22.6 25.6 34.0

(N) (221) (221) (218) (219) (219) (221)

Managers and

Administrators

Unweighted 18.4% 59.5% 12.4% 25.2% 28.4% 43.0%

Weighted 29.8 41.6 11.7 22.4 33.1 41.1 (N) (446) (449) (442) (445) (447) (449)

Note: Questions about internal staffing were asked only of informants who stated that current employees were 'sometimes' promoted or transferred to fill vacancies in a given occupation.

Informants were asked to indicate whether they used each method 'frequently,' 'sometimes,' or 'never.' The sequence of questions was repeated for each of the three occupations, if the informant stated, in response to a filter question, that the establishment ever fills vacancies in that occupation with people it already employs.3

Table 1 presents the percentages of informants who answered 'frequently' to the items in the sequence, separately for the three occupations studied in the NOS. It shows that formal staffing procedures-seniority and job posting-are most widely used. The un weighted percentages, which reflect the experience of the typical U.S. employee (see note 1) show that between three-fifths and two-thirds of internal vacancies are frequently advertised through posting of lists or circulation of vacancy notices. The seniority principle is frequently used for almost 40 percent of vacancies in core jobs, and even for 20 percent of managerial positions.

There is, nonetheless, appreciable use of informal methods in internal staffing. Only about a tenth of positions are frequently filled by asking the previous occupant of a position to recommend possible successors, but recruiting for nearly a quarter of them often draws on other interpersonal referrals. Employers make frequent direct approaches to candidates for promotion and transfer for about a fifth of core positions, and for almost 30 percent of managerial jobs. The final column shows that at least one of the three informal methods is frequently used for more than a third of core jobs, and for 43 percent of manageriaVadministrative vacancies. Results not displayed in Table 1 indicate that informal methods were 'never' used for less than 15 percent of jobs studied in the NOS.

The weighted results in Table 1 estimate the distribution of internal recruitment methods across establishments (see note 1). Percentages for job posting are lower in


the weighted results, indicating that this procedure is less often used in the smaller establishments which become more numerous when the data are weighted. Conversely, weighted percentages generally exceed the unweighted ones for seniority and the three informal methods, suggesting that these methods may be more often used in smaller establishments.

Crosstabulations of the frequency with which establishments use formal and informal methods of staffing (not shown) reveal negligible associations. The fact that these associations are not negative indicates that formal and informal methods do not constitute mutually exclusive approaches: informal channels are sometimes used alone, and sometimes as a supplement to posting or seniority. Establishments using one type of informal channel also tend to use others: there are substantial positive associations linking all three pairs of informal methods in Table 1.4

ORGANIZATIONAL AND OCCUPATIONAL CORRELATES OF INFORMAL STAFFING PROCEDURES

We have argued above that staffing methods are chosen in light of their anticipated information benefits and both administrative and opportunity costs, as well as internal and external pressures for equity and rationalization in the management of internal personnel actions. We suggested that as a result of the operation of these mechanisms, we should observe variation in the use of informal methods across establishments according to size, affiliation with a larger, multi-site organization, sectoral location, and the presence of personnel departments, specific job descriptions, and unions. At the occupational level, our three mechanisms predict variation in the use of informality across types of occupations, and according to the presence of training and job ladders. This section presents our findings; descriptions of our measurements of these explanatory variables appear in the appendix.

Bivariate Differences in Internal Staffing Methods Table 2 provides a preliminary indication of the relationships between our organizational and occupational covariates and the use of informal methods of identifying and selecting candidates for promotion or transfer. The first column reports percentages of establishment-occupation observationsS of a given type for which informants reported that at least one informal method was used 'frequently.' The other columns give similar percentages for the three specific informal methods.

The results in Table 2 are broadly consistent with the theoretical ideas developed above. As establishment size increases, informal methods are less often used to fill occupational positions; nearly half of observations from establishments size 10-49 use at least one informal method frequently, but only about 30% of those from establishments with more than 500 employees do so. About 44% of observations from independent establishments make frequent use of an informal method, as compared to a third of those from within multi-site organizations. These differences reflect the greater visibility of larger and multiple-establishment firms, as well as size differences in the information benefits, opportunity costs, and administrative efficiency of staffing methods.


Table 2. Bivariate differences in internal staffing methods (all occupations)

Percentage Using Method Frequentlya

Any Incumbent Referrals Direct

Variable Informal Referrals from Others Contact Nb

Method

Size

1-9 38.9% 14.9% 22.7% 30.0% (87-90)

10-49 46.0 9.7 26.5 32.4 (247-250)

50-99 38.2 9.1 22.9 23.2 (142-144)

100-499 34.0 10.6 21.9 19.8 (255-259)

500+ 31.3 8.9 22.0 14.1 (302-307)

Multi-site organization

No 43 .6 9.5 20.9 27.1 (346-351)

Yes 33.8 10.2 27.3 20.4 (696-707)

Auspices

Private 40.4 12.2 24.3 26.6 (615-622)

Public 32.2 6.9 21.3 17.5 (334-342)

Nonprofit 33.0 6.5 21.3 14.9 (93-94)

Union presence

None 41.0 11.0 25.1 20.6 (520-531)

Some 33.3 9.1 21.2 24.8 (518-523)

Personnel department

No 41.5 9.9 25.1 28.8 (523-530)

Yes 33.2 10.5 21.3 16.5 (497-506)

Job descriptions

No 43.2 13.2 22.6 34.5 (136-139)

Yes 36.1 9.5 23.1 20.8 (906-920)

Training

No 34.6 9.6 18.7 23.5 (239-243)

Yes 38.1 10.1 24.6 22.5 (791-803)

Multiple levels in

occupation

No 39.8 9.9 22.6 24.6 (322-327)

Yes 35.8 10.0 23.3 21.7 (720-731)

Occupation

Managerial 42.3 12.0 25.0 28.0 (518-525)

Professionalffechnical 34.7 9.2 26.7 15.2 (141-147)

Sales/Service 43.0 8.4 25.2 26.2 (107)

Administrative Support 27.6 10.5 18.6 14.0 (86-89)

Craft 43.3 20.0 26.7 30.0 (30)

Unskilled 22.2 3.1 13.7 13.0 (160-162)

a Based on unweighted NOS data.

b Observations are occupational groups within establishments. Owing to missing data, Ns for the different methods vary slightly.


Informal methods are less often used frequently when an organization has a personnel department, when most positions have written job descriptions, when an establishment operates under public or nonprofit auspices rather than in the private, for-profit sector, or when unions are present. These findings substantiate the arguments we offered above about staffing strategy and constraints as factors shaping the selection of staffing practices.

Turning to occupational characteristics, Table 2 shows slight tendencies to make more use of informal procedures when those selected to fill a vacancy are to receive formal training. Contrary to our expectations, informal methods are slightly less likely to be used in occupations with multiple levels, which offer the prospect of future promotion. As we anticipated, though, the use of informal methods differs appreciably across kinds of work. Internal staffing decisions for managers, professional and technical occupations, sales and service occupations, and craft occupations make use of informal methods substantially more often than those for clerical (administrative support) and semi- or unskilled occupations. This pattern suggests that organizations rely on informal methods when filling autonomous, structurally unique positions in which performance is not readily metered.

Multivariate Analyses We use ordinal logistic regression to examine the partial effects of organizational and occupational variables on the frequency with which informal staffing methods are used. This model is the appropriate choice for the study of ordered dependent variables like those studied here (Long 1997; Winship and Mare 1984). A positive coefficient indicates that an increase in the value of the explanatory variable is associated with a rise in the odds of using a given internal staffing method at all rather than 'never,' or 'frequently' rather than less often.6 Table 3 presents the results of these analyses.7

Beginning with organizational factors associated with the use of any informal method of internal staffing, we find statistically significant8 tendencies for establishments linked to multisite firms, those having personnel departments, and those in the public rather than the private sector to be less likely to use informal processes in internal staffing. The odds of using an informal method in multisite establishments, net of other factors, are 0.69 times as large as in independent establishments; in public sector establishments, they are 0.56 times as large as in the private sector. Of particular interest is the finding that the bivariate negative association of establishment size with informal methods in Table 2 becomes statistically negligible when we adjust for the other independent variables included. Thus, the size differences in staffing methods seen in Table 2 are attributable to the fact that size is associated with factors that have statistically significant coefficients in Table 3, including the presence of personnel departments, multi site affiliation, and occupational mix. Contrary to our expectations, the presence of a union does not seem to reduce the use of informal staffing procedures, net of the other factors we have controlled. Written job descriptions are negatively associated with the use of informality in promotions and transfers, but this association is statistically nonsignificant.


Table 3. Organizational and occupational correlates of informal internal staffing: ordinal logit

coefficientsa

Internal Staffing Method Explanatory Any Informal Incumbent Referrals Variables Method Referrals from Others Direct Contact

Organizational

characteristics

Estabishment size (log) -.002 (.047) -.014 (.048) -.007 (.046) -.002 (.047)

Multisite organization -.373** (.135) .069 (.138) -.190 (.133) -.511 *** (.134)

Public secto{ -.573*** (.149) -.520*** (.153) -.421** (.146) -.621*** (.150)

Nonprofit sectorC -.240 (.228) -.039 (.227) -.202 (.225) -.154 (.224)

Union presence -.030 (.107) .083 (.109) .038 (.104) .134 (.108)

Personnel department -.345* (.172) -.082 (.176) -.406* (.168) -.549** (.175)

Job descriptions -.213 (.199) -.201 (.204) .030 (.194) -.413* (.200)

Occupational

Characteristics

Formal training .500** (.158) .210 (.163) .476** (.154) .395* (.158)

Multiple levels -.260 (.142) .051 (.146) .217 (.140) -.138 (.141)

ManagerialC 1.042*** (.188) 1.222*** (.206) .366* (.181) 1.015*** (.190)

Professionalffechnicalc .810*** (.241) 1.139*** (.257) .731** (.233) .584* (.238)

Sales/Servicec .850*** (.255) .529 (.273) .683** (.244) .835*** (.253)

Administrative SupportC .216 (.262) .946*** (.282) .052 (.258) .281 (.263)

CraftC .574 (.398) .914* (.412) .346 (.378) .486 (.403)

Threshold parametersb

First cutpoint -1.821 .822 -.370 -1.194

Second cutpoint .644 3.082 1.667 1.230

Log-likelihood -966.89 -918.59 -1046.91 -987.48

N 1014 998 1006

Source: 1991 National Organizations Study. Standard errors given in parentheses. * p<.05 ** p<.Ol ***p<.OOl

1010

aBased on unweighted NOS data. Observations are occupational groups within establishments. Owing to missing data, Ns for the different methods vary slightly.

bConstant term is constrained to zero.

cReference category for auspice dummy variables is the private for-profit sector; reference category for occupation dummy variables is unskilled occupations.

The odds of using an informal staffing method grow by a factor of more than 1.6 when those in an occupation receive formal training, suggesting that employers seek more detailed information when considering internal candidates for such jobs. The presence of job ladders (as measured by multiple levels), on the other hand, has no statistically significant association with informality in internal staffing. Consistent with our hypotheses, informal procedures are used much more often for some types of occupations than for others. The findings show that the use of any informal channel is most common for managerial employees; however, such


methods are also more common for professionaVtechnical and sales/service occupations than for semi- or unskilled ones. As anticipated, the coefficient contrasting craft and unskilled occupations is positive, but it does not reach conventional levels of statistical significance.

Several differences in the effects of certain organizational variables across specific informal methods of identifying and selecting internal candidates are worthy of note. Differences are more strongly patterned for direct approaches than for either type of referral. Affiliation with a multi-site organization primarily reduces the frequency with which direct approaches to candidates are used, as does the presence of formal job descriptions. In contrast, location in the public sector has a statistically significant negative coefficient for all three informal methods. The presence of a personnel department is negatively associated with relying on referrals from employees other than the position's current incumbent, and with making direct approaches to candidates; but having a personnel office does not reduce the frequency with which employees leaving a job are asked to recommend potential successors.

DISCUSSION

This chapter examined the organizational side of promotion and transfer events, and the extent to which they involved individual-level social capital, in the form of interpersonal networks linking selecting officials to candidates. Using data from a national study of U.S. employers, we found that informal methods of recruitment and selection-direct approaches to candidates as well as referrals-play a substantial part in internal staffing actions; they were 'frequently' involved in promotions and transfers into more than a third of the occupations studied, and 'never' involved in only 15 percent of them. While formal procedures for allocation of internal vacancies, particularly job posting, were even more widespread, this nonetheless indicates that many U.S. employers find it advantageous to rely on interpersonal as well as impersonal sources of information when making internal staffing decisions.

Our multivariate analyses demonstrated that variation in the use of informal methods for internal staffing is patterned across types of organizations and types of work in a fashion to be anticipated, given the differential information benefits, administrative expenses and opportunity costs we associate with such methods in distinct situations. Particularly notable are the between-occupation differences, which suggest that employers seek to activate the social capital embodied within informal ties to obtain information about candidates for positions involving training, autonomy, and interpersonal skills, in which performance cannot be readily assessed. We found it interesting, however, that scale per se was not linked to informality in staffing, given our covariates.

Our data also indicate that an employer's use of social capital to inform promotion and transfer decisions is strongly constrained by forces both internal and external to an organization. The presence of a personnel department substantially reduces reliance on informal ties in internal labor markets; externally, location in the public sector has a similar effect. These findings are consistent with institutional theorizing (e.g., Dobbin et al. 1988) which stresses the maintenance of


organizational legitimacy as an important priority for organizations. Here, legitimacy is presumably enhanced through the use of staffing methods that assure equity and procedural rationality. We found it somewhat curious, therefore, that the presence of unions was not similarly associated with a decline in the use of informal networks in internal staffing decisions.

In closing, we offer two further observations about the part played by social networks in promotion and transfer processes. The first is that there is some reason to think that networks are even more involved in internal staffing processes than our data indicate. We have examined the procedures that establishments use to disseminate knowledge about vacancies and assemble information about candidates for intraorganizational mobility. Candidates need not necessarily hear about opportunities via these channels; even formally distributed information quickly enters and diffuses through workplace social networks. Suggestive evidence on this point comes from the 1991 ass, which provided the sampling frame for the NOS. ass respondents who had held more than one job with their current employers were asked about how they heard about their present jobs. The most common answers cited interpersonal contacts: more than half (51 %) of the respondents said that they had been approached and asked to take their new jobs; nearly a third (31 %) learned of their new jobs through a supervisor, while 15% referred to a coworker. Less than a fifth heard about their jobs by virtue of job posting or a vacancy list (17%) or as a result of unions or seniority (16%).9

Secondly, theoretical understandings of social capital stress the superior capacity of interpersonal ties to convey certain types of information. Taking this as a premise, we have invoked the usual 'norms of rationality' (Thompson 1967) to make predictions about which employers will make use of particular staffing methods. We enter the caveat, however, that informal methods may be used in ways that diverge from received theory. Indeed, this may occur for formal methods as well: Reskin and Hartmann (1986) review cases in which bidding restrictions or seniority provisions undermined efforts to broaden gender equality in access to desirable jobs through job posting practices. It seems especially likely, however, that informal methods of staffing might be used for reasons other than their superior information benefits. Though as social capital informal ties have the capacity to convey subtle and nuanced information about the capabilities of candidates, as social liabilities they simultaneously can convey other, less meritocratic information-about, say, interpersonal loyalty rather than technical ability. By using interpersonal channelsthat tend to link socially similar persons-when diffusing information about promotion and transfer opportunities and assembling information about candidates for them, employers run the risk of excluding those lacking such social capital from consideration. This at once raises equity concerns (from the individual's side) and opportunity cost worries (from the employer's).

We think that organizations face a continuing dilemma about how to balance the information benefits and risks associated with the use of individual social capital in personnel actions. To the extent that selection into jobs in contemporary organizations requires information that is accessible only through interpersonal ties, employers must also provide safeguards against supervisory malfeasance in using networks, guaranteeing that this takes place in an atmosphere of procedural


rationality. An intriguing direction for future research would be to determine what differences exist between employers relying on pure 'network staffing'-that is, those that use informal methods alone-and those using networks as an adjunct to more impersonal methods.

APPENDIX: MEASUREMENT OF EXPLANATORY VARIABLES

This appendix describes the measurements of independent variables that appear in Tables 2 and 3. Descriptive statistics reported are for the set of 1091 organizationoccupation observations examined in this chapter, for which internal vacancies were ever filled using current employees. More details on measures in the NOS appear in Kalleberg et al. (1996).

Size. Natural logarithm of the number of full-time employees in the establishment (mean, 4.98; standard deviation, 1.98). Table 2 uses actual size.

Multi-Site Organization. Dummy variable identifying the 67% of observations from establishments that are part of larger, multiple-establishment organizations.

Personnel Department. Dummy variable identifying the 52% of observations from establishments that have a separate department or section responsible for personnel and/or labor relations.

Job Descriptions. Dummy variable identifying the 87% of observations from establishments in which there are written job descriptions for most jobs.

Union Presence. Scale combining four items indicative of the presence of unions (mean, 1.50; standard deviation, 0.64).

Public Sector. Dummy variable identifying the 32% of observations from establishments operated by federal, state, or local governments.

Nonprofit Sector. Dummy variable identifying the 9% of observations from private, not-for-profit establishments.

Training. Dummy variable identifying the 76% of observations in which those in an occupation had received formal training within the past two years.

Multiple Levels. Dummy variable identifying the 69% of observations in which an occupation has more than one level.

Occupational Categories. Core and ass occupations were classified into threedigit 1980 Census codes by the NOS. We subsequently grouped them into the six broader classes used in this chapter. No specific occupational title was used by the NOS when asking about 'managers and administrators,' so we classified all observations for these occupations into the 'managerial' group. Overall, 50% of our observations are managerial, 14% are professional, 10% are in sales or service occupations, 8% are in administrative support occupations, 3% are in craft occupations, and 15% are in semi- or unskilled occupations.

Data collection and writing were supported by National Science Foundation awards SES-89-11696 and SBR-95-l17l5. For helpful comments, we are indebted to the editors, Roger Th.AJ. Leenders and Shaul M. Gabbay.

NOTES

I. A work. establishment refers to a specific geographic site or address. Some establishments are part of larger, multi-site firms or organizations. The sample of establishments was drawn as part of a topical module on 'Organizations and Work.' included in the 1991 General Social Survey (GSS; see Davis and


Smith, 1996). In 1991, the GSS interviewed a random sample of 1,531 English-speaking U.S. adults. At the end of the interview, each employed respondent was asked to give the name, address, and telephone number of herlhis workplace; married respondents were asked to provide the same information about the workplaces of their employed spouses. This generated a multiplicity sample in which work establishments have known, but unequal, probabilities of inclusion; the probability that an establishment is included in the NOS is proportional to its number of employees. Thus, there are more large establishments in the NOS than would appear if workplaces were to be drawn at random from some listing of establishments. The unweighted NOS sample describes work settings from the standpoint of a typical U.S. employee, since it gives each GSS respondent equal weight. To instead describe the population of U.S. work establishments, the data must be weighted inversely to workplace size. Most figures presented here are for the unweighted sample. 2. Owing largely to the clustering entailed in the area probability design of the GSS, some establishments were sampled more than once. The data reported in this chapter include only one record for such duplicated cases. Including duplicates, there were 1,127 interview attempts and 727 completions. 3. It did not seem sensible to ask organizational informants about practices that they do not enact. The fact that questions about internal staffing practices were posed only for occupations that were ever staffed internally raises the prospect of sample selectivity. The selection criterion-internal filling of a position-is central to definitions of internal labor markets, and we observe that many of the covariates in our empirical analyses below (especially size, formalization, and affiliation with multi-site firms) are also strongly linked to the presence of ILMs (Kalleberg et aI. 1996). Net of the effects of these common covariates (especially establishment size) on selection into our sample and on staffing methods, we believe that any association between a propensity to fill vacancies with internal candidates and the use of staffing procedures is weak, and therefore that any residual selection bias is slight. 4. For incumbent and other references, Goodman and KruskaI's gamma is 0.62; for both types of references and direct approaches it is 0.51. 5. That is, there may be up to three records for each establishment in this Table-<>ne for each occupation (core, GSS, or managerial) for which the establishment's informant answered the sequence of questions about internal staffing. 6. The coefficients can be understood in multiple ways. For an ordinal dependent variable with J categories, ordinal logistic regression models the natural logarithm of the odds that an outcome will be in category j or lower, rather than above j, as '1:J - pX, where '1:j is a 'threshold parameter' or 'cutpoint' and X is a set of independent variables. The log-odds of being in a category above j, rather than j or lower, are therefore -'1:J - pX. The proportional change in the odds of being in a category above j, rather than in category j or a lower one, that is associated with a one-unit increase in a given explanatory variable Xi is the same at any value of j and is equal to expC/3;). Thus, for example, the coefficient of -.573 for public auspices and the use of any informal method indicates that for public-sector establishments, the odds of using any informal method 'frequently' rather than 'sometimes' or 'never' are exp(-.573)=O.56 times as large as they are for private, for-profit establishments. The same proportional factor applies to the odds of using at least one informal method 'frequently' or 'sometimes' rather than 'never.' 7. Because of the clustering of observations on occupations within establishments, we estimated multilevel models including between-establishment random effects on the likelihood of using a given staffing method. The results of these analyses indicated that, net of our covariates, only small and statistically insignificant components of variance are associated with organizational differences. Accordingly, we elected to present findings using a simpler and more easily interpretable single-level approach. 8. Statements on statistical significance refer to the .05 level. 9. The percentages reported here total more than 100% because respondents were permitted to mention more than one source of information when responding to the question. Case bases (Ns) for the percentages range from 238 to 251.

Getting a Job as a Manager

•

ABSTRACT

11 HenkFlap

Ed Boxman

In his seminal study, Granovetter (1995) demonstrated how the job-attainment process is embedded within social networks. The ensuing research effort and theoretical discussion left two points unclear. To what extent do people with higher societal position use informal channels to find a job? Do they receive positive returns in terms of income? A replication of Granovetter's analysis in a large sample of Dutch managers at larger companies (n=1402) in 1986-1987 shows that Dutch managers generally rely on their social contacts to find a job, and they do so more frequently at higher executive levels. Moreover, using informal job-finding methods leaves them with higher earnings. Granovetter' s weak-ties argument has been refuted: although they are the most widely used, finding a job through weak ties does not produce a higher income level. It is not true that managers rely more on informal contacts later in their career.

Our own more general hypotheses on social capital have been confirmed. Managers with more social capital (association memberships and external work contacts) find a job more frequently through some informal channel. Moreover, they earn a higher income independent of their human capital. Burt's hypothesis that social ties enlarge the returns of human capital has been refuted. Human and social capital do interact in that social contacts help workers to earn more income at any level of human capital, but the returns of human capital decrease at higher volumes of social relations.

INTRODUCTION

Many members of the workforce in Western societies have found a job through some kind of informal contact. In the 1980s roughly a third of the employed people in the


Netherlands did so, and in the 1990s as many as half did so (Moerbeek, Flap, and Ultee 1997). Comparable figures for former West Germany and the U.S. in the 1980s are 42 percent and 59 percent respectively (De Graaf and Flap 1988). According to a recent study of the French labor market, 32 percent of all the employees used an informal channel to find their job (Forse 1997), and around 1990 this was also true of nearly half the employees in Spain (Requena 1991). These facts are at odds with the official universal ideologies in these countries.

In contrast to the official ideology, popular opinion holds that if you want to get a job, it helps to mobilize your networks. Networking also gets you a better job. Furthermore, social relations are important, especially at the top of the societal ladder, since the people there have business relations who bring them financial and technical information that can be useful in their current job and also labor market information that can be useful, inter alia, in learning about job-openings. Because of the often influential positions of these contacts, they can be instrumental in helping those at the top of the ladder get similar jobs. However, general social surveys in industrial countries such as the U.S., Germany, France, Spain, and the Netherlands have demonstrated again and again that it was precisely the people who are lower on the ladder who found their jobs through informal channels. In addition, this research has shown that the use of social contacts rarely brings a better job than the use of more formal channels or applying directly.

A number of empirical studies of specific occupational groups do, however, indicate that people with better jobs make extensive use of informal contacts. In his seminal study Getting a Job, Granovetter (1995) noted that over 60 percent of the people in higher positions (he studied a group of persons with technical, professional, or managerial occupation in Newton, Massachusetts, a suburb of Boston) found their jobs via someone they knew. That was also how they found the better jobs. Preisendorfer and Voss (1988) described how almost three-quarters (74 percent) of 200 employees at universities and vocational colleges in the former West Germany found their jobs through some informal channel, although networking did not help them find a better job-for example a steady full-time job.

Findings like these suggest a U-shaped association between a person's social status and the use he or she makes of informal contacts in finding a job. This type of U-shaped association probably remains hidden in random sample surveys of the general population because those surveys usually contain only a small number of people in higher positions. Findings like these also make us wonder about the circumstances under which it is profitable to mobilize contacts to get ajob.

A small research literature has recently emerged on the social networks of managers, their determinants, and the returns of their networks while doing their job as a manager. There is a special interest in the occupational group of managers, since their actions are supposedly more decisive for the fate of enterprises and of the economy at large. Managers are in the business of entreprendre, bringing people and other production factors together (Burt 1992: 274}-that is, they are in the business of networking. One can assume that if they are good at this job, their social networks will capture not only individual social capital but also ftrm-Ievel social-capital-and that the firm will benefit-for example, in terms of better survival chances, larger sales volume, or greater profit. Moreover, these managers will probably also benefit as

Getting a Job as a Manager - 199

individuals and receive a promotion, a higher salary, or other compensation because of the advantages they bring to their frrm.

This research concentrates more on the specific characteristics of managers' networks and less on the use and possible social capital returns of these networks, since data on actual use and outcomes for individual managers-and to an even greater extent on frrms-are hard to come by. Our contribution adds to this research by concentrating on how managers and would-be managers use their own social network in the job-finding process and it deals with the returns of their networks during their careers as managers.

Case studies with anecdotal material (Mintzberg 1973; Kotter 1982) have demonstrated that managers usually have large work-related networks. Managers talk with numerous people at their frrms to keep informed about upcoming business opportunities as well as potential threats. It is also clear that managers execute their agendas by mobilizing their social networks.

Other, more quantitative studies allow for more solid conclusions.l Ibarra (1992, 1997) has demonstrated that women often experience a sex-specific division within their network at work: for friendship and emotional support they turn to their female colleagues, and for work-related advice and other instrumental help they calIon their male colleagues. This probably puts them at a disadvantage if vacancies arise at a higher level, since they then have to compete with men who have more strong-that is, more multiplex-ties to those men who decide about promotions and filling vacancies. The result of these processes is that women who aspire to higher positions within their organizations, come up against a glass ceiling they are unable to break. Brass (1985a) has shown that for women, this sex-specific division is indeed not conducive to being promoted into a managerial position.

Yet there do not seem to be large differences in the networks of managers at different positions or of different sexes (Moore 1992; Burt 1996). Even the kind of work contacts cited by managers as being the most important or most troublesome are stable across kinds of managers. Managers as compared to ordinary employees, however, do have larger and more open networks, with more work colleagues. They also have stronger ties to their alters, but the alters are less strongly linked to each other (Carroll and Teo 1996). The fact that there are clear differences between the networks of managers and employees and no clear differences between the networks of managers themselves suggests that managers (including female ones) are selected, inter alia, for their network characteristics, making them more similar to their colleagues.

Once they have a job as a manager, what are the returns of their network in their job as a manager? Burt (1992) has demonstrated that male managers with a more autonomous position within their networks are promoted earlier, whereas female managers and males who have just entered the higher managerial ranks need a more cohesive network or even a high-prestige sponsor if they are to succeed in getting further promotions. In another study Burt (1996) described how these structural holes (meaning that one's alters are disconnected from each other) contribute to higher returns for executive managers at an investment banking division of a large American financial organization. They receive larger bonus compensation payments since their alters have no alternative for them, but they do have alternatives for their alters.


Boxman, De Graaf, and Rap (1991) have showed that a larger, more diverse external network of Dutch managers at large companies makes for a higher income, quite independent of the managers' human capital and number of subordinates. In a replication, Meyerson (1994) claims to have found that strong ties bring higher compensation for managers at Swedish public firms. Carroll and Teo (1996) have discovered that a greater number of colleagues in a person's network does not generate more income for managers, but it does so for other employees. They account for this by arguing that all managers have a sizable number of colleagues in their core network and that if they did not, they probably would not last long as a manager.

Apart from Granovetter's small study, the job-finding process of managers and the role that is played in job finding by their social networks has not been studied extensively. This state of affairs makes it interesting to inquire more deeply into the job-finding process of managers and the instrumentality of social networks in their getting a job. We test a number of hypotheses (mainly taken from Granovetter's study, as it is the landmark study on getting a job) in a large data set of Dutch managers. We describe the job-finding process of the managers and how frequently they found their current job through informal relations and, more specifically, through weak ties. In addition, we examine whether managers with more social contacts do indeed use these contacts to get their job. For a number of job-outcomes we also analyze whether there are positive returns to using or having a better network, i.e. greater social capital.

In the last section of this chapter, we discuss whether an individual's social network will also provide the firm with social capital.

RESEARCH TRADITIONS ON THE RELATION BETWEEN LABOR MARKETS AND SOCIAL NETWORKS

Within the sociological and economic research literature on labor markets, at least four traditions have paid attention to the influence of social networks on labor-market positions. The theoretical background of the research literature on the networks of managers is similarly heterogeneous, since it has been influenced by human capital theory, status-attainment theory, structural-organizational ideas, and social capital theory.

The status-attainment research studies the extent to which an individual' s chances of upward or downward mobility depend more on his or her achieved characteristics than on ascribed features such as family background. In the course of research, the awareness has grown that the chances of finding a job depend not only on social origin, education, and work experience but also on access to the kind of labor market information that is provided by social networks. The main findings of the general surveys incorporating these ideas are that the status of the contact person contributes to the occupational attainment independently of human capital and that part of the original effect of human capital is in fact to be attributed really to the social resources used in the job-finding process (Lin, Vaughn and EnseI1981).

The structuralist tradition assumes that there are barriers between different parts of the labor market, including cleavages between networks of relevant people. To find a job in more attractive segments, like an internal labor market, individual capacities are often not enough. One also needs social contacts that feed into this segment. According to this idea of social closure, outsiders are excluded if company recruitment


occurs along lines of preexisting social networks linked to the workplace-for example, by way of employee referrals (Grieco 1987; Windolf and Wood 1988). In addition to the social networks being instrumental as a channel of information, institutional and organizational factors thus influence social networks at the work place. Also see Marsden and Gorman (this volume) on internal hiring practices.

The neoclassic economic tradition of labormarket analysis has a number of varieties-human capital theory, job-search theory, and transaction cost theory. Human capital theory does not provide much information on the relation between networks and labor market position. It states that human capital, as indicated by education and work experience, constitutes the best explanation for attained occupational position. Job-search theory does offer a suggestion (Stigler 1961, 1962; for a review of job-search theory, see McKenna 1985; Devine and Kiefer 1991). It assumes that the collection of information on labormarket opportunities entails certain costs, like loss of time, money and other opportunities forgone. Various search strategies have different costs. The people who are the highest achievers are those who are able and willing to invest in the search for information. The job-search theory does not devote much attention to the embeddedness of search processes by employees and employers in the informal social networks of employees and employers (Granovetter 1985). Although the theory starts from the assumption that the use of personal contacts lowers the search costs if people exchange their information with each other (pooling), it does not specify the conditions under which the exchange will occur and with what result. In reality, search processes do not emerge at random in markets of anonymous, disconnected actors, but within the confines of preexisting networks. Interesting enough, job-search theory provides a mechanism that could explain the effects of social networks, but its weakest point is the lack of empirical testing.

A recent addition to this line of thought is transaction cost theory (Williamson 1994). It stresses that most of a [trm's costs are not production costs but transaction costs. They include not only the ex ante costs of the search for attractive exchange partners but also those of arriving at an agreement and the ex post costs of having to enforce an agreement. Networks lower not only the search costs but also those of contracting and enforcement (Granovetter 1985; Raub and Weesie 1990). Since they engender mutual trust, denser networks will make the management of business and labor relations less cumbersome, stabilize cooperation and indirectly improve performance (see, however, Burt and Knez 1995).

The three traditions described above focus on the matching process between people and positions. Neo-classic human capital theory, search cost theory and sociological status-attainment research focus mainly on the role of individuals with different characteristics. Structuralist theories look for the possible effects of industries and organizations. Transaction cost ideas are only just beginning to influence empirical research on the returns of work-related networks in the occupational career.

Network research has accepted suggestions from each of these three traditions in research on networks and their returns in occupational attainment, and elements of each can be recognized in studies on the networks of managers and their returns that were discussed above. Insights from these three traditions come together in Granovetter's Getting a Job. He explicitly links individual and positional factors. He is interested in the latent structure of social networks and its influence on labormarket


behavior. Although according to Granovetter information is mainly a byproduct of other social interactions that come without costs, he assumes that informal search efforts decrease search costs. This is why an employee's and an employer's interests are served if they search through informal channels, especially weak ties, since they provide access to information that is more likely to be new (Granovetter 1973).

Granovetter's ideas have constituted an important source for the recent fourth tradition focused on networks and labormarket chances-social capital theory (Bourdieu 1981; Flap 1988; Burt 1992). According to this theory, social capital produces a better life. People who have access to more people and to people who have more resources (including contacts) and who are more prepared to lend a helping hand are more successful in goal attainment, including in the labormarket. What is different, though, is that Granovetter sees networks mainly as a by-product of other kinds of interaction, while social capital theory suggests that networks are also the result of people investing in each other. Provided that there is a common future, people' invest in particular ties and types of networks to the degree that they are instrumental. In short, social capital is the present value of future help. Social capital theory specifies the strength of the weak-ties argument in that weak ties do give access to people with better second-order resources, but those people usually are less prepared to help.

We now present Granovetter's major hypotheses on the relation between networks, job search, and the attained position, as well as three additional hypotheses that have been the product of social capital theory proper. We specify Granovetter's hypotheses and test them consecutively for our large data set of Dutch managers.

GRANOVETTER'S HYPOTHESES AND THREE OTHER ONES BASED ON SOCIAL CAPITAL THEORY

HI: Search through social networks decreases search costs. People are more apt to find a job and to find a job with less effort if they search along informal lines instead of formal ones, since it enables them to hear about vacancies earlier. In addition to the advantage of timing-an advertisement informs potentially everyone at the same time-informal contacts provide more in-depth information on the particularities of the job in question and everything that goes with it, a kind of information that cannot be found in advertisements. Employees can also make their productivity and other capacities known through their social contacts to potential employers and get an introduction via them. Contact persons feel an obligation not to recommend people they know for bad jobs or refer poorly qualified people to employers who are looking to fill a vacancy. Looking at the process from the other side, employers can lower the costs of recruitment by searching informally for candidates. They get more in-depth information on the true productivity of potential candidates more quickly. This more intensive information prevents disappointment on both sides. Moreover, it saves on the costs of control and training (see Grieco 1987).

H2: Better jobs in terms of income and prestige are found through social contacts. People who enter the labor market want the best job they can get, and they will mobilize their resources accordingly, including their social resources. Since social contacts bring extensive and intensive information and possibly also other kinds of support, better jobs are found using contacts.


H3: Better jobs are found through weak ties. Weak ties-ties that are less intensive and less frequent, like those with acquaintances and colleagues-lead to the best jobs because weak ties can serve as bridges between different social circles (cf. Granovetter 1973). Information travels across these bridges over horizontal and hierarchical cleavages between different occupational groups and levels and leads sooner to people in higher social positions, who are better able to provide the relevant information on a vacancy or help in some other sense.

H4: Employees at later stages of their careers make more use of their informal ties, especially their weak ties, to attain jobs.

This hypothesis builds on the above three hypotheses by adding the auxiliary assumption that people acquire contacts, especially weak ties, on the job and even more so by changing jobs. Under this assumption, it is logical to expect informal channels to playa greater role in the later stages of an employee's career (see also Granovetter 1988: 193).

H5: Employees at higher occupational levels make more use of informal channels in the job-finding process.

Another plausible auxiliary assumption is that people in higher social positions have larger networks and networks containing relatively more weak ties. They consequently make more use of informal ties, especially weaker ones, to find ajob.

H6: Employees with better networks make more use of their contacts in finding a job.

Hypothesis 1 can be specified using the notion of social capital. If having a greater number of contacts with alters who are higher placed is instrumental, then people will employ these resources more frequently in finding a job.

H7: Employees with more social capital will have higher returns of their social capital.

There is no reason to assume the advantages of social relationships stop after a person has acquired a particular job. The basic idea is that informal social relations to relevant others not only help get a job but also help while doing a job, and improve the returns from this job (Han 1996; Flap, Snijders. and Van Winden 1996).2

H8: Employees with more social capital succeed in getting larger returns of their human capital.

This hypothesis was formulated by Burt (1992), who argues that social capital provides opportunities for applying one's human capital and thus promotes the returns of human capital.

DATA AND MEASUREMENTS

Our data set on Dutch managers is unique. Generally, data in the public domain on the recruitment of managers and their ensuing careers are scanty. To our knowledge there is no other sample of this size on the labormarket behavior of managers at many different firms that includes multiple indicators of human and social capital.


The actual research, a survey, was conducted in 1986-1987. A questionnaire was sent to 4000 large companies with fifty or more employees. These 4000 companies were randomly selected from the total population of 8746 large Dutch companies. Packets were sent to the companies, each containing one questionnaire. Above the company's address we printed the job title of a particular manager, for instance: To the Managing Director of General Motors Netherlands.' We included job titles in five categories of the most important decision-makers at Dutch companies: 1) managing director, 2) personnel manager (human resources), 3) commercial manager (sales manager or purchasing manager), 4) manager production/automation, and 5) financial manager. For each category, we mailed 800 letters. The letter of introduction requested that, if the company did not have the job mentioned, the questionnaire be completed by the personnel manager or the commercial manager. Since at some firms, someone in a middle-management position answered the questionnaire, these managers were classified as an additional category. The response rate was 35.2 percent (n=1402). Due to missing information for some respondents on specific variables, the number of respondents used for the analysis in this chapter is 1369.

The nonresponse is rather high. Although we expected managers of larger companies to be overrepresented in our sample because--e.g., smaller companies do not always have a personnel manager-we cannot exactly determine whether this is the case because we have information only on the size of the establishment our respondent works at. However, based on information on the whole sample of companies with more than fiftly employees, we know that 65 percent of them work at companies with more than 100 employees; 55 percent work at establishments with 100 or more employees. This was to be expected, since it is possible to have large companies with small establishments but not the other way round.

The key concepts in our theoretical arguments were operationally defined as follows:

Job-Finding Method. The channel through which the current job was attained is classified in two categories: 1) informal contacts (via relatives, coworkers, employers, acquaintances, or somebody working at a headhunter's office), and 2) formal channels (through advertisements and employment agencies). Of the three categories usually used in labormarket research-formal, informal, and direct application-the third was unfortunately not included. Rightly or wrongly, the reason for this omission was our fear of inviting socially desirable answers on the part of the respondent. Saying you found your job by yourself makes a much better impression than admitting that somebody helped you. Formal and informal means of finding a job together account for seven subcategories: 1) employment agency, 2) newspaper advertisement, 3) information from a relative, 4) information from an acquaintance, 5) information from a work contact, 6) being asked by the employer, and 7) being asked by someone from a headhunter's office.

Strength of Tie with a Contact Person. The strength of tie categories were: 1) strong tie (contact with a relative) and 2) weak tie (contact with a coworker, acquaintance, employer, or a headhunters). Regrettably, the questionnaire did not contain the category 'information from a friend,' so we had to work with the assumption that every relationship with a nonrelative is weak. Although this is


unfortunate, the hypotheses on the differential effects of strong and weak ties can still be tested since family is generally considered to be the strongest of all the existing types of ties.

Human Capital. Human capital was measured by two indicators-years of formal education and work experience. The first indicator simply consists of the total number of years at school. The second consists of the number of years worked after school.

Social Capital. Social capital was proxied by 1) the amount of work contacts with people in other organizations, particularly people with the same level of education, the same position at another company, and with people with many subordinates, 2) the number of association memberships-for instance, Rotary or Lions, professional associations, frequency of attending receptions (in five categories, ranging from no memberships to four or more memberships), and 3) the number of family contacts in managerial positions. The number of work contacts was actually measured on a Mokken-scale (H=0.61; rho=O.79) (Niem611er, Van Schuur, and Stokman 1980). To test hypothesis 7, we used the first two indicators-work contacts and membershipscomputed factor scores on social capital for all the managers, and divided the resulting scale in four groups of comparable size (339 with hardly any social capital, and then 420, 304, and 296 managers with increasing amounts of social capital).

Income. Income was measured in gross annual income in Dutch guilders, with the following categories: 1) less than Fl. 50000, 2) F1. 50000-70000, 3) F1. 70000-100000, 4) Fl. 100000-150000, and 5) more than Fl. 150000. One Dutch florin or guilder (1998) is worth about U.S. $0.50. For the analyses, we used the logarithm (In) of the average category income to create a normal distribution. Except for the function level variable, all other variables are quite normally distributed.

Function Level. Function level was measured by the logarithm (In) of the number of a manager's direct and indirect subordinates, which takes into account differences in the positions of managers in organizations of different sizes. It provides more information than just the job title. For example, a commercial manager at a large multinational is usually much higher functionally than a director at a smaller firm.

Job Satisfaction. Job satisfaction was measured by constructing a scale asking for level of satisfaction with several aspects of the current job (H=0.41; rho=O.86).

Perceived Chances of Mobility. Perceived chances of mobility were measured with a scale consisting of three items (H=O.51; rho=O.66).

Quality of Information on the Labor Market. This variable was established with a scale of items on information on relevant vacancies, working conditions, job contents, and possibilities of doing similar work at other organizations (Mokken-procedure on four items produced a scale with H=O.60; rho=O.75).

Importance of Advertisements. The measurement of importance of advertisements as a source of information on vacancies is straightforward.

THE ROLE OF INFORMAL CONTACTS FOR THE DUTCH MANAGER

Our research shows that personal contacts are a very important instrument for managers and would-be managers not only to get a job, but also to get a better job.

0-03

N

~ o 0

\

-~

-ti

Tab

le 1

. Way

off

mdi

ng c

urre

ntjo

b fo

r D

utch

man

ager

s of

larg

er c

ompa

nies

<::

l i3 M

anag

ing

Per

sonn

el

Com

mer

cial

M

anag

er

Fin

anci

al

Mid

dle

~

Dir

ecto

rs

man

ager

s m

anag

ers

prod

ucti

on/

man

ager

s m

anag

emen

t ~

<")

auto

mat

ion

§:

1. E

mpl

oym

ent b

urea

u 1%

1%

1%

1%

3%

4%

Q

'1::

1

2. R

espo

nd to

an

adve

rtis

emen

t 18

%

28%

34

%

32%

46

%

~.

38%

- ~

in d

aily

/wee

kly

;::

\::>..

t"-<

3. A

dver

tise

men

t in

prof

essi

onal

jour

nal

7%

7%

5%

11%

5%

9"

10

t;. ~ -~.

4. C

ompa

ny-i

ntem

al a

dver

tise

men

t 0%

3

%

1%

0%

1%

0%

5. F

amil

y 12

%

3%

7%

6%

4%

4%

6. A

cqua

inta

nce

3%

7%

4%

4%

5%

4%

7. W

ork

cont

act

6%

9%

4%

6%

4%

4%

8. A

sked

by

empl

oyer

49

%

36%

40

%

40%

27

%

30%

9. A

sked

by

an e

mpl

oym

ent a

genc

y 6%

9"

10

4%

1%

4%

7%

Tot

al p

erce

ntag

e:

100%

10

0%

100%

10

0%

100%

10

0%

N (

1369

) (5

17)

(167

) (2

32)

(278

) (1

04)

(71

)


Table 2. Network characteristics and information on the labor market among Dutch managers of larger companies, 1987

Importance of Information about advertisements for Perceived labormarket information about mobility opportunities the labormarket chances

Number of Tc= .08** Tc=-.l0** Tc= .06* memberships (n = 1394) (n = 1347) (n = 1162)

Number of Tc= .25** Tc=-.l0** Tc= .13** work contacts (n = 1380) (n = 1340) (n = 1149)

Number of Tc= .07** Tc= .OI Tc= .00 family-contacts (n = 1371) (n = 1330) (n=1143)

Kendall's Tau-test: * statistically significant < .05; ** statistically significant < .01

Sixty-two percent of the managers in our study did find their current job through some kind of informal tie, and up to 75 percent of the managing directors found their current job that way (Tables 1 and 4). The numbers are somewhat higher than those found by Granovetter in his research (58 percent), but his figures relate to another country and another time. Moreover, his respondents are not a random sample (Granovetter 1995: 7-10) and he did not inquire into the category 'asked by employer,' which is how, according to our research, most managers (40 percent) in the Netherlands get ajob.

Granovetter (1995 : 19) reports that 14.8 percent of the managers he studied found their job through direct application. The kind of frrms worked in by the managers who were studied by Granovetter is unclear. Our study focuses on managers of firms with fifty employees or more. Four out of ten of our respondents, and half of our managing directors, stated that they were asked personally by their current employers to take the job. Although around one-third of Dutch managers got their jobs through advertisements, advertisements do not seem very important to Dutch managers, as 66 percent of our respondents do not think they are relevant as a source of information (see Tables 1 and 2). In the recent National Organizations Study, 50 percent of all the employers in the U.S. said they used advertisements when they recruit managers from outside (Marsden 1995: 138). Referrals from current employees as well as referrals from business or professional contacts are used less often to recruit new managers from the outside. One difficulty interpreting the figures from that study is that employers were not responding to a question about an actual hiring but about what they typically do when they want to fill a vacancy through external hiring. Moreover, employers often use two or more recruitment methods simultaneously.3

To determine whether social networks lower the costs of search (Hypothesis 1), we looked into the influence of networks on the amount of labormarket information, the perceived importance of advertisements, and the perceived chances of upward mobility. If it is cheaper to acquire information on the labor market through informal channels, then people with more extensive and diverse social networks should have more labormarket information, and they should also perceive more opportunities for mobility. Advertisements would also probably be less important to them. Table 2


shows that people who have more work-related contacts, attend receptions and other social gatherings more frequently, are member of more associations and have more family members in managerial positions do indeed have more information about the labor market. In addition, they also have a lower opinion of advertisements as a source of relevant information on vacancies, and they see better chances for themselves on the labor market. The associations found also make clear why financial managers and middle managers have a greater use for formal channels than informal ones (see Tables 1 and 4); they have fewer contacts at other companies and they attend fewer receptions and the like than do managing directors or commercial managers. Although the indicators are only proxies for search costs, our results nevertheless confirm the hypothesis that networks lower the costs of job search.

The hypothesis on better placement through social networks (Hypothesis 2) has been corroborated by our findings. Table 3 shows a statistically significant4 positive association between informal ways of finding a job and income earned, level of the attained job, number of subordinates, and job satisfaction. If a managerial job is found through some kind of informal contact, this job will be at a higher level, bring a better income, entail responsibility for more subordinates, and give greater job satisfaction. Multiple regression analysis of income on education, work experience and use of personal contacts for finding a job has resulted in the following statistically significantly beta-coefficients: education = .42, work experience = .35, and use of informal contact to find the job = .13 (R2 = .20).

Hypothesis 3 on the strength of weak ties fared less well when confronted with the data. According to Table 3, it does not find support in our data. There is no statistically significant relation between the strength of a tie with the contact person and the income acquired (Tallc = 0.03). Moreover, managers who found their job through a weak tie are also not more satisfied with their job, nor do they have more subordinates or higher-level management positions. Yet, as can be seen from Tables 1 and 4, strong ties are rarely used to find a managerial job. Strong ties are more often used at smaller firms (see Table 4), a fact that is probably related to the fact that small firms are more often family finns.

The test of the career-cycle hypothesis (Hypothesis 4) shows no statistically significant associations between a more advanced career and a greater use of informal ties in general or weak ties in particular. The third panel in Table 4 shows that work experience also does not have a statistically significant effect on the use of informal ties in general nor on the use of weak ties in particular.

Table 3. Information channels used by Dutch managers of larger companies and characteristics of the job

Attained Number of

PQsition Income subordinates

Informal contacts Tc = .21 ** Tc = .15** Tc=.l2**

vs. formal channels (n = 1364) (n = 1364) (n = 1359)

Non-family Tc= .01 Tc= .03 Tc=.OO

vs. family (n= 843) (n = 843) (n= 840)

Kendall's Tau-test: ** statistically significant < .01.

Work

satisfaction

Tc= .09**

(n = 1368)

Tc= .02

(n = 844)


Concerning Granovetter's hypothesis on the effect of the job level on search behavior (Hypothesis 5), more positive findings can be reported. Table 4 demonstrates that managing directors usually make more ample use of personal contacts to find a job than lower-level managers. Personnel managers and commercial managers are special in that, together with managing directors, they have more work-related contacts at other companies and professional associations. For example, 47 percent of the personnel managers are members of a professional association and 44 percent have contacts with a professional recruitment agency. Most of the commercial managers (67 percent) state that they find advertisements unimportant as a source of information because they hear that information from others. The relation we have noted between the executive level and the use of weak ties is contrary to Granovetter's hypothesis. Although weak ties are more important to most managers than strong ones as an avenue for locating a job, they seem to be especially important to people in lower management positions.

We have also examined some special categories of managers. Managers with more human capital in the form of education generally make more use of informal ties. This human capital effect is even more clear in their use of weak ties. No statistically significant differences have emerged between male and female managers in how they acquired their present position. But since women are a very small part (4 percent) of our research group, not much weight can be given to this statistically nonsignificant difference. Of greater importance seem to be our findings on how a job was found by managers in large and in small establishments, combined with what was found on internal and external recruitment. Table 4 shows that the size of the establishment combined with whether recruitment has been internal or external makes a statistically significant difference. Managers who change firms and start at a smaller establishment of another frrm make more use of social contacts than those who are going to work at a smaller establishment of their own frrm. One explanation could be that companies with large establishments have ample relations with small suppliers or organizations servicing them. These contacts probably promote informal mobility among firms (see also the chapter by Higgins and Nohria).

The next question about the job-finding process is whether people with more social relationships have also more often found their present job through an informal channel. The answer to this question is important, not only as a preliminary test of the network-as-resource argument (Hypothesis 6) but also because a positive finding would indicate that our implicit assumption on the causal order of social capital and job outcomes has some truth to it-ties are used to find jobs and attain a decent income, rather than that ties are the result of a job and a particular labormarket position). This validation is needed because the research design was cross-sectional, meaning that the respondents' social capital was not measured before they entered their current job but that their social capital, as well as their human capital and income, were all measured in the same period.

Table 5 shows that managers with an average or large number of work contacts at other frrms have more often reached their current positions via informal channels. The same can be observed of managers with relatively numerous club and association memberships. Although the differences are not very large-the extreme categories differ only by 10 percent-they are statistically significant. This does not hold true for


Table 4. Group-specific differences in employment of social contact in finding a job among Dutch managers oflarger companies

Percentage employing social contacts Percentage non-family

Total external internal total Population recruittnent recruittnent population

(N=1402) (N=776) (N=618) N=1402) Gender

Women 63 62 67 94 Men 62 54 71 88

Formal education

Primary-extended primary 63 63 64 79 Grammar 67 **) 54 **) 79 **) 83 **)

Higher vocational 54 48 64 91 University 71 69 75 94

Work experience

0-10 years 58 50 73 88 11- 15 years 48 36 67 89 16-25 years 62 55 72 90 26-50 years 63 58 70 86

Size of firm or establishment

< 100 employees 64 64 **) 74 84 **) > 100 employees 60 52 70 92

Function

Managing director 74 72 76 84 Personnel manager 62 **) 55**) 69 89 Commercial manager 59 50 69 94 Manager production/automation 57 49 73 92 Financial manager 45 40 60 90

Middle manager 49 49 52 91

** statistical significance of chi2-test <.01.

the association between the number of managers in one's own family and the use of informal job entrance means (this absent association is not shown here). So, we have found some support for the network-as-resource argument. These results reinforce the confidence we have in our indicators of social capital.

The final question on the returns on social capital on the job (Hypothesis 7) and the related one on the relative returns of human compared to those of social capital, is answered in Table 6. This Table presents the average annual incomes of managers in four categories of human and social capital. More social capital clearly produces a higher income. There is also a clear positive income effect of having a better education.

Burt's hypothesis that social capital enlarges the returns of human capital (Hypothesis 8) was refuted. An examination of successive rows of the same Table


Table 5. Proportion of Dutch managers finding a job through social contacts by two indicators of social resources, 1987 (n=1359)

Percent Number of F-test

informal valid cases (significance)

Work contacts with managers

in other organizations 1. none 54.6 240

2. very few 55.4 276

3. moderate 67.9 274

4. many 65.0 546

(1336) 5.57 (.0008)

Memberships in clubs.

professional organizations. etc. 1. none 63.3 128

2. one 54.9 446

3. two 63.8 423

4. three 66.3 252

5. four or more 71.3 87

(\336) 3.80 (.0043)

shows that human capital and social capital interact: social relations help to earn more income at any level of human capital. The returns of social capital are about equal for all the educational categories but, as is shown in the successive columns of Table 6, the returns of human capital in terms of income decrease at higher volumes of social relations. The same pattern emerges, although somewhat less clearly, if human capital is measured by years of work experience: social capital does not multiply the returns of human capital. This is particularly the case for the managers with the most social capital. The difference in mean annual income between managers with primary school education and those with an university education is smallest for the managers with the lar~est social network (for more intricate analyses, see Boxman, de Graaf, and Flap 1991).

CONCLUSIONS AND DISCUSSION

Our questions addressed the extent to which managers employ informal contacts to find their job and whether informal contacts benefit managers in terms of income and the like. Our research among Dutch managers made it clear that managers generally rely on social contacts to find their jobs, especially at the higher executive levels. This finding gives further credence to the conjecture of a U-shaped association between job level and the use of social contact, a conjecture that explains the apparent contradiction between general social surveys that establish a negative association between job level and the use of informal job-finding methods, and our finding that managers, a well-defined group of high-status occupations, mainly get their jobs through social contacts.


Table 6. Average gross yearly income of Dutch managers of larger companies in Dutch guilders (x 1000) by social capital and formal education. 1987 (n=1359)

Social capital

low high

Formal education (I) (2) (3) (4)

1. Primary/low vocational 73 (n=45) 86 (n=41) 88 (n=26) 128 (n=18)

2. Extended primary-grammar 81 (n=llO) 100 (n=107) 104 (n=80) 114 (n=75)

3. Higher vocational 91 (n=14O) 99 (n=206) 104 (n=127) 117 (n=136)

4. University 108 (n=44) 132 (n=66) 132 (n=71) 142 (n=67)

Dutch managers do receive higher earnings if they have used informal job-finding methods. and their work satisfaction is higher as well. Moreover, the ones at higher executive levels have more frequently found their job informally. Contrary to Granovetter's hypothesis, weak ties do not bring better jobs, although they are the most important way managers have found their present job. His ideas that weaker ties are used most widely by higher-level managers and that managers at more advanced stages of their career make more use of informal job finding methods have also been refuted.

The specifications of Granovetter's hypotheses with the help of the social capital theory have held up well in our empirical analysis. Managers equipped with greater number of social ties (more memberships and external work contacts) make more ample use of their social contacts to find a job. In their jobs as managers, these contacts provide them with social capital as they earn a higher income if they have more social ties. Social capital and human capital both contribute to a higher income. Burt's argument that social capital multiply the returns of human capital has, however, been refuted. Human capital and social ties interact in their contribution in a particular way to income: social ties bring the social capital of higher returns at any level of human capital, but for managers with a larger volume of social ties, more human capital does not add to their income. Human capital and social capital are only partly interchangeable as regards earning an income as a manager: social capital can act as a substitute for human capital, but not the other way round. Contrary to the hypothesis forwarded by Burt (1992), human capital is not most valuable for people with the most social contacts but for those who are practically without any social resources at all; it is least valuable for managers with many social resources. Elsewhere we have established that human capital produces social capital, but the effect is not very strong (Boxman, De Graaf, and Flap 1991).

Apart from the confirmation that informal searches lower the costs of collecting labormarket information, there is more direct evidence to support the idea that is at the basis of most of the arguments on networks and labor market chances: managers with larger and better networks are better informed about the labor market and have a lower opinion of the value of advertisements.

A number of our findings call for further comments. The refutation of the weak ties argument stands out, or rather the question why. Of course our operationalization of strong and weak ties may be held responsible for the absence of any positive effects of weak ties in our study. But one should bear in mind that according to other studies, including general social surveys among the workforce, informal ties do not always


lead to better jobs for every group, and weak ties hardly ever do. For a review, see Flap (1991), Breiger (1995), and Granovetter's (1995) afterthoughts in the second edition of his 1974 book.

Probably the answer should be sought in the fact that there are structural and institutional restrictions that have influenced the search process as well as in the search outcomes. Important examples of disturbing factors include the situation of contact persons who act as go-betweens, the search and recruitment behavior of companies that search for candidate managers, and the situation in a particular sector of the labor market. As to the first factor, it is an empirically well-established relationship in studies on networks and social stratification that contact persons in higher social positions bring better returns. So perhaps the determining factor is not the informal tie as such or the strength of a tie but the person at the other end-what he stands for or the resources he or she provides access to. For example, individuals of higher social standing signal higher value: perhaps they are more trustworthy as judges of quality in others, or perhaps they do indeed have the necessary contacts with the employer who does the hiring and is looking for a candidate.

Furthermore, recruitment behavior differs between organizations and sectors of the labor market, depending on tradition or on what is needed on the job. A particular kind of search behavior by a job candidate-for example, an informal search-might have quite different outcomes, especially when a job is such that the person who occupies that job can do a lot of damage, employers recruit informally and preferably through stronger ties, thus enlarging the returns on informal search. For these jobs, they need more subtle information on the quality of a candidate to lower the risks of damage (Flap and Boxman 1996).6 By hiring someone informally, the employer has some assurance that the contact person will somehow vouch for the individual he has recommended and that the newly hired person will get along with sitting personnel. Tacit skills needed at the job will also be more easily transferred to the newcomer if there is some preexisting tie (cf. Grieco 1987; Marsden 1995). This explanation seems appropriate for the job-finding process of managers, especially given that 40 percent of the managers and about 50 percent of the managing directors in our sample were asked to apply for the job by their employer.

The value of informal search, or for that matter the value of a tie is contingent. Employers should recruit informally. If contact persons are involved, the contacts should have relevant information or contacts to still other people who do, and they should be prepared to share that information. Granovetter sometimes gives the impression that news flows along the ties in a network like water in a system of canals (cf. Frenzen and Nakamoto 1993). But contact persons are not always prepared to pass on job information to anyone they know because in doing so they vouch for the person they give this information to.

The career hypothesis has also been refuted. As to why managers at later stages of their career do not make greater use of their social contacts to find a job, it is conceivable that the number of relevant contacts does not increase with the number of years at the job. This can be the case when, for example, the job does not bring with it many contacts, the organization operates in a market segment with fewer contacts, or the employee has been working too long at the same company thereby ossifying his or her network so that fewer persons elsewhere know about the employee's qualities (cf.


Granovetter 1988). Yet another possibility is that for certain higher-level occupations, the specific labor market is small and circumscribed. The people in that market all know each other and the readiness to provide information about a particular vacancy to others might decline if a would-be information provider fancies that job himself (Spector 1973).

Much still remains to be explored in research on the instrumental role of networks in the life of managers. For example, longitudinal studies could establish in greater detail the extent to which a particular network comes with the job or is a prerequisite for getting and doing the job. Another question that might be answered with longitudinal data is whether individuals with more social capital hop faster from one job to another than those with less social capital.

A second item on the research agenda is the existence of organizational and institutional variation, which might make for different returns of social capital. The instrumental value of social relations is contingent partly on the institutional and organizational context and partly on the strategy chosen by the management of the firm. Our analysis has been practically devoid of any institutional context. Elsewhere (Boxman, De Graaf, and Flap 1991) we did check for company size, number of subordinates, and market sector (manufacturing or service), but that did not greatly alter the results presented above. Nevertheless, an important question we only touched on is: how do institutional conditions alter these effects? Bauer and Bertin-Mourot (1991) argue that in France, state-controlled firms and financial institutions often perform more poorly than do private banks because they parachute into top positions managers who are from outside and who lack ties to the people working at the firm or to relevant business partners outside the firm. In fact, specific institutional arrangements, such as appointment procedures and rules of succession, are responsible for this.

Most importantly perhaps, future efforts should concentrate on the mechanism that accounts for the social capital returns on social structure. What is the mechanism? There are many ways for social connections to provide advantages resulting in a higher income and not just at the moment of job-finding. Is it scarce information, for example, about the situation in the particular market the company operates on? Is it learning from the good and bad examples of other managers at other companies? Or should we look at the recruitment process? How do employers recruit managers? Do they 'buy' managers who are rich in social networks? Do they have more trust in the quality of a managers if referrals come from particular others? Do certain networks keep others from opportunistic short-term profit-making (Meyerson 1994)? Do managers indeed experience higher rates of return on their social contacts as a reward for the higher returns they and their contacts bring to the firm? Or are some relations more helpful in doing a job while others are needed to get a better compensation, at least partly independently of how the job is done (cf. Belliveau, O'Reilly, and Wade 1996; Lazega, this volume)?7

All this brings up the question we referred to above-the question of the relationship between individual-level and firm-level social capital. Managers usually use their own networks to promote company goals. D'Aveni (1991) demonstrated that managers with an extensive informal network in the business community can help ward off business failure and bankruptcy. But the agency problem of managers who


act opportunistically and misuse corporate social structure to advance their own goals to the detriment of the goals of the company they work for (cf. Meyerson 1994) makes it clear that individual-level and fIrm-level social capital are not identical. The latter is also not a simple aggregate of the former (see Leenders and Gabbay, this volume; Pennings and Lee, this volume) because the collective good aspect of corporate social capital creates the danger of underinvestment in corporate social networks. Managers sometimes do not share business and other work contacts with each other because they are competing for status, money, or other company bound rewards. Property rights to social capital are unclear, as can be seen when employees are lured away by a competing fInn, and take along their contacts with clients or suppliers. This is done, for example, by account managers at investment banks (cf. Eccles and Crane 1988), or by editors at publishing houses, who take along the authors whose work they have edited in the past (Powell 1984). Many companies have formal and informal rules forcing managers and other employees to share their contacts with the other employees (cf. Lazega, this volume).

It cannot be denied that companies do benefIt from the private part of a manager's network as well. It was recently demonstrated by Podolny and Baron (1997) that, for managers, having a tight core network of persons with whom to consult in the event of personal problems is conducive to promotion. Earlier research has shown that married managers are evaluated by their superiors as being more productive and are given higher functions (Korenman and Neumark 1991). A wife who does not have ajob and can devote more time to supporting her husband is especially advantageous to individual managers (Pfeffer and Ross 1982}-and probably also to some extent to the company he works for.

In a wider sociological context, it is also interesting to contemplate the possibility that the networks of managers help them reproduce the positions of their family group or class. Wright and Cho (1992) demonstrate that the lines of authority are quite permeable, at least as far as friendships between managers and nonmanagers are concerned. These lines are far less divisive than those of property and expertise. Wright and Cho account for this by noting the many opportunities for interaction that are forced on managers and other employees while at the workplace. In what could be called a semiquantitative, comparative study, Bourdieu and De Saint Martin (1978) describe how different factions of the French elite-either connected to state-owned or state-controlled enterprises or to large private fIrms largely controlled by a small number of families-try to preserve and reproduce their status and social capital in the next generation by way of various strategies. For the former, educational credentials plus friendly connections with other highly placed offIcials are their main resource. The latter typically try to hold on to their position by having more children and having them marry children from other families in the private faction of big business.8

We would like to express our gratitude to P. de Graaf and 1. Schmidt for contributions to earlier versions of this chapter. Vedior Personnel and Advice 8.V. at Almere Haven allowed us to collect the data. Please address all correspondence to H. flap. ICSlDepartment of Sociology. Utrecht University. Heidelberglaan 1.3508 TC Utrecht. Email: [email protected].


NOTES

I. A large literature exists on the networks of managers filed under the heading 'interlocking directorates.' With few exceptions, the numerous quantitative studies of interlocking directorates barely touch on the importance of networks of individual managers. Stokrnan, Van der Knoop, and Wasseur (1988) is such an exception. Their study of large corporations in the Netherlands suggests that the pattern of interlocking between large firms and banks is not only the outcome of corporate actors looking after their interests by using their directors and CEOs, but also the outcome of the actions of individual managers, who, looking after their own interests, create and use their own network. 2. As was noted above, our dataset does not include information about direct applications. In his study on the job-finding process of professional, technical, and managerial workers, Granovetter (1995: 19) reports that 14.8 percent of the people in the managerial category (there were only 81 managers in his sample of 282) succeeded in finding their current job through direct application. For a discussion of the difficulties under the heading 'direct application,' see Granovetter (1995: 154-156). There are indications that direct application brings in less income compared to the use of informal channels (see Faase 1980). 3. If we combine informal job-finding methods, human capital, and social capital into one multivariate analysis, all three of them have a statistically significant effect on the income managers attain. 4. Statements about statistical significance refer to the .05 level. 5. Their National Organizations Study in the U.S. showed that traditions within a particular firm seem to be a stronger detenninant of recruitment methods than characteristics of the firm or the occupation (Marsden 1995: 149). 6. There is also the possibility that the strength of a tie is not measured correctly if one equates colleagues with weak ties or for that matter friends with strong ties. In our research on how people got ahead in their occupational careers in former East Germany, people classified 24 percent of their ties to workrnates as strong and 15 percent of their friendship ties as weak (VOlker and flap 1998). 7. These questions lead to a wide range of other questions. For example, what do the networks of managers mean to the goal achievement and performance of their department or of the company as a whole (e.g., O'Aveni 1991)? Are different types of networks needed to accomplish different company goals (Briider! and Preisendorfer 1997)? And is social capital also managerial capital-that is, can it be willfully created and can it be steered (Kanter and Eccles 1992)? For a review of these questions, see flap, Bulder and Volker (1998) and Leenders and Gabbay (this volume). 8. A somewhat similar idea of Bourdieu is that there is a division between different types of jobs. Higher jobs, especially, differ according to whether they belong to the economic domain or the cultural domain. Earnings advantages are likely to be found among people who choose education and occupations in the sector of their origin and in occupations where the criteria for measuring work performance are unclear. Recent research by Hansen (1996) in Norway demonstrates that this indeed the case and is also true for managers in the cultural and in economic domains.

•

The Changing Value of Social Capital in an Expanding Social System: Lawyers in the Chicago Bar, 1975 and 1995

ABSTRACT

12 Rebecca L. Sandefur Edward O. Laumann

John P. Heinz

Social capital is 'some aspect of a social structure' (Coleman 1990: 302) that acts as a resource that individuals may appropriate and use for their own purposes. In this paper, we examine the economic value of ties to local professional elites: specifically, the income returns to Chicago lawyers of contacts among the elite of the Chicago bar. Contact with the elite of the bar represents a channel through which rank and file lawyers may 'tap in' to the social structure of the bar and acquire valuable resources. The information and influence lawyers access through contacts with notables are properties of the corporate organization of the bar, and, as such, are benefits of corporate social capital. We briefly outline a theory of social capital and suggest ways in which elite ties may act as social capital. We then discuss changes in the social organization of the bar and suggest how these changes may affect the value of the social capital represented in elite contacts. We reason that acquaintance with elites will become more valuable because of the relative scarcity of such contacts in larger social systems. In analyses of factors affecting lawyers' incomes, we find evidence consistent with the hypothesis that ties to elite system members are more valuable in a larger system.

INTRODUCTION

In addition to the obvious reasons for hiring a lawyer-i.e., to acquire expertise about the law and its procedures-lawyers are sources of a broader range of information about the community or the business environment, and they help corporate clients span the boundaries between their own organizations and other organizations. Lawyers are not the only agents employed by corporations for the


latter purposes, of course, but lawyers make an important contribution to these functions. When a business hires an attorney or a law firm, it gains access to courtroom skills and legal knowledge, which are aspects of the lawyers' human capital, and it also gains access to the lawyers' social capital. One component of this social capital is lawyers' connections to other lawyers, both within and outside their own firms (cf. Lazega, this volume). Contacts with other lawyers provide useful information about opportunities, contingencies and strategies, and they give access to persons who may be willing to use their influence to benefit the lawyer, whether by dropping the lawyer's name in tones of approval and regard or by intervening more directly on the lawyer's behalf. These connections may benefit the lawyers' clients, as well as the lawyers themselves.

This paper considers the returns that accrue to a particular form of social relationship, ties to local professional elites, held by lawyers in the bar of a major American city. 1 Our analysis builds on the fact that these urban lawyers constitute a large, loosely-bounded social system. The social system of the bar differs from the social system of a large formal organization such as a large corporation in the distinctness of its boundaries; but, the bar is not unlike many large corporations in the contemporary United States in its size and in the diversity of work that is conducted within its boundaries (cf. Burt's (1992) study of a U.S. high technology firm that is 'the size of a small city'). At the same time that the private practice lawyers we consider in this paper are members of the corpus of the bar, they are themselves members of corporate formal organizations: law firms ranging in size from a solo practitioner and her secretary to large firms employing hundreds of lawyers in offices around the country. The social capital we will consider is social capital from sources outside the lawyer's office, but it is social capital indigenous to the social system of the local profession.

Lawyers who possess valuable social capital, providing access to information and ties to influential others, should benefit from this capital in two ways that will be reflected in the rewards that accrue to their work. First, lawyers with superior social connections should have an advantage in attracting clients, since they have the benefit both of access to information about opportunities and of contacts potentially willing to refer business to them. Second, lawyers with superior social capital should be more valuable to their clients. Access to well-placed contacts gives lawyers access to resources useful not only in finding and courting new clients, but also in dealing with the legal matters of existing clients. Successful legal strategy requires not just technical knowledge of the law, but also other salient knowledge, such as knowledge of others' personalities, of others' likely strategies, of others' history of past behavior in similar circumstances. Such extra-legal knowledge aids lawyers in outwitting and outmaneuvering their opponents in the adversarial contests carried out in the texts of legal briefs and in conference rooms and court rooms. Wellconnected lawyers whose associates can pass along such useful extra-legal knowledge should be more valuable to their clients than equally competent lawyers without access to such useful contacts.

Below, we briefly outline a theory of social capital that identifies important differentiations in its forms by examining the mechanism through which it has its effects and the way in which it is acquired. In light of these observations, we suggest

The Changing Value of Social Capital in an Expanding Social System - 219

specific ways in which ties to the elites of one's profession may act as social capital. We investigate whether ties to elites are valuable to lawyers in increasing their incomes net of other characteristics and organizational contexts which might aid them in attracting clients and in being superior lawyers. We then ask whether the value of the social capital of elites ties changes when the social system in which it is embedded changes.

SOCIAL CAPITAL

Our conceptualization of social capital draws from the work of James S. Coleman (1988, 1990). In Coleman's formulation, social capital is appropriable social structure; it is 'some aspect of a social structure' that acts as a resource that individuals may use for their own purposes (Coleman 1990: 203). Such social structure may exist in relatively discrete forms, such as organizations, or in more diffuse forms, such as extended families, communities, or other loosely bounded social systems. Always, it consists of relationships. These relationships may be components of formal organization, such as the relationships of classmate, department head, co-worker, and instructor; or, the relationships that constitute social structure may be defined by other criteria, such as the relationships of neighbor, lover, uncle, co-conspirator, and friend-of-a-friend. These relationships may be characterized by both their structural form and the content that inheres in them; and, aspects of both their form and their content will condition their productivity as social capital.

An individual's stock of social capital consists of the collection and pattern of productive relationships in which she is involved and to which she has direct access, and further of the location and patterning of her associations in larger social space. That is, her social capital is both the contacts she herself holds and the way in which those contacts link her in to other patterns of relations. Social capital thus exists in an amazing multitude of forms; nevertheless, any form of social capital may be productive through one or more of three types of benefits: information, influence and control, and social solidarity. Information benefits arise when social relationships provide access to relevant, timely and trustworthy information of use to the actor in question (Burt 1992; Laumann and Knoke 1987). The influence and control benefits of social ties are obverse sides of one coin: the ability to influence others (Parsons 1963; Coleman 1990) and the ability to be free of others' influence. Solidarity benefits arise when there is some degree of mutual trust and commitment among a group of individuals. In this paper, we confine our discussion to the first two types of benefits.2 Our conception of social capital differs from the definition given by Gabbay and Leenders (this volume), who emphasize a distinction between social ties and social capital. In their view, social capital comprises the beneficial resources actors draw from their social networks, rather than the relationships that constitute those social networks. In our view, the aspects of social structure and social relationships that are potentially valuable through their provision of beneficial resources are forms of social capital; beneficial resources provided by forms of social capital are what we term benefits (Sandefur and Laumann 1998b).


As presented by Coleman (1988, 1990), social capital is defined by its function; thus, any form of social capital has a quality that may be called its valence. That is, forms of social capital which provide one benefit or are useful for one purpose may not provide a different benefit and may be useless or actively harmful for other purposes. This variable characteristic of benefit or liability, of positive or negative 'charge,' may change given the context in which a particular form of social capital is activated and the purpose for which it is employed. Among lawyers, for example, the solo practitioner is free of formal collegial entanglements and so has the opportunity to be a self-determining entrepreneur (Seron 1996). At the same time, however, he is a relative isolate, which means that, while he is relatively free of others' control, he may lack others on whom he can rely when he faces difficulties. Thus, Arnold and Kay (1995) in a study of lawyers' professional misconduct, find that solo practitioners are more likely than lawyers who work in law firms both to he charged with acts of professional misconduct, and to be convicted of them. The misbehaving solo practitioner might have been saved from disgrace if he had been prevented by vigilant colleagues from committing his crime. Post hoc, he might hope to mobilize others to collude to hide his misdeeds or to try to sway the decisions of members of the disciplinary committee considering his disbarment. Thus, social structure that is social capital for some purposes is social liability under other circumstances.

The third characteristic of social capital to note is the variation in its mode of acquisition. Some forms of social capital are acquired relatively passively (such as family connections acquired by birth), while others may be actively sought (such as acquaintance with influential colleagues). The acquisition and cultivation of such key relationships can be part of a deliberate strategy of personal advancement, as well as the unintended by-product of ordinary informal social contact.

The Social Capital of Ties to Professional Elites Lawyers who work in private practice are rewarded with income accruing as a result of evaluation by two audiences, their employers and their clients. Lawyers can increase their incomes by being superior performers and by appearing to be superior performers. Contacts who are knowledgeable about the bar or its clients can be valuable by providing a lawyer with information that helps make her human capital-her talents, abilities, and experiences-more productive (Burt 1992). Contacts who are influential in the bar or with its clients can promote a lawyer's reputation with clients and among other lawyers. Elite lawyers in a city bar can act as such informed and influential contacts.

The elite of the bar are not simply prestigious or respected members of the profession; they are leaders of various constituencies within the bar and liaisons with people and organizations outside it (cf. Laumann and Pappi 1976; Heinz et al. 1982; Heinz et al. 1997). As leaders, they are consulted because they are knowledgeable and their acquaintance is sought because they are influential. A notable lawyer is not just then a prominent member of his local profession, he is also a 'relay point' among his associates. Considered from a network analytic perspective, the notable lawyer's prominence reflects his advantageous structural position as a leader of some segment of the bar or the bar at large-his position at the center of a pattern of


branching relationships. Through his ties to constituents and his relations with other prominent members of the bar, he can synthesize a picture of the bar and its clientele that a less prominent lawyer would not have the information to construct. Further, his status as a leader may give him access to particular, possibly sensitive items of information--either about the larger bar or about matters of interest to those whose work intersects with his own. Thus, contacts with elites can provide a rank and file lawyer with information about the general 'lay of the land' and about specific opportunities.

Acquaintance with a member of the elite can also benefit a lawyer through the influence elites may use on her behalf. Such sponsorship by an elite or 'notable' lawyer may benefit a rank and file lawyer in three ways. First, the judgment by a respected member of the profession that one is a valuable colleague or a key player enhances one's reputation among peers and superiors. Second, the elite lawyer can circulate his favorable evaluation among his other constituents and other prominent lawyers; in such a way, his position in the network of relations in the bar can augment the benefit of his favorable evaluation.3 Thus, one's reputation benefits both from the seal of approval of a respected judge and from the ramifying nature of the elite lawyer's social network. The third way in which an elite lawyer may use his influence to benefit a constituent is by a specific intervention. For example, 'making a call' is a common practice in which elites use both their social connectedness and their influence to benefit their constituents by intervening in another's decision (about, for instance, whom to hire, whom to promote, or how much to reward someone).

Even quite unintentional actions by elites, such as the passing mention of a name to a potential client or a gossipy story about a competitor can benefit a rank and file lawyer. There is no doubt great variation in the intensity and type of elite sponsorship and support. In the analyses to follow, we do not have measures of the specific mechanisms through which the social capital represented in elite contacts has its effects. We simply argue that, on the whole, such contacts should provide access to information and to someone potentially willing to use his influence on one's behalf.

Social Capital and Changes in the Social System By this reasoning, elite contacts would be a form of social capital in a city bar with 1500 lawyers or a bar with 15,000 lawyers. But, we argue, in addition, that the rate of income return on such capital will change as characteristics of the system change. That is, since social capital is an aspect of social structure, as the social structure itself changes, the value of its appropriable components may change. In particular, we hypothesize that the value of elite ties will be greater in larger systems than in smaller ones because the social capital of elite ties is likely to be more scarce in larger systems.

Let us consider elites for the moment as groups with some degree of internal social cohesiveness (rather than a priori designated proportions of the population, such as the top 1 %). For example, consider an elite such as the generally recognized group of respected scholars in some academic field, for instance the top researchers of poverty in sociology or of Anglo-Saxon in English. In addition to being aware of


one another's work, these scholars meet at conferences and on review panels and they correspond about their students, their research and other concerns of their field. Each of them may not personally know all of the others, but each is acquainted with a (probably very high) proportion of her colleagues. Consider a situation in which such more or less cohesive elites are relatively open to communication with rank and file system members. The hypothetical respected scholars exemplify such an elite, as do the leaders of a community or town (Laumann and Pappi 1976) and prominent lawyers in a city bar (Heinz et al. 1982; Heinz et al. 1997).

As a social system expands, such elites are unlikely to grow at a corresponding rate. Two different approaches suggest support for this proposition. Mayhew (1973) presents a formal proof of the proposition that the rate of elite expansion will be slower than the growth of the social system the elites inhabit. Employing Mosca's (1939) definition of a ruling elite as some numerical minority of the popUlation, Mayhew shows that the ruling elite of a social system will comprise a decreasing proportion of the popUlation as the system increases in size. A similar prediction follows from a network analytic perspective. Again, let us assume that a certain degree of social connectedness among members of the elite is necessary for the elite to persist as a relatively defined group. As an elite grows in size, the ability of individual members of the elite to maintain ties of a specified intensity with other members will decrease. Any individual can maintain only a certain number of relationships; as the number of potential interaction partners increases, the average proportion of those potential partners known by individuals will likely decrease. Individuals' 'carrying capacity' for relationships thus places an upper bound on the size of an elite with a specified degree of internal cohesion.

From the perspective of 'ordinary' system members, the question may be framed somewhat differently. In a smaller social system, acquaintance between a rank and file member and a member of the elite is more likely than in a larger one, if only because acquaintance between any two randomly selected members is more likely. Because elites comprise a smaller proportion of the population in larger systems, the chance probability of a tie between a randomly selected member of the population and a member of the elite is lower. To the extent that members of the elite possess special influence or other resources that they may use on behalf of their constituencies, access to such resources is then more restricted (in the sense of being less widespread) in larger systems. Thus, the resources are more scarce and consequently may be more valuable.

Changes in the Social Organization of the Bar and the Value of Elite Contacts Since 1970, the number of lawyers in the United States has roughly doubled; the rate of expansion of the bar has far outstripped growth in the population (Abel 1989: Table 23). This tremendous growth in the number of lawyers has been attended by changes in the way law practice is organized. Lawyers and their work have become increasingly incorporated into organizations such as large law firms. Many lawyers now occupy work roles more similar to that of employees than to the tradition of free-standing professionals. Job placement has become increasingly mediated by law school placement offices and by groups of firms that join forces in their search for qualified students (Abel 1989: 224). Law firms themselves have grown larger, and


have experienced attendant organizational changes that arguably erode the professional autonomy once characteristic of work in the law (Galantar and Palay 1991). Thus, lawyers' opportunities to recruit clients and to increase their own incomes have become increasingly structured by the formal organizations that train, sort, employ and reward them.

Reflecting national trends, the Chicago bar has doubled in size in the past twenty years and law practice has become increasingly concentrated in large organizations. In a random sample of Chicago lawyers taken in 1975, 19% of licensed lawyers were solo practitioners, while a similar survey conducted in 1995 found that solo practitioners constituted only about 13% of lawyers. The twenty years saw no change in the proportion of lawyers working in law firms (67%), but a great increase in the average size of the firms in which lawyers were employed (37 lawyers in 1975, 178 lawyers in 1995). As these firms have grown, they have in many cases formalized procedures for hiring, evaluation, and promotion. In such a context, Chicago lawyers' income attainment chances may have become increasingly 'bureaucratized: in the sense that a greater proportion of lawyers are subject to formalized procedures applied with supposedly less partiality.

One could argue that the increasing formalization of lawyers' income attainment chances should lead to a decrease in the importance of elite contacts and other particularistic factors, in favor of the increasing importance of formal qualifications, such as education. Such would be a classic Weberian argument about bureaucracy (Weber 1946). However, one might also argue that key contacts would remain important or in fact increase in value with the increasing bureaucratization of mobility contests. Even in more formalized settings, the information accessed by ties to elites should still be valuable to lawyers in increasing their incomes by providing notice of opportunities, contingencies or complications. And, clients and employers will continue to encounter situations in which there is some uncertainty about someone's past or future performance that available information 'in the file' cannot resolve. In such cases, elite influence or the 'reference' provided by association with elites can help resolve this ambiguity in favor of the lawyer who counts elites among his acquaintance. Further, since employers (and, likely, some clients) have implemented formal procedures intended to reduce the role of non-productive factors in hiring and reward decisions, they will find themselves faced with several formally equivalent candidates. Here again, elite influence can be a crucial factor in determining who among formally identical lawyers is to receive scarce and valuable opportunities and rewards. In sum, it is not clear that increasing rationalization will lead to a reduction in the value of elite ties.

Below, we present a preliminary test of the two hypotheses that we have outlined above. First, we ask whether, net of more evident productive characteristics, elite contacts are associated with higher income. Second, we ask whether the value of elite ties increases as a social system expands. We consider the specific case of the Chicago bar as a large, loosely-bounded social system in which a group of elite lawyers is identified as leaders of the bar and of various constituencies within it (Heinz et al. 1997). Employing two surveys of the bar, the first conducted in 1975 and the second in 1995, we investigate changes in the value of contacts with elite lawyers that coincide with changes in the social organization of the bar.


DATA AND METHODS

The 1975 study of Chicago lawyers includes 777 randomly selected respondents with offices within the city limits (Heinz et al. 1982). The 1995 survey includes 788 randomly selected lawyers drawn from the same area (Heinz et al. 1997). The response rates for both surveys were just over 82% of the target samples. The lawyers in each survey were asked a series of questions about their career history, their educational and family background, the nature and structure of their current work and work experiences, their social and political attitudes, their social participation, and their relationships with specified other groups of lawyers. Analyses presented below are for private practice lawyers (those working in law firms or as solo practitioners). The sample includes 522 such lawyers in 1975, 35 of whom were dropped from the analyses due to missing data, resulting in a sample of 487 respondents. In 1995, the sample includes 521 such lawyers, 33 of whom were dropped due to missing data, resulting in an analysis sample of 488 respondents.4

In consultation with informants knowledgeable about the Chicago bar, the principal investigators identified lawyers who appeared to be representative of several salient elites. The elites selected were drawn from a large range of social and professional categories: bar association officers, political partisans, type of work and kind of clientele, practice setting, gender, race, and ethnic background (Heinz et al. 1997). Lawyers who were currently in public office were excluded from the identified elites, to avoid conflating official power or influence with prominence among colleagues. We refer to these selected elite lawyers as the 'notables.' The 1975 questionnaire included the names of 49 notables, 6 of whom were dropped from subsequent analyses.5 The 1995 survey included the names of 68 notables, three of whom were dropped from subsequent analyses because they proved to be relatively unknown (for details, see Heinz et al. 1997). The notables of 1995 are more diverse demographically than those included in the 1975 study, as is consistent with the increase in the demographic diversity of the Chicago bar. The 1995 list of notables also includes lawyers who work as the internal counsel of non-legal organizations and those who work for legal aid organizations, while the 1975 list did not.

As part of each survey, respondents in the random sample were given a list of the notables and asked to indicate which notables they knew at two specific levels of association. On their first pass through the list, they were asked to mark those notables with whom they were 'personally acquainted.' Next, they were asked to mark the notables of their acquaintance who they believed would take the time to advise them if asked. Analyses in this paper will focus on the notables nominated at the second, stronger level of connection.

A variety of aspects of lawyers' personal and work characteristics affect their income attainments (d. Hagan and Kay 1995; Abel 1989; Sandefur and Laumann 1998a). In this analysis, we are interested in the relationship between ties to elites and income net of these other characteristics. In our models, therefore, we include measures of respondents' law school quality, achievement in law school, practice setting, practice type, and position in their firm. We include a control for age, which is here a proxy for experience in the law. The relationship between age and income is modeled by a term for age and a term for age-squared. This latter term permits the


positive effects of age (and, by proxy, experience) to decrease over time, and so acts as a statistical control for the obsolescence of skills acquired early in one's professional career. We also include a control for gender; female lawyers have been shown to receive lower pay than male lawyers, on average and net of certain productive characteristics (cf. Hagan and Kay 1995). We are unable to include a measure of hours worked, since this question was not asked of respondents in 1975.

The distribution of ties to notables is quite skewed in both periods, and we have no reason to believe that the relationship between the number of notables known and income is linear; therefore, we constructed dichotomous measures-whether the respondent reported knowing 1,2,3, or 4 or more notables well enough to ask them for advice. In the regression analyses, nomination of 0 notables is the omitted category. Age is the respondent's age calculated from her self-reported year of birth. As a measure of the lawyer's ability (or human capital) as indicated by her law school performance, we include a pair of dummy variables indicating whether the respondent was in the top 10% of her class and/or on the law review of her law school or in the top 11-25% of her class. In the regression analyses, class ranks below the top 25% are the omitted category. Whether the respondent attended an elite law school (e.g., Harvard), a prestigious law school (e.g., Northwestern University), or a local law school (e.g., John Marshall Law School in Chicago) is indicated by a trio of dummy variables; regional law schools (e.g., University of Illinois) are the omitted category in the regression analyses.6 Practice setting is measured by a dummy variable indicating whether or not the respondent is a solo practitioner; effects presented in the regression analyses have as their referent respondents working in law firms.

Practice type is measured slightly differently in the two surveys. In the 1975 survey, respondents were asked what proportion of their income they earned from work with clients who were businesses. Two dichotomous variables indicate whether 25% or less of their income or 75% or more of their income comes from work for businesses (as opposed to non-profit corporations or personal clients). In the regression analyses, the omitted category is lawyers receiving 26-74% of their income from clients who are businesses. In the 1995 survey, respondents were asked what proportion of their clients were businesses; using this information, we constructed a pair of dichotomous variables parallel to the indicators constructed for 1975.7 A lawyer's position in her firm is modeled by a dummy variable indicating whether or not the respondent is a partner. The respondent's gender is modeled by a dummy variable indicating whether or not the respondent is a woman. 1975 income is converted to 1995 dollars using the Consumer Price Index, and its natural log is taken before computation of the regression equations. Modeling the natural log of income reduces the impact of outlying observations (in this case, respondents with very high incomes) on the coefficient estimates, and it permits straightforward comparisons across time. In the regression equations, the estimated metric coefficient of a variable predicting income may be interpreted as the proportionate change in the dependent variable given a one-unit increase in the predictor. The magnitude in dollars of the effect of a change in the value of an independent variable will depend upon the value at which it and the other independent variables in the model are evaluated (Hauser 1980; Stolzenberg 1980).


Table 1. Descriptive statistics for analysis variables

1975 1995

n 487 488

Logged Income (st.dev.) 11.6 (.73) 11.4 (.83)

I Notable Advisor 18% 11%

2 Notable Advisors 14% 13%

3 Notable Advisors 10% 6%

4 + Notable Advisors 27% 17%

Age (st.dev.) 44.7 (14.0) 42.0 (10.4)

Top 10% of ClasslLaw Review 33% 29%

Top 11 - 25% of Class 29% 32%

Elite Law School 23% 14%

Prestigious Law School 16% 14%

Local Law School 45% 43%

Solo Practitioner 27% 20%

Clients 0-25% businesses 31% 24%

Clients 75-100% businesses 47% 58%

Partner 42% 40%

Female 3% 21%

Table 1 presents means and standard deviations for continuous measures and percentages for dichotomous measures used in the regression analyses. Recall that this sample is restricted to private practice lawyers. It is worthwhile to discuss briefly some of the changes in the Chicago bar between 1975 and 1995. The most striking change is in the proportion of women. In 1995, 29% of all and about 21 % of private practice lawyers in Chicago were women, nearly ten times their presence in 1975. (Currently, women are over-represented in government employment and underrepresented in private practice.) Businesses constitute a larger proportion of the client base of private practice lawyers in 1995 than in 1975. Solo practitioners, who serve mostly individuals and small businesses, have declined as a proportion both of lawyers in general (see above) and of lawyers in private practice. The share of Chicago lawyers who graduated from elite schools decreased. This is likely related to two factors. As law firms grew during the 20 years between surveys, they broadened recruitment beyond graduates of top schools. Further, as the number of applicants to law schools increased during the same period, elite schools held their enrollments relatively constant while lower-ranked law schools increased the size of their incoming classes. Thus, changes in both supply and demand are reflected in the educational backgrounds of Chicago lawyers.

Acquaintance with notable lawyers is clearly much more common in 1975 than in 1995. Fifty-three percent (53%) of lawyers know no notables in 1995, while only 31 % of lawyers lack such connections in 1975, even though the list of notables was longer (50% longer) and more inclusive in 1995. As we suggested earlier, ties to elites should be a rarer commodity in a larger system. A different and convenient summary measure of the degree of interconnectedness in the bar is a simple measure


Table 2. Regressions for logged income. 1975 and 1995. Metric coefficient d d h estImates. stan ar errors In parent eses.

1975 1995 Intercept 8.49*** (.37) 7.73*** (.49) 1 Notable -.08 (.08) .03 (.09) 2 Notables .00 (.08) .25** (.09) 3 Notables .02 (.10) .13 (.11) 4 Notables .35*** (.08) .43*** (.08) Age .11*** (.02) .13*** (.02) Age2 -.00*** (.00) -.00*** (.00) Top 10% of Class .08 (.06) .25*** (.07) Top 11-25% of Class .06 (.06) .21*** (.09) Elite Law School .02 (.09) .21* (.09) Prestigious Law School -.09 (.09) .10 (.09) Local Law School -.00 (.08) -.15* (.07) Solo Practice .15 (.08) -.38*** (.09) Clients 25% or less business .01 (.07) .02 ( .09) Clients 75% or more business .18** (.07) .24** (.08) Partner in firm .52*** (.07) .45*** (.08) Female -.21 (.15) -.14* (.07) Adjusted R2 .40 .50

*p < .05. ** P < .01. *** P < .001. Italic-face type indicates a statistically significant difference between the two periods at the level of p < .05. Bold-face type indicates a statistically significant difference between the two periods at the level of p < .10.

of the density of ties between elites and rank and file lawyers. In this case, the density is simply the ratio of the number of observed nominations of elites by rank and file members to the number of possible nominations; as such, it represents the degree to which possible connections have been realized. In the Chicago bar, the density of ties between elite lawyers and the rank and file has declined considerably. In 1975, the sample density of such ties at the level of simple acquaintance was .117; the density at the stronger level of connection was .065. In 1995, the sample density of ties of simple acquaintance was .059, while the density of advisor ties was .029.8

ANALYSIS AND FINDINGS

Table 2 presents results for OLS regressions of logged income in 1975 and 1995. Estimated metric coefficients and their standard errors are presented separately for each model in each year.9 In the body of Table 2, statistically significant differences between the coefficients in the two periods are indicated by differences in typeface; statistical significance within each period is indicated by the familiar asterisk designations.

The findings support our first hypothesis that the social capital represented by ties to elite lawyers would be positively associated with higher incomes net of other characteristics. In both periods, counting four or more notable members of the bar as advisors is positively associated with higher income (p < .001); in 1995, counting two notable members of the bar as advisors is positively associated with higher income (p < .01). The pattern of findings is also consistent with our hypothesis that


elite ties would be more valuable in the larger bar. The income returns to the highest degree of elite connection appears to have increased slightly, but this increase is not statistically significant. The income returns to less extensive connections to the elite have increased to a degree that is statistically significant (p < .05).10 In neither period does 1 tie to a member of Chicago's legal elite evidence an appreciable relationship to income.

To get a sense of the 'real money' value of ties to notable lawyers, let us compare the predicted incomes of similar lawyers with and without ties to local legal elites. In 1975, the predicted income of a thirty year-old male graduate of the elite Harvard Law School, who graduated in the top 10% of his class and worked as an associate in a law firm where his clientele consisted of a roughly equal balance of businesses and persons would be essentially the same whether or not he counted two notable lawyers among his closer acquaintances. In 1995, an identical lawyer would gain roughly $20,000 by acquaintance with two notable members of the bar. To take a different example, consider a fifty year-old man who graduated in the middle of his class from a local law school and operated a solo practice where he served largely personal clients. In 1975, if he counted four or more elite members of the bar among his acquaintance, his predicted income would be 44% higher (in 1995 dollars) than that of an identical lawyer with no notable contacts. In 1995, a similar lawyer gains 58% more income by acquaintance with four or more elite lawyers. The increase in income associated with elite contacts varies in size depending on characteristics of the lawyer and her practice, but the returns to notable ties are considerable and, on the whole, larger in the larger system than in the smaller.

Specific other findings in these models are interesting and worthy of discussion, both as they inform us about changes in the determinants of lawyers' income and as they suggest the workings of other forms of social capital. Earlier, we suggested that work in the law have become increasingly structured by organizations such as law firms and law schools, and that procedures for hiring, promoting and rewarding lawyers have become more formalized. This observation is borne out by inspection of the effects of class rank and law school prestige on lawyers' incomes. Changes in the relationship between legal education and income are striking. When partnership status, client base, law school, practice setting and gender are controlled, class rank has no direct effect on lawyers' incomes in 1975. In 1995, however, class rank is a statistically significant predictor of income, net of all controls (p < .001). The effects of law school prestige are more than twice as large in 1995 as in 1975 (p < .10 for the test of statistically significant difference between the two periods for the effect of elite law school attendance). It appears that income attainments may have become more keyed to the human capital signaled by law school performance and attendance at a well-regarded institution.

Net of age, elite ties, legal education, client type and partnership status there is a negative relationship between solo practice and income in 1995 (p < .001), while the positive coefficient in 1975 does not attain statistical significance. In both periods, solo practitioners have, on average, lower incomes than lawyers in law firms (Sandefur and Laumann 1998a). Part of the deficit experienced by solo practitioners stems from the fact that their clients tend to be individuals and small businesses. The most lucrative of private practice legal work, work for large corporations, is


performed by lawyers in large law firms. Thus, even when solo practitioners have businesses among their clients, these clients tend to be small, locally owned businesses that require neither the volume nor type of work done for large corporations. In addition, solo practitioners must rely on referrals and advertising to attract clients,l1 while firm lawyers have the benefit of association with an organization with an established reputation and of formal organizational contacts with other lawyers who can share their expertise and who may make referrals that benefit them (cf. Lazega, this volume).

Given these general considerations about the different situations faced by solo practitioners and lawyers in firms, there are two possible reasons for the difference between the two periods in the effect of solo practice on income. It may be the case that, in 1975, the best lawyers in Chicago were more evenly distributed between positions in firms and solo practice, while in 1995 the best lawyers were more strongly attracted to work in firms. Our measures of lawyers' ability are restricted to measures of their education, and so do not capture all the attributes that make for a successful lawyer. Thus, stronger selection of talented lawyers into firms in 1995 than in 1975 might account for the different relationships between solo practice and income. An alternative explanation has to do with the increasing concentration in law firms both of opportunities to do lucrative work and for professional advancement. Solo practitioners may be at a greater disadvantage in the larger, more formally organized bar not only because of the greater organizational resources and more lucrative client bases of larger firms, but because the career mobility pathways provided in large firms are now more remunerative than those open to solo practitioners.

Unsurprisingly, partners in law firms make more money than associates in both periods. As we noted above, work for business clients is more lucrative than work for personal clients and non-profit organizations; thus, those lawyers whose practice consists largely of work for businesses have higher incomes. In 1975, the negative effect on income of being a woman does not reach statistical significance (there were very few women in the Chicago bar in 1975). In 1995, the negative effect on income of being female is statistically significant, net of controls for practice setting, client base, education and elite contacts.

Further Considerations The results from our statistical analyses of the relationship between social ties to elites and income in the two periods are consistent with our suggestion that such ties are useful in increasing income and with our hypothesis about the increasing value of elite contacts in expanding social systems. Nevertheless, competing explanations for our findings deserve attention, and we turn now to discussion of three classes of competing explanations.

The first set of alternative explanations has to do with causal ordering. We have discussed our findings as though elite contacts were a 'cause' of income attainments. Of course, the relationship is most likely non-recursive; that is, the causal arrows run in both directions. More successful lawyers are likely more attractive associates for notable lawyers and are perhaps more likely to come into contact with them in the course of their work and work-related socializing. At the same time, acquaintance


with notable lawyers likely affects income attainments in the ways we have discussed, and perhaps in other ways which our theory did not address. Further, it is unlikely that most of the lawyers in our sample became acquainted with the elites they nominated only in the year of the survey; thus, for many (if not most) lawyers, we are investigating the relationship between contemporary income and relationships formed some time before income was measured. It is highly unlikely that the opposing hypothesis about causal order-success in law as measured by income affects the likelihood of elite acquaintance, but the social capital of elite acquaintance has no effect on income-is the truth. Nevertheless, if such were the case, these findings would then tell a story about the social stratification of the bar that would be more about the social distribution of success than about the resources which contribute to success.

The second set of alternative explanations concerns the issue of selection or omitted variables in an analysis. This concern is ubiquitous in social scientific work. Any or all of our findings which we have interpreted as the 'effect' of one factor on another may in fact be the result of selection into certain roles and circumstances on the basis of some other factor we did not model or did not measure. For example, if the most motivated lawyers go to work for firms, while the less motivated lawyers become solo practitioners, then all or part of the effects of firm practice which we have argued are due to organizational factors are in fact due to unmeasured characteristics of the individuals who enter firm practice. Likewise, if the most motivated and consequently successful lawyers are those who are most likely to be associated with members of the elite, then what we interpret as the effects of social capital are simply the social organizational results of individuals' traits. If such selection processes differ in the two periods under study, then effects that we have argued are due to specific changes in the social and formal organization of the bar could in fact be due to other changes in the bar's organization that affect the ways in which particular individuals are selected into particular circumstances. In this paper, we can only appeal to the plausibility of our theory to counter this type of alternative explanation. We note that, if selection were the only operative factor, the findings would remain an interesting descriptive characterization of changes in the social organization of the bar, though our explanation would require reconsideration.

The last set of potential competing explanations for our findings has to do with the identification of the notables themselves. We have operated under two assumptions about the notables identified by the survey directors. First, we have assumed that the notables were correctly identified in each study, or at least that any error in identifying them was not systematic in either period. Our arguments do not require that the notables be an exhaustive or perfect sample, but they do require that they comprise an accurate and reasonably representative characterization of the elite of the Chicago bar. If any error in identifying the notables is systematic-for instance, if the leaders of some sector of the bar are excluded-then our findings may be compromised by the invalidity of the notables as representatives of the bar's elite in one or both periods.

It is difficult to know for certain how the exclusion of some proportion of the bar's elite would affect our findings. However, let us consider the case in which this exclusion did occur in both periods. Some substantial proportion of respondents in


each period then possess elite contacts which we have not measured. In the regression analyses, these respondents who do have the unmeasured advantage of notable ties are included in the reference category against which the effect of notable ties is presented. This should bias the regression coefficients for notable contacts downward in both periods, a bias which is not fatal to our hypothesis about different effect sizes in the two periods.

The second assumption about the notables is specific to the second period under study. While the Chicago bar doubled in size between 1975 and 1995, the number of notables that respondents were permitted to nominate was only increased by an additional 50% or so (65 notables in 1995 versus 43 notables in 1975). We argued above that the elite would not grow at the same rate as the bar; however, even if such is the case, that does not mean that the list of notables for 1995 was increased by a factor corresponding to the true increase in the size of the bar's elite. If the process by which elites were selected in the second period more strongly excluded relatively unimportant prominent lawyers, then our findings could be an artifact of the greater importance of the individuals chosen for the 1995 study elite, rather than a result of the greater importance of elite contacts.

We are very doubtful that the second survey's group of notables results from more stringent skimming of the cream of the bar. Neither group of notables was the result of an 'attempt to create a list of the most notable, successful or influential lawyers in Chicago. Rather, the list[s] include[s] a selection of lawyers, of varying types, who are prominent in one respect or another, but not necessarily more prominent than others' (Heinz et al. 1997: 447). We find no reason to believe that the selection procedures resulted in a systematically more influential and informed group of notables in the second period. The second possibility, that the selection procedure did not result in skimming, but did fail to increase the list of notables by a sufficient amount in the second period, is less cause for concern. If such were the case, we would again have a number of respondents who have the unmeasured benefit of notable ties, but are included in the reference category against which the effects of notables ties are estimated. Such an occurrence would bias the coefficients downward in the second period, in a direction unfavorable to our hypothesis. Thus, if such bias were present, the relationships we report in support of our hypothesis would appear weaker than they actually are.

CONCLUSION

Contact with the elite of the Chicago bar represents a channel through which rank and file lawyers may 'tap in' to the social structure of the bar and acquire valuable resources. The information and influence lawyers access through contacts with notables are properties of the corporate organization of the bar, and, as such, are benefits of corporate social capital. The business firms and individuals who retain a lawyer gain not only from the lawyer's human capital-her skills, abilities, talents, and industry-but also from their social capital, which she may use as a resource in her work for them. The findings of this paper suggest that the social capital of ties to legal elites is associated with valuable income rewards for the lawyers who possess it. Changes in the organization of a social system-in this case, a tremendous increase in the size of the Chicago bar-are linked to changes in the value of given


forms of social capital. Our findings are consistent with our hypothesis that the greater scarcity of contact with elites would lead to increases in the value of elite ties.

As formal organizations have become increasingly salient in structuring Chicago lawyers' access to opportunities and rewards, formal qualifications, such as educational credentials and law school performance, have become more important predictors of lawyers' incomes. Employment in law firms, which provides access to organizational resources, such as office equipment, support staff and colleagues, and to a more lucrative client base, seems to confer greater benefits in increasing income for contemporary lawyers than for lawyers of 20 years ago. Yet, at the same time that the distribution of rewards appears to have become in some ways more 'universalistic' and structured by formal organization, the economic value of ties to specific others, such as those between a rank and file lawyer and legal elites, has not declined and, in fact, appears to have increased. In the contemporary urban bar, the extra-organizational social capital provided by ties to local professional elites remains valuable, both to the lawyers who have such capital, and, by implication, to the firms in which they work and to the clients who pay for their services.

This research was supported in part by grants from the American Bar Foundation and the National Science Foundation (#SBR-9411515). Additional support for Heinz was provided by Northwestern University's Institute for Policy Research. We thank Charles E. Bidwell. Ronald S. Burt. Jeffrey A. Hayes, Ray Reagans. Ross M. Stolzenberg. Christopher B. Swanson. Jeffrey Y. Yasumoto. Ezra Zuckerman. and an anonymous reviewer for helpful comments and useful discussions. This paper is a revision of a section of a paper presented at the Conference in Honor of James S. Coleman. Mannheim. Germany. November 2. 1996. The discussion of the benefits of social capital draws heavily on Sandefur and Laumann (l998b). The views presented in this paper are the authors' alone. and neither the supporting agencies nor supportive colleagues are responsible for any errors.

NOTES

1. The term 'bar' refers to the lawyers licensed to practice in a particular community. not to any particular formal bar association or affiliation of lawyers. 2. For a more thorough discussion of this paradigm for social capital. see Sandefur and Laumann (l998b). 3. Of course. it could also augment the impact of his negative evaluation and so have negative effects on the rank and file lawyer's attainments (cf. Brass and Labianca. this volume). In this paper. we measure relationships with elites which are. on the whole. likely to be of a positive character. 4. In both samples. the largest number of cases was lost due to missing data on income (33 cases in 1975. all cases in 1995). 5. Two notables were never intended to be included in analyses; the investigators knew them to be relatively unknown. and included them on the list only for purposes of gauging a baseline probability of chance acquaintance. One notable died during the collection of the data, and the other three were elderly and relatively unknown to respondents (see Heinz et al. 1982. Ch. 9). 6. The principal investigators coded law schools into categories indicative of their relative prestige and selectivity. For details. see Heinz et al. (1982). 7. Since work for businesses is generally more lucrative than work for personal clients. the incomparability in measurement between the two periods probably results in an underestimation of the increase in business clients as sources of lawyers' income. 8. Sample densities were calculated in the following manner:

Density = observed ties/possible ties Observed ties = total number of nominations from rank and file to notables Possible ties = (sample n)*(number of notables).

Thus. the sample density of 'advisor' ties in 1975 is 1923/(43)*(692) = 1923129756 = .065. where 692 is the number of practicing lawyers (in all practice settings. including private practice. internal counsel and


government in employment) in the 1975 sample. The number 1923 is the number of advisor nominations of notables made by rank and file practicing lawyers in the 1975 sample. 9. The dependent variables are in a common metric, logged 1995 dollars. In comparison of the metric coefficients for each period, the test of statistically significant difference is as follows: [Coefficient(75) - Coefficient(95)]/[std err(75)2 + std err(95)2]112, which is approximately distributed as Student's t. 10. In previous analyses (Sandefur, Laumann, and Heinz 1997), we estimated similar regression equations in which we modeled elite ties with a pair of variables indicating whether the respondent nominated 1-3 notables or 4 or more notables. In these models, the effect of knowing 1-3 notables was statistically significantly larger in 1995 than in 1975. In the present paper, it is likely that the coefficient for 3 notables in 1995 fails to achieve statistical significance because of the small number of observations in that category (29 respondents). 11. In fact, the advertising of lawyers' services became legal only in 1977. Lawyers in the earlier period had to rely on referrals by existing clients, contacts made through membership in local voluntary organizations, such as the Kiwanis or the Knights of Columbus, and the strength of their local reputations as ways of attracting new business.

• SECTION III

Structure at the Individual Level social capital and management

•

Generalized Exchange and Economic Performance: Social Embeddedness of Labor Contracts in a Corporate Law Partnership 13

Emmanuel Lazega

ABSTRACT

This chapter examines the relationship between social structure and economic performance at the intraorganizational level. It attempts to identify a few conditions under which individual social relationships are most productive for the firm in collegial organizations-where the production process is difficult to routinize, where professional expertise and advice cannot easily be standardized, and therefore where internal transaction costs for the firm as a whole can be assumed to be a large part of total costs. An empirical study of a medium-sized northeastern u.s. corporate law firm is used for that purpose. In this firm, attorneys are shown to be bound by a labor contract that is difficult to sustain on pure economic terms: partners can easily free-ride, and associates can threaten the quality of work. Against this damage potential, a social system sustains their commitment. Using network data collected in the firm, social capital is described and measured at the individual, workgroup, and structural levels to show that the more constraining the member's coworkers network, the easier it is for the firm to extract higher economic performance, including from tenured partners, by controlling the time put into work. With regard to partners, such teams represent an element of self-entrapment compensated by status and professional recognition. Examples of low and high economic performers, and their respective combinations of social resources, are provided as illustrations. A locally multiplex generalized exchange system is then described as providing firmlevel social capital. Its existence is viewed as a precondition for individual-level and group-level social structure to be productive because it maintains the circulation of social resources in the firm. A multilevel form of embedded ness is thus revealed here and shows the importance of taking into account a meso level when measuring the relationship between social capital and performance. In this particular case, the


notion of multilevel embeddedness advances our understanding of economic performance: the latter is rooted in individual social capital, that is itself rooted in workgroup and firm social capital, which in tum helps individual members in being more productive. Finally, however, this virtuous circle is shown to be fragile. First, because group-level social capital also threatens the cohesiveness of the firm: wellknitted teams can defect and take away with them valued members and clients. Second, given the dependence of economic performance on firm social capital, the relative contribution of individual ties and collective social structure to economic performance is also a highly politicized issue.

INTRODUCTION: STRUCTURAL SOLUTIONS TO STRUCTURAL PROBLEMS

Current managerial conceptions of corporate social capital often build on basic tenets of transaction-cost economics. They contend that economic growth increasingly depends not only on production costs but also on a critical factor, the so-called transaction costs. Transaction costs arise when market situations produce information asymmetry and therefore need visible signs for building confidence between economic actors. In fact, such a perspective stems from a more general tradition that focuses on social mechanisms supporting and enhancing economic performance, beginning with Durkheim (1893) and by now strongly established (Macaulay 1963; Bourdieu 1980; Coleman 1990; see Flap, Bulder, and VOlker 1998, and Gabbay 1997 for a review). Following this general perspective, social networks of ties within organizations are assumed to enhance performance because they reduce transaction costs by decreasing information asymmetries and opportunism. Thus, maximizing performance means not only improving technology, product and organizational innovation, managerial coordination, or financial management but also changing informal social restrictions, improving solidarity, and investing in social capital. These means introduce a long-term time frame in exchanges and make it possible for people to invest in a starting or an ongoing relationship. In this view, large amounts of social relationships are a basis for efficient cooperation and for solidarity within organizations.

This chapter examines the question of the relationship between performance and social structure at the intra-organizational level. It attempts to identify a few conditions under which individual social relationships are most productive for the firm in collegial organizations (Waters 1989)-where the production process is difficult to routinize, where professional expertise and advice cannot easily be standardized, and therefore where internal transaction costs for the firm as a whole can be assumed to be a large part of total costs. In this organization, social capital is located at different levels: individual level, group level, and firm level. Individual social capital is defined as the resources flowing through an individual's contacts. At the workgroup level, social capital is a function of high density and cohesiveness in strong work relationships. Firm social capital is provided here by a locally multiplex and generalized exchange system-a pattern of ties among members that helps them exchange various resources indirectly, supports cohesive work ties, and maintains a specific form of solidarity.

Generalized Exchange and Economic Performance - 239

To address the issue of the relation between performance and social structure, I look at the effects of these levels on one another. Following other scholars' more general work (Burt 1992; Flap 1990; De Graaf and Flap 1988; Flap and De Graaf 1986; Lin 1982, 1995a; Lin and Durnin 1986; Lin, Ensel, and Vaughn 1981; Campbell, Marsden, and Hurlbert 1986; Marsden and Hurlbert 1988), I do so by reporting an empirical network study of a corporate law firm focusing on intraorganizational performance data. This study shows that one of the conditions under which individual social capital is most productive for the firm is precisely that such a locally multiplex and generalized exchange system constrains some of its members-partners and associates with specific relational patterns and workgroup membership--in reaching higher economic performance (defined in terms of number of hours worked and dollar amounts brought into the firm). In other words, the members' labor contract (the partnership agreement for partners and the employment contract for associates) is combined with other social ties-such as strong collaboration, advice, and friendship, both at the dyadic level and the structural level.

This approach makes particular sense in collegial organizations (Waters 1989, Lazega 1998b). In effect, in the economic conditions of 1990, it can be assumed that when performance of partners and associates is good, firm performance is also good. However, in this firm, getting practically tenured partners to work well is sometimes a problem: there are enormous incentives to free-ride. Getting associates to work well is also a problem: although they are well paid, there is a very low chance for them to become partners. If partners can free-ride and associates threaten the quality of work, members have a labor contract that is difficult to sustain on a purely economic basis. Because, as Durkheim (1893) pointed out, a contract is always incomplete, members need the expectation that this contract will be fulfilled, andin case this exchange does not work-the flexibility that multiplexity provides to enforce these contracts. Getting cooperation and keeping production going are also a result of exchanging these resources in a mUltiplex way. Therefore, a generalized exchange system can be assumed to relate the three levels of capital to one another by maintaining the circulation of social resources in the firm. Several steps are taken to show that this exchange system, a component of this firm's corporate social capital, increases the efficiency of individual performance and collective action.

First, I show that a social system sustains members' commitment to their labor contract: members who are strongly socially integrated perform well economically; others perform less well. On the one hand, formal dimensions of structure, in particular hierarchical status and seniority, have the greatest influence on economic performance. In general, partners put in less hours but collect more dollars than associates because they charge more; in this firm, the more senior attorneys are, the higher their hourly fees. Associates collect less although they put in more time than partners. On the other hand, however, the nature of the individual social network also has an influence on economic performance: attorneys informally sought out for advice and for collaboration by many others tend to bill and collect considerably more than others. In addition, when social capital is proxied in terms of Burt's (1982) constraint scores, results confirm that members (both partners and associates) with a constraining coworkers network are pressured to put in more time, thus


collecting more dollars. The more constraining one's coworkers network, the higher one's economic performance. Thus, position in relational structure and social capital accumulated in this position do count for explaining performance, although these effects are weaker when compared to the weight of institutionally defined hourly rates.

Second, the effect of individual-level social capital on economic performance is decomposed at the dyadic level by looking at specific combinations of ties that provide a decisive push in performance increase (or that represent a liability that decreases performance). For example, specific configurations of social ties, such as mutual triplex ties, are strongly correlated with high performance. This fleshes out the positive effect of constraint scores in the coworkers network. Examples of partners with low and high economic performance (in terms of dollars brought in), and their respective combinations of social resources, are provided as illustrations. High performers draw heavily on their social resources. In that respect, the firm as a whole benefits from the networks that some individuals have (dense multiplex networks, especially constrained at the group level in the coworkers network) and suffers from the networks that other individuals have (sparser networks and especially weakly constrained in the coworkers network).

Third, strong and stimulating (that is, constrained) coworkers ties combined with advice ties or with both advice and friendship ties have a chance of being economically more productive than other ties and more likely to happen in dense workgroups. At the structural level, a locally mUltiplex and generalized exchange system encouraging the emergence of workgroups is then described as a precondition for individual social network to be productive because it maintains the circulation of social resources in the firm. This structure integrates work and social ties in a way allowing strongly knitted positions to perform better by extracting and facilitating higher efforts from their members. By doing so, it makes it possible to improve performance for members whose work ties are embedded in a way offering access to advice (which, in this firm, is a form of free collaboration), creating economies of time, providing flexibility in exchanges by allowing a resource of one type to be engaged for a resource of another type, and helping in foregoing immediate self-gains for a smoother and longer-term collective action. This social structure, to put it in Uzzi's (1997a) words, governs the intervening processes that regulate performance outcomes, both positive and negative. This system is a feature of the firm as a whole: it is a component of its corporate social capital. It can thus be asserted that if economic performance is rooted in individual social capital, the latter is itself rooted in collective social capital or firm social capital.

In sum, firm social capital, that requires many kinds of contributions, is also key to maintaining collective action and production. It is made more visible in the description of the firm as a locally multiplex generalized exchange system. Generalized exchange is strong at the level of workgroups. In particular, cycles characterizing local (that is, group-level) and multiplex generalized exchange can be found in this system, especially in the flows of resources among teams within the same office and specialty. This form of embeddedness is shown to relate three levels of social structure and help enforce the labor contract between partners and associates. It makes them more productive by relying on chains of mutual


obligations and debts. In particular, partners are especially well positioned to play on resource dependencies to get associates' commitment to their labor contract (which is not necessarily in their narrow and short-term self-interest).

I then argue that, in many ways, this exchange system is good for both the firm and the individual. Individual social ties help the individual perform and this network transforms itself into firm-level social capital because it produces an exchange system that makes the firm as a whole more successful in billing and in helping members maintain their commitment to their labor contracts and solidarity. In other words, the firm has found in this exchange system a structural solution to the structural problem of cooperation and commitment to the labor contract, a sort of partial equilibrium in the circulation of resources needed to fulfill it. I By allowing such a system to exist, this firm maintains certain forms of resource circulation, a precondition for group solidarity.2

However, this virtuous circle is also fragile. First, group-level social ties also threaten the cohesiveness of the firm: well-knitted teams can defect and take away with them valued members and clients (Lazega 1992a, 1999). Second, the relative contribution of individual ties and collective social structure to economic performance can be a highly political issue. Allocation of credit in teams (that is, clearly disentangling members' contributions) is often difficult. This study therefore confirms that looking at how an organization, especially a collegial one, extracts economic performance from its members requires a sociological conception of performance that must pay attention to the micropolitical context of members' action and to their strategic behavior in this context (Weber 1978 edition; Coleman 1990; Granovetter 1985; Hechter 1987; Lindenberg 1990, 1996; Raub and Weesie 1990; White 1981). This approach therefore entails a micropolitical conception of economic performance. Narrow conceptions of performance ignore the fact that-as Crozier and Friedberg (1977), Friedberg (1993), or Meyer (1994) put it-no measurement of performance in organizations ever goes unchallenged. To some extent, criteria used to measure efficiency of actors are negotiated by members themselves. This negotiation means that measurements of efficiency are strategic and politicized. In effect, they are always multidimensional; research in this area is complex, and multivariate approaches rarely conclusive. Practitioners know that it is impossible to find simple measures of performance for organizations with mUltiple and often conflicting goals. Meyer (1994), for example, shows that performance measures can be considered to be temporary constraints to which members of the organization adjust. It is thus impossible to define absolute measurements of performance, outside of a strategic context or institutional conventions. In the context of the law firm examined here, team work, autonomy and flexibility in selection of coworkers, and a weak hierarchy unable to force partners to cooperate all make it harder to provide one best measurement of performance by individual employees and workteams. This is why a study of the relationship between generalized exchange and economic performance necessarily leads to issues of fairness and to an examination of the policitized nature of performance measurements.

In conclusion, the forms of embeddedness discussed in this chapter help members perform and reach a form of solidarity, but do not necessarily produce, by


themselves, a self-sustaining social order. The structural solution provided by this exchange system raises new problems related to the balance of power within the fIrm (Lazega 1992a, 1999): it is a politicized issue, all the more easily politicized, because members can make their relative contribution to fIrm performance unmeasurable-for example by bringing more resources into the exchanges-which is made possible precisely by the existence of a multiplex and generalized exchange system. This topic of politicization of performance measurements, however, is beyond the scope of this chapter.

PROFESSIONAL LABOR CONTRACTS IN A CORPORATE LAW PARTNERSHIP

Fieldwork was conducted in a northeastern corporate law fIrm (71 lawyers in three offIces located in three different cities, comprising 36 partners and 35 associates), in 1991. All the lawyers in the fIrm were interviewed. In Nelson's (1988) terminology, this fIrm has characteristics of both a traditional and a bureaucratic law fIrm. 3 It is closer to the traditional type, especially because the tensions between bureaucracy and partners' participation in management are open and explicit and because it does not have formally defIned departments.

The fIrm services mainly corporations and other organizations prepared to bear its fees. Interdependence among attorneys working together on a fIle may be strong for a few weeks and then weak for months. As a client-oriented, knowledgeintensive organization, it tries to protect its human capital and social resources, such as its network of clients, through the usual policies of commingling partners' assets (clients, experience, innovations) (Gilson and Mnookin 1985) and the maintenance of an ideology of collegiality. Informal networks of collaboration, advice, and friendship (socializing outside), are key to the integration of the fIrm, that has to deal with many centrifugal forces created by differences in status, geographical location, and specialties (Lazega 1992a, 1992b, 1999).

The partnership agreement applies to the various aspects of a fIrm's life the prevailing fIrm philosophy regarding legal practice and how it should be undertaken. In doing so, it brings an element of predictability to fIrm operations and minimizes the room for disputes regarding issues such as fIrm management, compensation decisions, and withdrawal terms. The agreement accomplishes this by setting clear ground rules as to each partner's rights and responsibilities in connection with these issues and with the operation of the fIrm itself (Rowley and Rowley 1916 and 1960).

The law fIrm is a relatively decentralized organization that grew out of a merger, but it does not have formal and acknowledged distinctions between profIt centers. Although not departmentalized, the fIrm breaks down into two general areas of practice: the corporate litigation area (half the lawyers of the fIrm) and the more strictly speaking corporate area (anything other than corporate litigation). Work is supposed to be channelled to associates through a specifIc committee of partners, but this rule is only partly respected. Sharing work and cross-selling among partners is done mostly on an informal basis. Given the classical stratifIcation of such fIrms, work is supposed to be channelled to associates through specifIc partners, but this rule is only partly respected.


A weak administration provides information but does not have many formal rules to enforce. The firm has an executive committee made of a managing partner and two deputy managing partners who are elected each year, renewable once, among partners prepared to perform administrative tasks and temporarily transfer some of their clients to other partners. This structure was adopted during the 1980s for more efficient day-to-day management and decision making. The current managing partner is not a rainmaker who brings in important clients and does not concentrate strong powers in his hands. He is a day-to-day manager who makes recommendations to functional standing committees (finance, associate, marketing, recruitment, and so on) and to the partnership as a whole during partnership meetings.

This specific law firm is very much a stratified organization, in spite of a set of rules that tries to smooth the hierarchical nature of its business. Consistent with Nelson's (1988) more general terminology, its authority system is based on a distinction between finders, minders, and grinders. With a few exceptions, the finders are partners who find new and lucrative clients and bear the greatest responsibility for them. Their governing authority is not as formal as that of their analogues in corporations. Directives are reached by a form of gentlemen's agreement. The minders are partners with managerial roles and responsibility for long-established clients. The managerial role in large firms arises from the necessity of coordinating diverse practice areas, promoting an efficient organization of work, and decentralizing control over a large professional staff working on highly specialized matters. The grinders are other lawyers~ither partners who function as little more than salaried staff or associates-who are subject to the demands of partners and perform the actual legal work.

Partners' compensation is based exclusively on a seniority lockstep system without any direct link between contribution and returns. The firm goes to great lengths-when selecting associates to become partners-to take as few risks as possible that they will not contribute enough to firm's revenue or pull their weight. Partners may argue informally about what contribution might fairly match one's benefits, but the seniority system mechanically distributes the benefits to each once a year. Great managerial resources are devoted to measurement of each partner's performance (time sheets, billing, collecting, expenses, and so on), and this information is available to the whole partnership.

A low performance cannot be hidden for long. However, such firms usually make considerable profits, which may help partners overlook the fact that some voluntary contributions to shared benefits may not always be consistent with the successful pursuit of long term self-interest:

Our compensation system has no built-in peer review process. There is no committee meeting with each partner, no interview devoted to pulling out from that individual his or her state of affairs. The peer review that we have right now is everyone sits down in the partners' meeting, and you have in front of you the printout that shows how many hours I worked, how many hours I billed, how many hours I collected, and how outstanding my account receivable is, and then you get people grumbling at the meeting about the account receivable going up and not coming down. (The managing partner at the time of the study)


The firm does not have a formal peer review system that could provide intermediate steps between informal control and formal court procedures:

With the compensation system there is no built-in financial incentive for people to do things. If you have people who are motivated by other things, like self-respect, pride in craftsmanship, intellectual curiosity, competitiveness-whatever those different personal attributes ar~hat's not a problem. There are people who aren't as motivated by those other things as certain other people and may wind up resting on their laurels, sitting on their hands, whatever euphemism you want to come up with for becoming lazy both intellectually and how much they are willing to work. (The managing partner at the time of the study)

Before expulsion, partners have the power to punish each other seriously by preventing one of their own from reaching the next seniority level in the compensation system. As mentioned above, a partner can be expelled only if there is near-unanimity against him or her. Buying out a partner is very difficult and costly:

The rule is 90 percent affirmative votes of all partners to expell a partner. Abstention counts against expulsion. We have expelled only two partners in twenty years. Both were extremely serious situations. For one of them, there was a unanimous vote: this person went way around the bend with a number of things. The second guy didn't want to be part of us. A petition was circulated, saying 'I would vote to expell him.' He was presented with that and resigned. But expUlsion is extremely hard to exercise. Three persons can block it. At least two of our partners are good examples of that. One of them is a decent guy on a personal level. He says that under pressure he does things but that he is a corporate lawyer, and there is no pressure on him usually, so he just doesn't do it. The other partner is in a different situation. He was at the low end (of the performance scale) for a long time. Partner 5 went to talk with him. He claims that there isn't anything to do. He says: 'I am a corporate lawyer, my kind of work has dried up, 1 am out there in the bushes, hustling, doing everything 1 can.' That is difficult to check. But there is also the fact that part of the people he has worked with do not want to do it again because they think his competence is in doubt. Partner 17 is extremely good in his field. He once volunteered to go see this partner to share some work with him. He went to see him, but he says this partner did a horrendously poor job. So with our 90 percent rule, we don't cover for that. Apart from these examples, all the other partners do carry their weight. Of course, you're hot this year, you'll be down next year. But there is no need for a compensation committee that would just do what it wants to do, with all the subjectivity involved. (The managing partner at the time of the study)

It should also be mentioned that, since partners cannot be forced out unless there is near-unanimity against them, most partners in this firm manage to have at least one safety partner-a friend who would presumably side with them unconditionally and become their insurance policy against this consequence (Lazega 1992a, 1995b; Lazega and Lebeaux 1995; Lazega and Krackhardt 1998). Therefore, despite the existence of direct financial controls, the firm does not have many formal ways of dealing with free-loading. The harm that a single partner can inflict on others might become very substantial in the long run. Conversely, partners can try to isolate one of their own informally by, at the very least, not referring clients, not lending associates, and not providing information and advice. This strengthens the suggestion that performance depends on the social circulation of resources in the firm.


A few general indications about the relational climate and structural tendencies in this firm are needed here to understand the way in which the notion of embeddedness of labor contracts is used below. The climate and tendencies can be summarized by two separable forms of interdependence describing the interlocking of social relations such as being strong coworkers, advice relations, and socializing outside work. One of these forms concerns the interplay of friendship and advice, wherein an individual's friends may be a source of further friendship and advice ties (the friends of friends are often friends, and the advisors of one's friends are often friends or advisors, or possibly both). Advice ties, that will be shown to be key to economic performance, are thus often driven by personalized ties. The second form concerns the interdependence of advice and coworker ties, with the possibility that advice ties play an important role in generating further advice ties and, in conjunction with coworker ties, further coworker ties. Finally there is a weaker association between coworker and friendship ties in that the configurations comprising two friendship and one coworker tie are unlikely. There is a taboo with strongly personalizing work ties, particularly for partners with regard to associates who might leave the firm before coming up for partnership, as well as with associates who will eventually come up for partnership. Partners often feel that such relationships would tie their hands on the day of the vote. Associates know that they are being kept at arm's-length and express it more openly. Associates 58 and 51 express this in the following ways:

I am a big believer in keeping your job and personal life separate. You have to have your job to fall back on if you have a personal problem in your life. I would respect that privacy. I don't think friendship could matter here. I think the partners share their personal life with other partners, and associates with associates. But I don't believe that partners and associates share their personal life. It is probably also the associates' fault. Their goal is to make partner, and you don't want to show your vulnerability. You think it will affect your chances to become a partner--especially during the first four years, when you are not secure about your job, and your knowledge is limited. (Associate 58)

An interesting change in relationships comes when some people who were senior associates two years ago are now partners, and we don't spend as much time together outside work any longer. Since they will have to make a decision concerning us, we don' t socialize anymore. But I still consider them to be friends because I feel close to them. (Associate 51)

Associates also feel that partners don' t let them in more generally:

Between partners and associates, there is a gulf-regardless of how friendly we are. As an associate, I am left in doubt. I know that some friends I have were not made partners. There have been people who have gone all the way to the eight years and were not made partners. I will feel less constrained to talk about finn matters with other associates. There are a few partners, however, to whom I would talk about sensitive finn matters. I would actually never bring up the subject myself. But when they bring up the subject, I listen. It is difficult to realize how associates grapple and work just to be able to understand where they are, the kind of finn they really are in. (Associate 48)


These networks are thus partly dependent on each other, even though they each do have a life of their own, mainly because they each solve different problems of cooperation (Lazega 1992a; Lazega and Pattison 1998).

WHAT MAKES TENURED PARTNERS WORK?

In any organization, measurements of performance are intrinsically difficult to interpret and their informative value can change from one year to another.4 As seen above, performance data are never as hard and undisputable as one often expects them to be. In fact managers know and learn that such data must be handled with great care. A narrow conception of organizational efficiency ignores the fact that no measurement of an actor's performance goes unchallenged within the organization. Therefore, using performance measurements is not easy and rarely provides spectacular results. In spite of correlation between some forms of human capital (for example, years with the firm) and performance, there is no easy and clear-cut correlation between performance and social networks.

Members of this firm know that measuring economic performance describes only one aspect of contribution to collective action. This is why managing partners often emphasize spreading credit around generously. Many factors account for members' individual performance. These factors can be external or environmental (some areas of practice provide more work, some markets are currently more lucrative) and individual (some attorneys are personally more motivated or hard working). For these reasons, the following analyses link information on members' economic performance-narrowly understood as the amounts of dollars brought into the firm at the end of the year-with information on social status and relations among them. Differences in such performance may be explained, in part, in terms of relationships within the firm-for instance, because relational factors can help gain access to needed resources, reduce transaction costs with coworkers, or help pressure colleagues back into more productive behavior. To examine such effects on performance, I used information collected in early 1991 about each attorney's (partner and associate) relationships within the firm and combined it with information on their individual performance for the year before the study was conducted.

Variables and Analysis Specifically, to study the effect of position in firm structure on this type of economic performance, I look at the correlations between measurement of economic performance and various social factors related to firm structure, work process, and members' ties in 1991. I first use as covariates three dimensions of formal structure of this firm that were expected to be the most important (status, office, specialty), as well as two attributes of members defined from outside the firm (gender and law school attended). Table 1 presents the distribution of lawyers in this firm per variable.


Table 1. Distribution of lawyers per variable

Status

Seniority

Levell

Level 2

Level 3 Level 4 Level 5

Office

I (Boston)

II (Hartford)

III (Providence)

Specialty

Litigation

Corporate

Gender

Man

Woman

Law school

Ivy League

New-England non-Ivy League

Other

Total

Partner

14 13 9

22 13

20 16

33 3

12 11

13 36

Associate Total

7 10 5 7 6

26 48 6 19 3 4

21 41 14 30

20 53 15 18

3 15 17 28 15 28 35 71

The first covariate is status, a variable with two levels-partners and associates. We can hypothesize that status matters for economic performance in the sense that firm rules require associates to put in more time than partners. This variable is elaborated on in the second covariate-seniority. We can hypothesize that seniority matters for economic performance in the sense that the more senior members are, the higher the hourly rates systematically charged to clients. This second covariate is a variable with eight levels, indicating the three possible levels of seniority for a partner,S and five levels of seniority of associates. For associates, seniority has the meaning of being member of a cohort recruited the same year. We can thus look at gradual effects of numerical rank on economic performance. Office membership and practice are the third and fourth covariates. Office is a variable with three levelsOffice I, II, and III; practice has two levels-litigation and corporate. They are expected to have an effect on economic performance as indicators of variations in market demand. The next covariates are other actors' attributes-gender and lawschool attended. These attributes are included as control variables representing two characteristics of the outside world that could have an influence on economic performance. In this firm, women attorneys are mostly associates and often feel that they need to work harder than their male colleagues to reach the same economic results-for example, because they mostly have to deal with male clients or partners. Law school attended, a typical human capital variable, has three levels, indicating whether a lawyer went to an Ivy League law school, to a New England non-Ivy


League law school, or to another law school. This variable is introduced in the model to look at the extent to which a form of prestige acquired outside the firm may have an effect on performance-for example, via more lucrative clients.

To locate members in the informal structure of the firm and proxy their individual social capital, I use six variables based on standard sociometric information on three types of relations collected in this firm in 1991: coworkers, advice, and friendship (for name generators, see Appendix A). From these data, I derived proxies of individual social capital by looking at two types of measurements. The first was their individual indegree centrality scores in these networks.6 Indegree centrality represents a measurement of the extent to which members are popular in these networks and therefore accumulate work-related resources circulating in them (Wasserman and Faust 1994: 169-219). One can therefore hypothesize that they will be in a better position to perform economically.

The second was their individual constraint scores as defined by Burt (1992) in the same networks. For Burt (1992), network constraint measures social capital as a form of network structure. Specifically, constraint is a function of network size, density, and hierarchy (that measures the extent to which relations are directly or indirectly concentrated in a single contact). A contact in which relations are concentrated is a knot in the network, making it difficult for negotiations to proceed independently in separate relationships. Constrained networks leave little opportunity for individual initiative and little chance to withdraw from difficult relationships. Difficult relations persist because they are interlocked with cooperative relations. The higher the constraint, the fewer opportunities for alternatives offered by one's contacts or contacts' contacts, and the lower the performance. In our case, constraint represents a measurement of the extent to which colleagues can exercise unobtrusive but insistant pressure on a member. High constraint in a specific network means that clique members in that network have high investments in each other and high expectations from each other. The denser a member' s personal network of coworkers, the more his coworkers can coordinate their informal efforts at prodding him or her back into performing more. They can, for instance, try to increase their own collaborations with him or her, and exercise unobtrusive but insistant pressure to put in more time. One can therefore make the hypothesis that, in the case of this firm, high constraint facilitates high economic performance.7

In other words, in this type of collegial organization, a constraining network of strong ties, in which members overinvest, tends to create teams of partners and associates who rely on each other, at least for work. With regard to partners, such teams represent an element of self-entrapment compensated by status recognition: teams allow partners to create emulation and redefine what is an acceptable performance within their workgroup, including themselves. But coordination is also made easier in such workgroups: they bring associates together productively. The following is an example of how partners in this firm typically like to talk about their associates and the kind of intellectually challenging attitude they encourage within their teams:


Ours is a fascinating structure built on, to some extent, maximizing a certain type of efficiency. All are encouraged to think hard. You look good as an associate if you can convince a partner that he is wrong about something. You have the freedom of thought within legal problems. There is a great intellectual freedom here. An associate yesterday told me that she didn't think that a decision of mine in a file was correct. She stuck to her guns, and fifteen minutes later I called her to tell her that she was right. In other places, if the boss says something, everone says 'good idea, boss.' Not here. (Partner 19)

Thus proxies of social capital add a set of covariates to the baseline model. The effect of centrality and of constraining ties on economic performance is expected to be positive.

Using these covariates, several models were estimated to explain economic performance measured as the amount of dollar fees brought to the firm (managing partner not included) in 1990. It is important to realize that not all the covariates representing various dimensions of position in firm structure can be used at the same time because of strong dependency between them. This is typically the case for status and seniority; in the models, the most refined covariate, seniority, is used. In addition, status and seniority overlap with the number of hours worked and hourly rates as explanatory variables. The more senior attorneys charge more per hour. Associates work longer hours than partners.s Therefore, to avoid this problem, analyses below test the robustness of social capital effects using three different models. This multicollinearity will be taken into account in the interpretation of results. In terms of economic and relational variables, the best overall models achievable with this data set predicting the number of hours worked and the amount of dollar fees brought in are presented in Table 2.

Coworkers' Constraints and Economic Performance As predicted, model 1 contains an important effect confirming our expectations. It shows that, in this firm, constraint by teams of coworkers has a positive and statistically robust effect on the number of hours that members put in. The density of members' network of coworkers helps the latter control and increase the amounts of effort invested into hard work. In that respect, the effect of members' social network on their performance is confirmed. However, beyond this confirmation, an additional and unexpected dimension of social structure also emerges here. Interesting negative effects of dense friendship ties (and centrality in the friendship network) on performance are apparent. One can speculate about the reasons for these effects, such as spending more time on personalizing work relationships, or providing role-distance, outside work than on actual work.9 But this shows that individual and group-level social ties can also take less productive forms-forms that decrease performance. In terms of economic outcome, structure that brings social capital to the individual may thus bring social liability to the firm. Social capital can be a double-edged phenomenon.

Focusing back on the positive effect of constraint in one's network of strong coworkers, models 2 and 3-where the weight of economic variables such as hourly rates is more visible-show that such an effect is robust. Indeed, it is also present as a good predictor of the amount of fees brought into the firm at the end of the year. It is thus relatively independent of the way performance is measured.


Table 2. Variables explaining economic performance measured by the Number of hours worked (Model 1) and by the amount of dollars brought into the firm in fees in 1990 (Models 2 and 3) in 1990

Effects

Seniority

Hourly rates

Time input'

Office

Specialty

Gender

Lawschoolattended

Centrality

Friendship

Coworker

Advice

Constraint

Modell

0.01

0.24**

-0.16*

-0.03

-0.14

-0.27*

0.17

-0.02

Standardized Estimates

Model 2

0.76***

O.IS*

0.01

0.00

-0.03

-0.11

0.01

0.27*

Model 3

0.78*·*

0.40***

O.OS

0.07

0.02

0.02

-0.01

-0.02

0.23*

Friendship -0.81**· -O.IS O.ll

Coworker 0.23* 0.16* 0.13*

Advice -0.04 O.OS 0.04

*p<O.OS, **p<O.OI, **·p<O.OOI. Statistical significance levels are purely indicative, since observations were collected for the population, not for a sample. Adjusted Rsquared are 0.66, 0.86, and 0.89, respectively (and 0.63, 0.84, and 0.87, respectively, without the coworker constraint variable). The managing partner, who concentrates on firm policy and administrative work and is not a timekeeper during his tenure was not included in the computations of these parameter estimates. '. Including the interaction effect of time input and hourly rates does not provide additional insights here because senior partners who charge high rates are not among the members who put in the greatest number of hours.

In model 2, status and seniority count. Partners do collect more money than associates, and the more senior lawyers are, the more money they tend to collect. In general, partners put in fewer hours but collect more dollars than associates because they charge more. Associates do collect less money although they put in more time than partners. The more senior associates are, the more hours they bill and the more they collect; senior partners charge more per hour, put in less time, and collect more money. Market conditions have a mixed effect on economic performance: in terms of office membership, Hartford attorneys put in more time than Boston attorneys, and collect more; but in terms of division of work, no specialty (corporate or litigation) seems to be systematically more lucrative than the other-in spite of a general advantage for litigation in the early 1990s in U.S. corporate law firms.

Given our interest in the effect of informal relationships on performance, it makes sense to look into the latter effects in more detail for associates and for partners separately. The following results are then obtained. For partners only, effects considered here for hours billed are at best unstable. Practice has a very weak effect: litigation partners bill slightly more than corporate partners. Again, partners popular as friends tend to bill slightly less (as for associates mentioned below). For


associates only, centrality in terms of friendship affects collection negatively. This means that associates very active socially and providing on average more emotional support or role-distance to others may end up being handicapped in terms of collecting. Obviously, the link between the two phenomena is indirect and remains to be explained, but it is nevertheless persistent. The more central associates are in terms of their number of coworkers (that is, the more colleagues they work with), the more hours they bill. The only associates who collect more hours are those who are also central in the coworkers network (they tend to be senior associates). For associates, seniority is the best predictor of performance in terms of hours collected, but seniority and centrality as a coworker are the best predictors of performance in terms of hours billed.

In this context, the analysis of the effect of individual social ties on economic performance also shows that attorneys informally sought out for advice and for collaboration by many others tend to bill and collect more than others. This extra effect is partly due to the fact that senior partners are central advisors. As already mentioned, the effect of social capital measured in terms of Burt's constraint scores is still present in model 3, in which seniority is replaced by hourly rates and time input. This confirms our expectations: members with a constraining coworkers' network put in more time, bill more hours, and collect more dollars. The more constraining one's coworkers' network, the higher one's economic performance.

Thus, overall, multivariate analyses show that the effect of position in informal structure and social capital accumulated in this position do count for explaining performance, but these effects are weaker than the weight of seniority or hourly rates as defined by the institutional setting. to The next sections provide a more detailed illustration of such effects.

STATUS, SOCIAL CAPITAL, AND ECONOMIC PERFORMANCE: AN ILLUSTRA TION

As suggested, in particular, by the effect of constraint in the coworkers' network, economic performance is indeed rooted in the structure of individual social ties in the firm. This raises the question of the relative weight of specific combinations of ties in providing a decisive push in performance increase (or representing a liability that decreases performance). In this section, I look at similar effects from a different angle by showing that a specific configuration of social ties is required, at the dyadic level, for high economic performance in this firm. This is due to the fact that maintaining production (collective action) requires many kinds of contributions and resources. Describing this configuration is equivalent to understanding how social resources combine with one another to increase or decrease individual performance. Results obtained above are translated into a description of specific configurations of ties that are associated with economic performance. These associations can be measured by the correlations between the frequency of more or less multiplex combinations of dyadic ties for each actor and his/her performance measurements.

~ ;- (M

Tab

le 3

. Cor

rela

tion

tabl

e fo

r pe

rfor

man

ce m

easu

rem

ents

in 1

990

and

mos

t fre

quen

t typ

es o

f com

bina

tion

of t

ies

amon

g dy

ads

of a

ctor

sa

Hou

rly

rate

s T

ime

Inpu

t T

ime

Inpu

t F

ees

Bil

led

Fee

s B

ille

d F

ees

Col

lect

ed

Fee

s C

olle

cted

D

olla

rs

Dol

lars

H

ours

D

olla

rs

Hou

rs

Dol

lars

H

ours

No

ties

-.3

4"

-.31"

-.0

8 -.

31

" -.

15

-.3

1"

-.17

iFj

and

jFi

-.36

" -.2

0 .1

1 -.

26'

.05

-.26

' .0

1

JAi

.47

'"

.36"

.1

1 .4

2'"

.19

.41'

" .2

0

jCi

and

jAi

.36"

.1

5 -.

08

.21

-.01

.20

.01

iCj

andj

Ci

and

jAi (

i.e. '

Bla

u ti

e')

.54

'"

.44'"

.1

0 .4

3'"

.18

.45*

**

.24'

iCj

and

jCi a

nd iA

j -.

52'"

-.3

4"

.04

-.40

'"

-.05

-.

39'"

-.0

4

iCj

and

jCi a

nd iA

j an

d jA

i .2

2 .2

8'

.10

.24

' .1

0 .2

5'

.16

iCj

and

jCi a

nd iA

j an

d jA

i an

d iF

j an

d jF

i .3

9'"

.43

'"

.19

.46

'"

.25

' .4

8'"

.30

'

a O

nly

com

bina

tion

s o

f tie

s w

ith a

sig

nifi

cant

cor

rela

tion

wer

e ke

pt in

the

tabl

e. (

Sta

tist

ical

sig

nifi

canc

e le

vels

are

pur

ely

indi

cati

ve, s

ince

obs

erva

tion

s w

ere

coll

ecte

d fo

r th

e po

pula

tion

, not

for

a sa

mpl

e).

·p<

o.O

S, ·

'p<

O.O

I, "·

p<

O.O

Ol.

'F' r

efer

s to

Fri

ends

hip,

'C

' to

Cow

orke

r, '

A' t

o A

dvic

e. 'i

Fj'

rep

rese

nts

a fr

iend

ship

tie

fro

m i

to j

. 'iF

j an

djF

i' r

epre

sent

s a

reci

proc

ated

fri

ends

hip

choi

ce.

For

exa

mpl

e, h

avin

g m

any

ties

of t

he 'j

Ai'

type

--w

hich

mea

ns t

hat m

any

peop

le G

) se

ek o

ut y

our

advi

ce--

is s

tron

gly,

st

atis

tica

lly

sign

ific

antl

y an

d po

siti

vely

cor

rela

ted

(0.4

1"')

wit

h co

llec

ting

hig

h fe

es m

easu

red

in d

olla

r am

ount

s.

~

N ~

-ti o i:l ~ ~ .., 5' - ~ ~ [ I::>

;:s

I::>..

t--o

5'

[ ~.


Configurations of Dyadic Ties and Economic Performance As mentioned in the presentation of the firm, many combinations of ties are possible at the dyadic level in three networks. The most frequent type of combination (besides the 'no tie' possibility) between two persons in the firm is a mutual cowork tie with one unreciprocated advice tie (from j to i)-I will call this compound a Blau-tie in reference to Peter Blau's (1964) work on exchange of advice for status. Many partners have such ties with themselves as advice distributors and associates as advice seekers. In contrast, there are no duplex mutual cowork and friendship ties. The complete absence of such ties is interesting. As seen above, although friendship and work can go hand in hand when an advice component is also present, the two resources seem to have distinctly different natures.

Table 3 presents correlations between specific configurations of dyadic ties for each individual in the firm and several measurements of performance. This exploratory analysis provides several results. First, there is a negative correlation between having many 'no ties' (or many 'empty' ties-potential ties not actually realized) and putting time in, billing, and collecting. The less one exchanges resources with others, the lower one's performance. Second, there is a positive correlation between having many ties such as 'being sought out (unreciprocated) for advice exclusively' (no cowork or friendship component) and dollar amounts invested (time worked by hourly rate), billed (time billed by hourly rate), and collected (fees actually collected). The same correlation holds when the tie includes an unreciprocated friendship component. This is partly related to the fact that senior partners, who are often sought out for advice and cited as friends (without reciprocating), bill and collect more given their higher hourly rates. Third, and more suprisingly, correlations between having many ties that are exclusively reciprocated work ties and any performance index are all really small. 'Work only' relationships are quite neutral in terms of their association with economic performance. Such ties are more frequent for associates than for partners. However, adding one component to this combination changes this result. Given that partners work with associates more than with other partners, there are positive correlations between having many such ties with the added component 'being sought out for advice' (unreciprocated) and performance indexes. Such Blau-ties are much more frequent for partners than for associates. In my view, this confirms that self-entrapment by partners in teams of coworkers is compensated by Blau-type status and professional recognition. When adding friendship components to such ties (such as citing j as a friend or citing each other as friends), positive correlations become stronger, although there are less occurrences of that type. 11 Fourth, and conversely, there are mostly negative correlations between having ties with components such as 'having a reciprocated work relationship and seeking out advice (unreciprocated)' and dollar amounts put in, billed, and collected. Such ties are more frequent for associates (particularly nonsenior associates). Finally, there are strong and positive correlations between having individual triplex mutual reciprocated ties and dollar amounts put in, billed, and collected. Exchanges of that type are much more frequent for partners. In short, the results translate our previous statements into relational terms. Partners' higher economic performance, for example, shows in the correlation between having many Blau-ties and the amounts of dollars collected.


To illustrate this analysis, the data provide examples of low and high economic performers and their specific combinations of social resources. In these examples, each attorney has a relatively different profile, but important common characteristics are related to low or high performance.

Low and High Performers: The Importance of Blao-ties Among low performers (still in terms of dollars brought in in 1990), one partner has, in terms of compounds of ties, a higher than average proportion of persons that he cites as strong coworkers who do not reciprocate, and he has a lower than average proportion of Blau-ties, and of triplex reciprocated ties. He considers many people to be his friends, but they do not reciprocate. His partners do not listen much to his opinions about firm management and policy issues. In terms of his configurations of dyadic ties, this partner-who is the least productive or performant one-has the following combinations. Four types of ties in which he says that he is friends with others who do not reciprocate, four types of ties in which he is sought out for collaboration and that he does not reciprocate, four types of ties in which he is sought out for advice (lower number than the average partner), and five types of ties in which he seeks out others for collaboration but is not cited by them. The strong specificity here is also the absence of direct reciprocity. There are many relations with a transfer of resources, but few relations where there is direct and multiplex exchange. Notice that none of the configurations associated with strong economic performance is present in this partner's profile. Recall the managing partner's comment about this partner:

[Jack] is a decent guy on a personal level. He says that under pressure he does things but that he is a corporate lawyer, and there is no pressure on him usually, so he just doesn't do it.

A roughly similar pattern characterizes a second low-performing partner. He has a much higher than average proportion of 'no ties' and a higher than average proportion of persons that he cites as strong coworkers who do not reciprocate; he also has a lower than average proportion of Blau-ties, and of triplex reciprocated ties. He nevertheless has an average number of reciprocated friendship ties. Most of his partners do not listen much to his opinions about firm management and policy issues. Very few come to him for advice, and he has a relatively low number of coworkers. At the dyadic level, the specificity of this partner's relational profile is the contrast between his very few ties with others in the firm and two very strong (triplex reciprocated) ties with unconditional partners (his insurance policy against expulsion). He also claims three friendships with other partners, but they are unreciprocated. Recall the managing partner:

[Frank] was at the low end (of the performance scale) for a long time. Partner 5 went to talk with him. He claims that there isn't anything to do. He says: 'I am a corporate lawyer. My kind of work has dried up, I am out there in the bushes, hustling, doing everything I can.' That is difficult to check. But there is also the fact that part of the people he has worked with do not want to do it again because they think his competence is in doubt. Partner 17 is extremely good in his field. He once volunteered to go see this partner to share some work with him. He went to see him, but he says [Frank] did a horrendously poor job.


In sum, the two low performers are less involved than others in exchanges of production-related social resources in the firm. They do not create teams of associates who can rely on them for work and provide status recognition.

Among high-performing partners, there are broadly speaking two of types of relational profiles. The first type includes some of the most senior partners, especially given their high hourly rates. Their ties often seem to combine unreciprocated transfers toward j more than exchanges with j. They are more often sought out than seeking someone out-providers rather than beneficiaries. This makes economic sense, since they are in charge of running the organization that ultimately produces their larger share of the pie. But it also makes sense socially. For example, Partner 1 is highly sought out for advice. He cites few coworkers because he mainly worked, during previous year with a junior partner acting as a 'foreman' (Partner 26) concentrating coworkers' exchanges. Partners 2 and 4 have the same relational profile, except for their more frequent mutual involvement in mutual coworkers' ties and a higher than average proportion of Blau-ties. In addition to the work-oriented components, Partner 4 has an exceptional five triplex mutual ties with other partners. Thus, in general, unless they work with a 'foreman,' senior high performers are strongly work-oriented persons intensely involved in coworkers' ties and highly sought out for advice. Their compounds do not include many friendship components, except with a very few select other senior partners with whom reciprocation is taken for granted.12 The taboo concerning friendship with associates seems to be extended to young partners with a few exceptions for contemporaries.

The second type of high-performing partner also includes work-oriented persons with an above-average proportion of Blau-ties. They do, however, diversify their exchanges of resources-their types of ties-much more than their senior partners. Their compounds are more personalized and include more friendship components. \3

Typical of this relational profile is Partner 26. He is hard working, heavily involved in the business of the firm, one of the persons with the most ties in the firm altogether, among the record holders for the Blau-ties and for the triplex mutual ties (with his Boston contemporaries who-for reasons linked to the history of the firm-form a very cohesive group of partners). But in addition to many monoresource uniplex ties (such as being often sought out for advice and only for advice, although not as much as Partners 1 and 6), he is involved in many duplex or triplex ties with colleagues who work with him but also like him (often unreciprocated) and come to him for advice (unreciprocated). He is thus more relaxed about personalizing work ties and shows more social openness (than senior partners) to colleagues working with him. He is a work-oriented all-around exchanger and investor of resources.

The contrast between low and high performers' relational profiles shows that high performers (in terms of dollars brought in) draw heavily on their social resources and have common specific relational characteristics, in particular those involving them-not surprisingly-in task-related exchanges with associates. In that respect, the firm benefits from the networks that some individuals have (dense multiplex networks, especially constrained at the group level in the coworkers network and rich in Blau-ties) and suffers from the networks that other individuals


have (sparser networks and especially wealdy constrained in the coworkers' network).

However, as suggested by the effect of constraint scores in the coworkers' network and by the illustrations above, such configurations of ties (described at the dyadic level) favoring high economic performance are not distributed randomly in the firm. Their distribution depends on a wider pattern of social ties, a pattern that represents a multiplex generalized exchange system. Recall that, as described above, the structural tendencies in this firm can be summarized by two separable forms of interdependence: first, the interplay of friendship and advice, wherein an individual's friends may be a source of advice ties; second, the interplay of advice and coworker ties, with the possibility that advice ties play an important role in generating further advice ties and coworker ties. Thus mutual strong work ties occuring in conjunction with either unidirectional (partner to associate Blau-tie) or mutual advice ties-which are positively correlated with economic performance in Table 3-have a higher chance of occurring in workgroups with constrained coworkers ties. In that respect, productive coworker ties are embedded in advice ties. In tum, advice ties are often driven by more personalized ties. While there is a very weak association between duplex coworker and friendship ties, there is a strong one between advice and friendship ties. This configuration also has a higher chance of occurring in cohesive workgroups, and it is especially consistent with the strongest positive correlation in Table 3 between triplex reciprocated ties and high economic performance. Thus strong and stimulating (that is, constrained) coworkers ties combined with advice ties or with both advice and friendship ties have a chance of being economically more productive than other ties and more likely to happen in dense workgroups. Mentor relationships, for example, exemplify one way in which closely selected associates can work with some partners, combine work and personal ties, and create teams.

The existence of such workgroups and the multiplex generalized exchange system in which they are themselves embedded are structural features of the firm as a whole: it is a component of its corporate social capital. It can thus be asserted that if economic performance is rooted in the structure of individual social ties and in cohesive workgroups, the latter are themselves rooted in collective social structure, yielding firm-level social capital. This firm-level social capital, that requires many kinds of contributions, is also key to maintaining collective action and production. It is made more visible in the following section.

CIRCULA TION OF RESOURCES AND ECONOMIC PERFORMANCE

Firm social capital is partly reflected by a multiplex generalized exchange system. 14

The existence of this system in this firm has already been suggested by an abundance of density effects and scarcity of direct reciprocity effects at the dyadic level across the three networks (see, for example, Lazega and Van Duijn 1997). Such a general pattern of relations is made possible, in part, by reliance on the institutional framework of the firm as a guarantee that resources contributed will eventually come back (although not necessarily in kind) and by the fact that the longer members stay in the firm, the more personal relationships develop among


Figure 1. Relationships between positions of approximately structurally equivalent actors in three networks of relations (advice, coworker, and friendship) in the law firm (adapted from Lazega 1998)

Thick lines represent reciprocated choices. Grey line represent advice ties, black lines coworker ties, and dotted lines friendship ties. Roman figures represent the number of the position. Letters have the following meaning: H for Hartford, B for Boston, P for partners, As for associates, L for litigators, C for corporate. Thus position One represents BLP-Boston litigation partners. The bigger circles are positions of partners; the smaller circles are positions of associates.

them. For this pattern to encourage partner self-entrapment in their own workgroups. for example. it has to be shown to reflect and sustain the tendencies described above. Productive members can share several types of resources without immediate reciprocity. The system involves forms of indirect reciprocities that take into account several resources.

A 'Locally Multiplex' Generalized Exchange System When staying at the dyadic level. it is difficult to get a sense of how exchanges use different resources at the same time. and of the overall pattern of exchanges of the three resources in the firm. Recognizing this more generally. Levi-Strauss (1949) distinguished two forms of exchange: direct or restricted exchange (dyadic) and indirect or generalized exchange (structural). An analysis approximating structural


equivalence among members of the firm across the three networks provides such an overall view, represented in Figure 1. A detailed presentation of this Figure and the density tables and image matrices on which it is based are available from the author. To summarize, it shows that the three networks stacked together break down into nine positions of approximately structurally equivalent members. Asymmetries in the transfers of resources, along with the dependencies attached to them, create a mUltiplex generalized exchange system as shown in Figure 1.

Positions One, Two, and Three are positions of partners: all the others are positions of associates. The thick gray lines (reciprocated advice ties) reflect the backbone of the firm: three partners positions and their senior associates. Notice that requests for advice converge toward them. Many of the relationships between positions are not symmetric. Position One is a socially dominant group of Boston litigation partners who get advice, strong collaboration and friendship from Positions Two and Five. Its members get almost what they want from the people they choose. Many positions of associates are directly indebted to it for advice and collaboration, but it is not the top-performing position (third in average individual dollar collection). Position Two is a group of Boston corporate partners almost in the same dominant situation as Position One, except that it has an exchange of advice for friendship with Position Five. Many positions of associates are directly indebted to it for advice but not for friendship. It is nevertheless the top economically performing position. Position Three is a group mixing Hartford corporate and litigation partners, in the same category as Positions One and Two in terms of dependence on others for resources. It claims strong collaboration from Position Two but is unreciprocated in kind: Position Two members tend not to rely on Position Three for strong collaboration but do so for advice and friendship. Here reciprocity tends also to be direct but not necessarily in kind. Note that Position Three has direct exchanges of strong collaboration only with two positions of associates, Positions Six and Seven, and not with other positions of partners-but cross-selling can still take place, for example, since it is a mixed position (in terms of specialties). As for Position Two, many positions of associates are directly indebted to it for advice but not for collaboration or friendship. It ranks second in average individual dollar collection.

If we come back to the individual and dyadic levels, this picture confirms that partners' positions are characterized by a comparatively very high proportion of Blau-ties-whereas associates' positions are characterized by the corresponding high proportion of ties in which they are the advice-seeking party-and are highly cohesive given their high representation of mutual triplex ties. Thus, at the aggregate level, all three resources tend to circulate within positions of partners and among the two Boston positions. Two additional forms of embedded ness indicate that the partnership agreement is enforced through both economic interest and social ties. Partners from Position One are in a socially advantageous situation to remind partners from Position Two of their commitment, or to calm down status competition among them. Partners from Position Two, in tum, are in a socially advantageous situation for pressuring Position One and Position Three partners back to good conduct. Although they are not often asymmetrically indebted to each other and dependent on indirect reciprocity from each other, economic relations among


partners are overembedded for Position One and underembedded for Position Three, with Position Two members playing a key role of balancing the two forms.

This also confirms that the forms of embedded ness of the labor contract for partners and associates are radically different. Here, multiplexity is also more extensively used to enhance the productivity of this economic tie. Associates tend to feel indebted to partners for strong collaboration, advice, and sometimes friendship (mostly unreciprocated). Two senior associates positions, Positions Five and Six, have a less clear profile-in terms of direct reciprocity-than partners' positions. Positions Five exchanges the three resources with Position One partners. In that sense, it has almost a partners' profile. It has other uniplex and directly reciprocal exchanges and does not have to exchange one resource for another (with the exception of friendship for advice with Position Two). Two positions of associates are directly indebted to it for collaboration, advice, and friendship. Positions Six is in a different situation. It has more direct exchanges of strong collaboration with positions of partners and has also a direct exchange of advice with a position of partners. But it does not exchange friendship with partners (remember that it is made of lateral associates who did not grow up through the ranks but were hired away from other firms). It claims friendship with Position Two but this is not reciprocated-unlike Position Five's access to Position One's friendship. Position Six thus gets friendship from colleagues (Positions Five, Seven and Eight) sometimes different from those to whom it provides it.

Small cycles characterizing local and multiplex generalized exchange can be found in this system. Multiplex and local cycles reflect the existence of highly embedded strong coworkers' teams, for example, between Positions One, Five and Nine, or between positions Two, Six and Eight, or One, Two, and Four. This is due, in particular, to the fact that indirect reciprocity is at its strongest with less senior associates who are never in a position, for instance, to reciprocate for advice. They most often have to reciprocate in Blau-type status recognition and strong commitment to work. This reciprocation in commitment to work is not necessarily directly directed to sources of advice but to others. For example, Position Eight gets advice from Position Three but is only indirectly involved in strong work ties with Positions Three through Six. Another example is Position Seven, that gets the three resources from Position Five but is not in a position to reciprocate directly at all, and thus remains indebted but provides Position Three with strong commitment to work and friendship. A similar short cycle is also present between Positions Three, Six, and Seven. Position Four is in the same dependent situation as Position Seven with regard to the three partners ' positions it has ties to. Note that it does not get friendship at all from other positions, not even indirectly, which shows that the generalized exchange system fails to provide this resource to some of its members.

A large spectrum of forms of embedded ness is present here to enforce the labor contract for partners and associates. It would be too simple to summarize the situation by saying that performance of individuals is affected by these forms of embedded ness because high performers are always involved in very multiplex cycles and low performers in less mUltiplex cycles, or because resistance to the general exchange circulation rule always produces low performers. For some associates high productivity goes hand in hand with strong friendship ties with


partners: Positions Five and Seven overbill and overcollect compared to the other associates. Associates may often feel kept at arm's-length and that 'partners don't let them in,' but here we also see that some partners are nevertheless in a favorable social situation to extract high commitment from these associates. And at the other extreme, Position Four associates are also highly productive but much less socially connected to the partners with whom they work: requests for advice still indicate social embedded ness through status recognition, but partners are less in a position to extract work for friendship. In contrast, junior associates are less productive economically and claim friendship with one another and with senior associates. Here partners can play on resource dependencies to get commitment from associates to their labor contracts in different ways, sometimes indirectly through the dependence of junior associates on senior ones. Senior associates can also play this resource dependence game with strongly socially embedded junior associates (in Position Eight, pure friendship ties are highly overrepresented, and Position Seven sends a record proportion of unreciprocated triplex ties) but not so much with partners.

Back to the dyadic level, we have seen that there is very often-in Positions Two and Seven, One and Nine, Two and Four, and Three and Six-a Blau-tie. This compound reflects one of the most frequent types of embeddedness. Partners also concentrate requests for advice and unreciprocated citations as friends and advisors or as coworkers and advisors. Such citations converge towards all partners' positions but are strongly overrepresented for Position One. Also associates ' positions are much less cohesive, which is confirmed by a low proportion of mutual triplex ties. In Position Three, empty ties and mutual duplex (cowork and advice) ties are overrepresented.

Multilevel Embeddedness: A Virtuous Circle? Exchanges and transfers of resources, along with asymmetries and dependencies attached to them, create a local and multiplex generalized exchange system. This is a social ecological system that has grown around the formal dimensions of the organization and constraints imposed by interdependence in production. It is part of firm-level corporate social capital: it helps members accept terms of the labor contract that they do not necessarily have a narrow interest in accepting. In other words, the firm has found in this exchange system a structural solution to the structural problem of cooperation and commitment to the labor contract, a sort of partial equilibrium in the circulation of resources needed to fulfill it. Given informal constraints guiding the choices of coworkers, advisors, and friends in this firm, and given the existence of such multiplex and local cycles reflecting the existence of highly embedded strong coworkers' teams, givers have a guarantee that they will become receivers, although not necessarily in kind. By allowing such a system to exist, this firm maintains certain forms of resource circulation, a precondition for group solidarity.

Thus, individual commitment to labor contract and individual economic performance are increased by membership in dense workgroups that themselves need a wider and more multiplex exchange system to operate. Individual performance that benefits the firm as a whole is driven by workgroup pressure and the social system of the firm: the organization helps its members perform, thereby


helping itself through aggregated performances. In many ways this mechanism is good for the firm as a whole as weII as for the individual. It aIIows very different individuals to carve out a place for themselves in the organization, based on the exchange of various resources. Social networks of individuals make them more productive in the organization-creating individual-level social capital; but it also transforms itself into firm-level social capital because the firm as a whole is more successful in billing and maintaining labor contracts and integration.

However, the embeddedness of individual economic performance in a social exchange system also raises the issue of the relative contribution of individual social capital and firm social capital to collective action and performance. 15 When several resources circulate, exploitation is present but not easily measurable. For example, if firm-level social structure helps individual members perform, then members who participate in building up this social capital are also entitled to some of the credit that goes to high performers. In this situation, the restraint of members (both in keeping track of all their contributions and returns, and in politicizing the use of measurements) is a micropolitical condition for the efficiency of this system. This is especially the case with teamwork, where others' behavior can easily be perceived to be opportunistic. In fact, members' restraint is often weak: many conflicts in corporate law firms or other private professional services firms are disputes about fairness in compensation to individual partners. But many cultural aspects of such organizations, such as an ideology of collegiality, are explicit exhortation to such a restraint.

Thus measuring only economic performance tells only one side of the story of contribution to collective action. This limiting condition for the efficiency of such a virtuous circle is the reason for managing partners often to emphasize 'spreading credit' as widely as possible. Nevertheless, however, recognizing and measuring the relative and specific importance of corporate social capital (as compared to other forms of capital) cannot be done without understanding this political negotiation that enables members to evaluate their contributions. This means that the relative value of social resources is negotiated, and that this negotiation is political. FoIIowing these politicized negotiations and how they try to disentangle the merits of the firm as opposed to that of its individual members in the production of a specific performance is beyond the scope of this chapter, and perhaps beyond the possibilities of structural analysis.

CONCLUSION

In summary, this chapter examined the question of the relationship between social structure and economic performance at the intraorganizational level. I have identified a few conditions under which individual social ties are most productive for the firm in collegial organizations-where the production process is difficult to routinize, where professional expertise and advice cannot be standardized, and therefore where internal transaction costs for the firm as a whole can be assumed to be a large part of total costs. An empirical study of a medium-sized northeastern U.S. corporate law firm was used for that purpose. In this firm, attorneys are shown to be bound by a labor contract that is difficult to sustain on pure economic terms. Partners can easily free-ride; associates can threaten the quality of work. Against


this damage potential, a social system sustains their commitment. Using network data collected in the firm, I describe and measure social capital at the individual, workgroup, and structural levels to show that the more constraining the member's coworkers' network, the easier it is for the firm to extract higher economic performance, including from tenured partners, by controlling the time put into work. Thus, position in the relational structure and social capital accumulated in this position do count for explaining performance, although these effects are weaker when compared to the weight of hourly rates as defined by the institutional setting. With regard to partners, such teams represent an element of self-entrapment compensated by stimulation, status and professional recognition.

Second, the effect of the individual social network on economic performance is decomposed at the dyadic level by looking at specific combinations of ties that sustain this commitment and provide a decisive increase in performance (or that represent a liability decreasing performance). For example, specific configurations of social ties, such as mutual triplex ties, are strongly correlated with high performance. This fleshes out the positive effect of constraint scores in the coworkers' network. Examples of low and high economic performers, and their respective combinations of social resources, were also provided as illustrations.

Third, back to the structural level, a locally multiplex generalized exchange system is described as a precondition for individual and group-level social structure to be productive because it maintains the circulation of social resources in the firm, thus making it possible to improve performance for structurally well located members and workgroups who can benefit from this circulation. A multilevel form of embedded ness is thus revealed here, which shows the importance of taking into account a 'meso' level when measuring the relationship between social structure and performance. In this particular case, the notion of multilevel embeddedness advances our understanding of economic performance: the latter is rooted in individual social capital, which is itself rooted in workgroup and firm social capital, which in tum helps individual members in being more productive. Individual social ties can be most productive-for the individual and/or for the firm-when they are part of a favorable system accumulating collective social capital. By favorable, I mean that the individual social network is located at the right place and mobilized at the right time in a wider production and exchange system.

Finally, however, the politicized nature of performance measurement prevents us from claiming that a permanent virtuous circle is produced by the fact that social capital is located at the individual, group, and structural levels. Exploitation and behavior perceived to be opportunistic are also corollaries of this multilevel embeddedness. A final example illustrates one possible implication of these results. As many managing partners in law firms know, the importance of constraint at the group level is not necessarily an encouragement for management to create dense and permanent workgroups in collegial organizations. The existence of such groups is risky for the firm. They can threaten the firm with disintegration when entire teams consider themselves exploited (relatively to others in the firm or to others outside the firm), decide to defect, and take away with them part of the firm's human and social capital. As shown elsewhere (Lazega 1992, 1999), what the managing partner of this firms says about individuals is also true about workgroups:


There are client loyalties to individual lawyers within the firm, but among ourselves we view all clients as clients of the firm. And indeed, if you are an individual to whom the client has demonstrated a great degree of loyalty, one of your responsibilities is to make sure that there are other partners to whom that client may also look and rely on, not necessarily on an ongoing basis. But if for some reason, for example, I am away, if I were suddenly to decide to go pump gas for the rest of my life-any number of thingsthat client loyalty is not an asset that belongs to me. If I were to go to another firm, if I have done my job well here at S,G&R, if I call my client and say 'I want you to know that I am in firm X,Y &Z now,' that client's response should be 'Whom at S,G&R should I call now?' It shouldn't be 'What's your new number?' Whether that would be the case in all cases, who knows? That's what ideally it should be. (Managing partner at the time of the study)

Social structure can thus produce a large amount of social capital for collegial organizations, but its manipulation is double-edged.

APPENDIX A

SOCIOMETRIC NAME GENERA TORS USED TO ELICIT COWORKERS, ADVICE, AND FRIENDSHIP NETWORKS

Here is the list of all the members of your firm.

Coworkers network: Because most firms like yours are also organized very informally, it is difficult to get a clear idea of how the members really work together. Think back over the past year, consider all the lawyers in your Firm. Would you go through this list and check the names of those with whom you have worked with. (By 'worked with' I mean that you have spent time together on at least one case, that you have been assigned to the same case, that they read or used your work product or that you have read or used their work product; this includes professional work done within the Firm like Bar association work, administration, etc.)

Basic advice network: Think back over the past year, consider all the lawyers in your Firm. To whom did you go for basic professional advice? For instance, you want to make sure that you are handling a case right, making a proper decision, and you want to consult someone whose professional opinions are in general of great value to you. By advice I do not mean simply technical advice.

Friendship network: Would you go through this list, and check the names of those you socialize with outside work. You know their family, they know yours, for instance. I do not mean all the people you are simply on a friendly level with, or people you happen to meet at Firm functions.

I would like to thank Henk Flap. the November 1997 ICS-seminar participants in Groningen. The Netherlands, and the book editors for comments on a previous draft.


NOTES

1. Elsewhere I have provided other examples of the fact that this finn finds structural solutions to structural problems (Lazega 1992a, 1995a, 1995b, 1997; Lazega and Vari 1992; Lazega and Lebeaux 1995; Lazega and Krackhardt 1997). 2. This is not to say that this ecological system, which makes the partnership agreement enforceable, disciplines all the members equally strongly. Some pay a higher price to be part for it. For example, some associates are put in a better position to try to build their competitive advantage on the use of these embedded ties. Close demonstration of this is, however, beyond the scope of this article. 3. Nelson (1988: 91-92) defines traditional management as characterized by 'I) ad hoc and reactive policy-making, with little long-range planning; 2) direct administration by leading lawyers, aided only by a part-time managing partner, with no regular monitoring of internal perfonnance measures or financial infonnation; and 3) infonnally defined and shifting work groups.' Bureaucratic management is defined by '1) a specialized policy-making group that actively engages in strategic planning; 2) a developed administrative component consisting of a managing partner and a mechanism for collecting and analyzing data on the financial perfonnance of individual lawyers and work groups; and 3) well defined work groups (usually taking the fonn of departments) with recognized heads who supervise the group and report to the central policy-making group.' 4. For example, given the way a partner is compensated in the finn, looking at the dollar amount actually collected in 1991 does not indicate exactly how productive this attorney was in 1991. Work done in 1990 can be compensated in 1991 (or perhaps even later), and such overlaps make it difficult to disentangle an attorney's productivity in one year as opposed to his or her productivity in another year. Simultaneously, looking at the number of hours billed in 1991 gives an idea of an attorney's productivity in 1991 but does not mean that all the work was done in 1991. 5. Seniority is defined by the rank of partners in the letterhead, which is mainly based on age and years with the finn (with the exception of four partners who were hired away from other finns). Coding of seniority levels in senior, medium seniority, and junior partners is based on cutoffs between Partners 14 and 15 (a difference of eight years in age) and between Partners 27 and 28 (a difference of nine years in age). These categories were explicitly used by the partners themselves. 6. For more infonnation on these networks, see Lazega (1992a, 1992b, 1993, 1994, 1995a, 1995b), and Lazega and Van Duijn (1997). 7. This is partly at odds with Burt's (1992) general statement about association between low constraint and high perfonnance and more consistent with Coleman's (1990: ch. 12) ideas on the benefits of closure and embeddedness of ties. In this chapter, I mainly analyze economic perfonnance understood as the amount of fees brought into the finn at the end of the year. Such amounts depend minimally on the amounts of time worked and on hourly rates. Nevertheless, the more members work, the more they perfonn in that sense. My point is that extracting work from them should be easier in a constrained network of work ties. Analyzing the determinants of other types of individual perfonnance, such as promotion to partnership, could presumably yield different results and an opposite sign to the association-which would be more in line with Burt's results. About this issue, see also Gabbay (1997). 8. There are many reasons for the inability of hourly rates and numbers of hours worked to explain all the variations in financial perfonnance. First, billing partners do not bill all the hours worked by their team. There are various fonns of nonchargeable time, and very often there is a negotiation between the finn and the client as to what is an acceptable price for the services rended. The billing partner then writes off a considerable proportion of hours worked before the bill is sent. Second, corporate law finns have notorious difficulties collecting what was billed, and many partners choose to live with high account receivables rather than antagonize a client from whom they expect more business in the future. 9. However, causal links are difficult with non longitudinal data; it is impossible to know here whether members are low performers because they establish different types of relationship with their colleagues or whether they establish these relationships to try to mitigate the effects of their low perfonnance and carve out a different place for themselves in the group. 10. A longitudinal approach could bring more insights into these effects and especially enable researchers to identify causal relations. A look at perfonnance data for the next year (1991) suggests the possible existence of a cyclical and infonnal mechanism in which attorneys who work and collect a lot on big cases tend to bum out temporarily, to attract much social approval, to slow down a bit, and work with fewer colleagues, until they are ready to pick up again, thus triggering a new phase in which they start working with more colleagues again, increase their work load, and bill more until they collect again.


Needless to say, the existence of such a mechanism remains to be proven, and, if so, the cycle should vary from one attorney to another. II. The type of tie in which i and j consider each other as strong coworkers and in which j seeks advice from i whereas i seeks j for socialization outside the fmu is a frequent type of compound. 12. There are exceptions, of course: Partner 12, for example, has more than ten ties including unreciprocated (by j) friendship ties. Declaring more friends than one may actually have nevertheless characterizes associates' profiles much more. 13. Practicing very diverse and more complex-less task-oriented-types of exchanges also presupposes more flexibility with rules of exchange of resources. Description of such normative games, however, is beyond the limitations of this chapter. 14. For a discussion of the relationships between networks and generalized exchange systems, see for example, Levi-Strauss (1949), Ekeh (1974), Bearman (1997), Breiger and Ennis (1997). 15. The fact that this system helps some individual members reach high performance does not mean that it is egalitarian in the distribution of resources and in the provision of structural solutions to individual problems. This can also be illustrated by looking at the relative chances of senior associates to become partners. In their competition for the attention of partners, associates with the right connections to the right partners-with a specific position in this pattern and a specific type of individual social capitalhave a clear structural advantages in the highly selective race to partnership. They are in the fast lane because these connections, among other advantages, allowed them to play with organizational rules in an rewarding way, in particular to cross internal boundaries (for example, seek advice from very senior partners), provided that such 'infractions' are limited and well localized (Lazega 1995a).

•

CEO Demographics and Acquisitions: Network Effects of Educational and Functional Background 14

ABSTRACT

Pamela R. Haunschild Andrew D. Henderson

Alison Davis-Blake

This study investigates the effects of CEO educational and functional background on corporate acquisitions. Educational and work-related functional backgrounds are likely to come with interorganizational networks, networks that stay with individuals over long periods of time and have the potential to affect acquisitions. We argue that these networks constitute a form of interorganizational social capital, which directs acquisition activities along certain channels. Hypotheses are tested on 449 firms and their acquisitions during the 1986-1993 period. We find evidence that the networks that come with different CEO education and functional backgrounds are related to the type of acquisition completed by that CEO's firm. Obtaining a degree from an elite school. for example. is related to engaging in acquisitions in unrelated industries. We also find functional background effects are strengthened under conditions of uncertainty and educational background effects are weakened with tenure. These results suggest the importance of personal networks in affecting major firm strategic actions. and highlight the contextual nature of acquisition decisions.

INTRODUCTION

What factors affect a firm's choice of strategic actions? This question lies at the heart of research on strategic management, and factors external and internal to the firm have been proposed as answers. Several recent studies have shown that key strategic actions such as engaging in acquisitions. adopting a poison pill. and implementing the multidivisional form are affected by external factors. These factors include the actions of other firms in the same industry (Fligstein 1985. 1991; Palmer, Jennings. and Zhou 1993). the actions of interlock partners (Davis 1991;

CEO Demographics and Acquisitions - 267

Haunschild 1993; Palmer et al. 1993), and the constraints imposed by resource dependencies (Pfeffer and Salancik 1978).

Although external factors undoubtedly affect strategic actions, there has been a renewed interest in internal factors as determinants of strategy. In particular, several recent studies have investigated the effects of CEO demographics on a firm's strategic actions (Fligstein 1990; Finkelstein and Hambrick 1990; Palmer et al. 1993; Davis, Diekmann, and Tinsley 1994). Most of this work has focused on the relationship between functional background and strategy. Early foundations for this work came from the Carnegie School theorists. They argued that, by creating a cognitive map or way of thinking about issues, functional background affects managerial decision making in complex situations (Dearborn and Simon 1958; Cyert and March 1963; but see Walsh 1988 for an alternative point of view). Later, Hambrick and Mason (1984) advanced the upper echelons perspective in which they argued that organizational strategies reflect the cognitive bases and preferences of the top management team. Various demographic variables, including functional background, were proposed to represent these cognitive bases. The relationship between various demographic variables and strategy has been empirically demonstrated by several researchers (e.g., Norburn and Birley 1988; Bantel and Jackson 1989; Grimm and Smith 1991; Wiersema and Bantel 1992). Demographic variables have also been proposed to be indicators of power, and power affects strategic action in predictable ways. For example, Fligstein (1990) showed the wave of conglomerate acquisitions that occurred in the 1960s can be partially explained by finance CEOs that rose to power and executed acquisitions consistent with their functional specialty.

Past research on the link between managerial background and strategic action is important because it shows that the backgrounds of top managers can have an effect on major aspects of corporate strategy. The theoretical foundation of prior work, however, has been limited to arguing that managerial background affects managerial cognition and power. Yet certain educational and work-related backgrounds are likely to come with interorganizational networks, networks that stay with individuals over long periods of time and have the potential to affect major strategic decisions. The impact of these networks, in turn, is likely to be affected by other contextual factors. The purpose of this study, then, is to examine the network effects of CEO education and functional background on a major firm strategic action: corporate acquisitions. We argue that the networks that come with functional and educational background provide a form of interorganizational social capital, which directs acquisition activities in predictable ways. By studying the networks that come with education and functional background, we provide a link between the internal (CEO characteristics) and external (interorganizational factors) perspectives on strategic action. A second purpose of this study is to investigate contextual conditions likely to moderate these network effects. We study two such conditions: decision uncertainty and CEO tenure. We choose uncertainty and tenure because each have been shown to moderate the impact of networks in other settings.

Corporate Acquisitions Corporate acquisitions are an interesting and important arena in which to explore the network effects of CEO educational and functional background. Acquisitions represent a critical strategic choice, one that typically involves the commitment of


substantial resources and the expenditure of political capital. Acquisitions are also highly complex events (Haspeslagh and Jemison 1991) that can generate a great deal of decision uncertainty (Haunschild 1994). Some researchers (Dearborn and Simon 1958; Cyert and March 1963; Hambrick and Mason 1984) have proposed that demographic indicators like CEO educational and functional background are particularly likely to be associated with complex, uncertain decisions. Thus, corporate acquisitions are an important setting in which to test whether CEO background affects strategic action and whether these effects vary with contextual conditions such as decision uncertainty. Although the effects of functional background on acquisitions have been the subject of a few previous studies (Song 1982; Fligstein 1990, 1991; Davis et a1.1994), these studies have been limited to proposing that the relationship between functional background and acquisitions are due to its effect on managerial cognition and in determining who has power in organizations. By examining network effects, we present a very different view of acquisitions from what has been presented before.

What causes firms to do acquisitions, and do acquisitions of certain types? This is an interesting and important question, especially in light of evidence indicating that acquisitions often reduce the value of the acquiring firm (Ravenscraft 1987). Much of the work on acquisition motivations is financial and efficiency-based, proposing that acquisitions are driven by such things as the search for synergy (e.g., Jensen 1984; Ravenscraft 1987). Resource dependence theory presents an alternative perspective, proposing that acquisitions are a response to the constraints imposed by organizational interdependence (Pfeffer and Salancik 1978). The managerial and agency theories propose that top managers, especially CEOs, have a great deal of influence over acquisition activities, and thus their motivations and incentives can playa significant role in predicting corporate acquisition activity. Some managerial theories, for example, propose and find evidence that acquisitions are driven by managerial desire for the power, prestige, and financial rewards associated with managing large companies (e.g., Marris 1964; Baumal 1967). Agency theory proposes that poorly-monitored managers and managers whose incentives are out of alignment with shareholder interests will engage in acquisitions consistent with their own interests and inconsistent with shareholder interests (e.g., Morck, Schleifer and Vishny 1990).

While these latter theories present evidence for individual effects on corporate acquisition activity, they are largely asocial, considering only the impact of individual incentives and desires. The same is true for the few studies looking at the relationship between functional background and acquisitions. This means that most existing theories of acquisitions do not consider the social context in which firm managers are embedded. Firm managers exist in a social world, and carry with them interorganizational networks that provide them with social capital in the form of the information they receive, and their ability to engage in acquisitions. This study provides social context by investigating whether the interorganizational networks associated with CEO backgrounds affect their firms' acquisition activities.

We examine CEO background rather than the background of the entire top management team (TMT) because we expect that the number and strength of interorganizational network ties is greater for CEOs than for other members of the top management team. CEOs are more likely than other top management team members to be on corporate boards and to participate in other arenas in which they


might learn about acquisition targets (Useem 1982). Thus, examining only the CEO is adequate to test the background effects in which we are interested. Our approach is also consistent with several recent studies of the effects of managerial background on firm action. For example, studies of M-form adoption (Fligstein 1985; Palmer et al. 1993), diversification activities (F1igstein 1990, 1991) and acquisitions (Davis et al. 1994) have looked at the effects ofthe CEO, not the entire top management team.

Our theoretical arguments are organized as follows. Educational and functional backgrounds are demographic indicators of managerial networks. We start by discussing the effects of CEO educational background on corporate acquisitions, focusing on the network effects of an elite graduate degree. Second, we discuss the effects of CEO functional background on acquisitions, focusing on the network effects of an output and peripheral function background. We conclude by discussing how these background effects are likely to vary with decision uncertainty and power.

HYPOTHESIS DEVELOPMENT

CEO Educational Background and Corporate Acquisitions We argue that the interorganizational networks that come with graduate degrees from elite schools provide executives with information about acquisition targets and access to financial resources necessary to engage in acquisitions. These networks thus provide corporate social capital, allowing these CEOs to engage in certain types of acquisitions. Graduate degrees from elite schools have been said to provide individuals with lifelong business contacts among the corporate elite and there is evidence that these contacts affect firm actions (Useem 1982; Palmer, Jennings, and Zhou 1993). Only one study, however, has empirically examined the relationship between elite education and firm strategic actions. Palmer et al. (1993) proposed those graduates of elite business schools are especially likely to maintain contacts with each other and thereby learn about accepted business practices such as the adoption of the multidivisional form (M-form). They found a positive relationship between a CEO having an elite business school degree and a firm's likelihood of adopting the M-form.

Contacts formed during an elite graduate education may also affect acquisitions. As discussed by Palmer et al. (1993), these ties may provide information about business trends. Elite ties are thus likely to provide information about acquisitions, either general information about the appropriateness of acquisitions as a strategy or specific information such as information about the availability and suitability of particular acquisition targets. Elite education also shapes corporate social capital that may provide access to the financial resources and social networks necessary to engage in acquisitions. Useem and Karabel (1986) showed that a degree from an elite graduate school facilitates corporate ascent. They also found that holders of elite degrees were more likely than graduates of nonelite schools to become directors of multiple corporations and leaders of major business associations. Thus, elite education provides social ties with others that control large firms (Useem and Karabel 1986; see also Useem 1979; D'Aveni 1990; Palmer, Jennings, and Zhou 1993).

The information and interorganizational corporate social capital provided by elite ties are likely to be especially important for domain-expanding (conglomerate and vertical) acquisitions. Conglomerate acquisitions occur when firms buy


unrelated firms in industries that are neither potential buyers, suppliers, competitors, nor complements to the acquiring firm's existing business. Vertical acquisitions occur when firms acquire their suppliers or distributors. Conglomerate and vertical acquisitions are domain-expanding acquisitions, because the firm is entering entirely new lines of business, or moving up or down the supply/distribution chain.

The contacts provided by an elite education are likely to be helpful for CEOs engaging in domain-expanding acquisitions for two reasons. First, because these contacts are likely to cross industry boundaries (Mintz and Schwartz 1985), CEOs with these contacts are probably more able to identify acquisition targets outside of the industries of their current business than CEOs without these contacts. Second, consolidating firms across sectors is likely to require more power and control in the intercorporate network than consolidating firms within sectors, and CEOs with elite degrees are particularly likely to have this power (Useem and Karabel 1986). We thus hypothesize: H1: Firms in which the CEO received an elite graduate degree will be more likely

to engage in domain-expanding acquisitions than firms in which the CEO did not receive an elite graduate degree.

CEO Functional Background and Corporate Acquisitions Following Hambrick (1981), and Hambrick and Mason (1984), functional backgrounds can be classified into four general categories. Output junctions, which include marketing, sales, and product R&D, involve an emphasis on growth and the search for new opportunities. Throughput junctions, which include production, process engineering, and accounting, involve a focus on improving the efficiency of the input-to-output transformation process. Peripheral junctions, which include law and finance, are areas that are not integrally involved in a firm's core activities (cf. Hayes and Abernathy 1980). Finally, general management involves a concern with the functioning of the firm as an integrated whole and a focus on what businesses the firm should be in (Andrews 1971; Asnoff 1965).

There are several possible mechanisms through which CEO functional background might affect patterns of acquisitions. To date, researchers have focused on the effects of functional background on managerial power and managerial cognition. According to the power perspective, CEO functional background affects acquisition strategies because individuals who have the skill to engage in acquisitions gain power when acquisitions will help the firm deal with critical strategic contingencies (Pfeffer 1981; Fligstein 1990, 1991). Consistent with this perspective, Fligstein found that the functional background of the CEO varied with critical contingencies that firms faced over the past century (Fligstein 1990).

According to a cognitive perspective, functional background causes individuals to interpret an organization's problems and solutions to those problems in characteristic ways (Dearborn and Simon 1958). Fligstein (1991), for example, proposed that executives with finance backgrounds possess a 'finance conception of control,' in which the firm is viewed as a bundle of assets to be bought and sold, and the finance conception of control has been prevalent since the 1960s. Consistent with this, Fligstein (1991) found that firms with finance CEOs engaged in more conglomerate acquisitions in the 1960s and 1970s than firms with CEOs without finance backgrounds. Using similar logic, Mizruchi and Stearns (1994) found that


firms with finance CEOs tended to use more external borrowing as a financing strategy than firms whose CEOs had other backgrounds.

However, we propose that the effects of CEO functional background go beyond power and cognition. Functional background may also lead to systematic patterns of external corporate ties, resulting in opportunities to engage in certain strategies. The functional backgrounds that are most likely to be associated with external ties that affect acquisitions are output and peripheral functions. Yet, these two backgrounds are associated with different types of ties: the social capital inherent in CEO networks varies with the CEO's functional background. Peripheral function CEOs are likely to have wide-ranging networks, including somewhat extensive contacts in the investment banking and legal communities. These contacts are likely to provide access to acquisition opportunities that are based more on capturing financial value than on creating operating synergies. Thus peripheral function CEO are more likely to do conglomerate acquisitions since their networks produce information and access to target firms in unrelated industries. H2a: Firms in which the CEO has a peripheral function background will be more

likely to engage in conglomerate acquisitions than firms having CEOs with other backgrounds.

CEOs with output function backgrounds have spent much of their careers in marketing and sales oriented jobs. These jobs require extensive interactions with suppliers and distributors. Thus, output function CEOs are likely to have many more contacts with firms in distributor (also possibly supplier) industries than CEOs with other backgrounds. These contacts in distributor industries are likely to facilitate vertical acquisitions--drawing attention to targets in these industries and producing contacts that facilitate acquisitions in these industries. Thus, the interorganizational links associated with an output background are likely to affect the number of vertical acquisition targets noticed and to improve access to those targets through social ties. If they do, then we expect that CEOs with output function backgrounds are more likely to engage in vertical acquisitions than CEOs with peripheral backgrounds. H2b: Firms in which the CEO has an output function background will be more likely

to engage in vertical acquisitions than firms in which the CEO has a peripheral function background.

Moderating Effects of Decision Uncertainty The relationships between networks and acquisitions outlined above are likely to be moderated by an important factor: uncertainty. While other possible moderators exist, we chose to study uncertainty because it is suggested by current theory, yet has not been studied in this context. For example, uncertainty has been shown to moderate the impact of interorganizational networks on strategic action (e.g., Haunschild 1994).

Several theories support the idea that uncertainty surrounding a decision is likely to affect that decision in two ways. First, uncertainty causes individuals to rely on cognitive shortcuts and heuristics (Cyert and March 1963; Hambrick and Mason 1984), such as those produced by one's educational and functional background. Second, uncertainty increases the extent to which individuals engage in social comparison (Festinger 1954). This means that managers experiencing uncertainty are likely to look to others outside their organization for ideas and information about business practices. Consistent with this latter idea, Haunschild (1994) found that


uncertainty about acqUisItion targets causes firms to establish more interorganizational relationships, and to be more influenced by the information obtained through these relationships. Both the cognitive shortcut and social comparison arguments suggest that the interorganizational networks that arise from educational and functional background should be more influential under conditions of uncertainty. Thus, the acquisition decisions of CEOs experiencing a high level of uncertainty should be more affected by the CEO's educational and functional backgrounds than the decisions of CEOs experiencing a low level of uncertainty. H3: The effects of educational and junctional background on acquisition activity

are strengthened by firm uncertainty.

Moderating Effect of CEO Tenure CEO tenure should weaken the relationship between CEO background and acquisition strategies. As tenure increases, a CEO's management paradigm is likely to become more fixed and less open to diverse input (Finkelstein and Hambrick 1990; Miller 1991; Hambrick and Fukutomi 1991; Davis, Tinsley, and Diekmann 1997). Consistent with this, studies have found that CEO tenure is associated with strategic persistence (e.g., Finkelstein and Hambrick 1990). One reason strategic persistence occurs is that information accessed for decisions tends to become routinized over time, reflecting past experience more than new information (Katz 1982). This means that long-tenured CEOs are likely to become more parochial in their decision making, relying more on internal organization factors, and less on external factors. If this is true, then the interorganizational networks associated with an elite education and functional backgrounds should have weaker effects on longtenured CEOs. H4: The effects of educational and junctional background on acquisition activity

are weakened with CEO tenure.

METHOD

Sample The sample for this study consisted of all firms completing an acquisition between 111186 and 7/15/93. The acquisition had to meet the following four criteria: 1) the acquiring firm bought a controlling interest in the target firm; 2) both acquirer and target were U.S. based, publicly held companies; 3) data about the acquiring firm's CEO was available; and 4) at least ten acquisitions occurred in the acquiring firm's industry during the period studied. The fourth criterion was used to allow us to control for industry effects. The following 13 industries were included in the sample: oil and gas extraction; food and kindred products; chemicals and allied products; fabricated metal products; electrical and electronic machinery; measuring, analyzing and controlling instruments; lum6'er and wood products; transportation equipment; communication; electric, gas and sanitary services; insurance; banking; and business services. Including only acquisitions for U.S. based, publicly held companies was necessary to insure the availability of data on CEO background and several of the control variables. The data were obtained from the Merger and Corporate Transaction Database maintained by Securities Data Corporation (hereafter, SDC). This produced a sample of 271 acquisitions, completed by 228 firms. To rule out the possibility that results of our tests for different types of acquisitions were affected by the fact that only firms completing acquisitions were


in these analyses, we also collected a sample of nonacquirers. The 221 nonacquiring firms were randomly selected from the 1985-1992 Business Week Corporate Elite survey, producing a final sample of 449 firms, 221 of which did not complete acquisitions, and 228 of which completed one or more.! We then tested for effects of sample selection bias using methods described in Berk (1983). The results of these tests show no evidence of selection bias due to including only the acquirers in the analyses (results available from the authors).

Measures A dummy variable for each firm was coded one if the firm completed one or more acquisitions during the study period, zero otherwise. For each acquisition, three dummy variables were created to indicate whether the acquisition was related, vertical, or conglomerate. Each variable was coded one if the acquisition was of that type, zero otherwise. A classification scheme similar to those used in other studies of acquisitions (e.g., Davis, Diekmann, and Tinsley 1994) was used to classify acquisitions. An acquisition was coded as related when the two-digit SIC code of the acquiring firm matched that of the acquired firm. SIC codes were obtained from the SDC database. An acquisition was coded as vertical when the industry of the acquiring firm either sold more than 5% of its output to, or received more than 5% of its input from the industry of the acquired firm. Input-output numbers were obtained from the U.S. Department of Commerce' s Survey of Current Business. All remaining acquisitions were coded as conglomerate.

Explanatory Variables CEO biographical data were obtained from Business Week's Corporate Elite survey, which includes CEOs of the Business Week 1000 (Business Week 1986-1993). These data were obtained by questionnaire from either the CEO or a designated corporate representative. The biographical data include the CEO's self-reported education, primary career path, and tenure as CEO. While there are limitations to this type of self-report data, any errors in such reporting are unlikely to be systematically related to acquisition activity.

The educational data from the Business Week survey were used to construct dummy variables for education level (undergraduate, masters, PhD), education type (MBA, other), and whether the CEO has a graduate degree from one of the eleven elite schools identified by Useem and Karabel (1986). As preliminary support for the idea that CEOs with elite graduate degrees have more network ties than CEOs with nonelite degrees (see HI), we obtained data on the interlocks of the CEOs included in this study. CEOs with elite graduate degrees participate on more boards than CEOs without elite graduate degrees (F(1,245)=3.98, p<.05). As interlocks are only a subset of all possible interorganizational networks, this data does not provide conclusive evidence that elite degrees are associated with more interorganizational networks, and thus we use the elite degree dummy variable rather than the number of interlocks in the analyses.

The primary career path data in the Business Week survey were used to construct dummy variables for output function, throughput function, peripheral function, and general management backgrounds. Each CEO's primary career path was used to construct the dummy variables. For example, if a CEO's primary career path was finance, the peripheral variable was coded 1, and the output, throughput,


and general management variables were each coded O. In all analyses, peripheral function background is the omitted category. As preliminary support for the idea that CEOs with output function backgrounds have more network ties to firms in vertically related industries, we obtained data on the interlock ties of CEOs. We then classified these interlocks according to whether they were to firms horizontally related, vertically related, or unrelated to the CEO's own firm, using the classification scheme described earlier. We then tested to see whether CEOs with output-function backgrounds sit on more boards of firms that are vertically related to their own firm than CEOs with other backgrounds. Results show CEOs with output function backgrounds do tend to sit on boards of firms that are vertically related to their own industry (F(1,487)=3.80, p<.05). They also tend to sit on boards of firms that are unrelated to their own industry (F(1,487)=4.21, p<.05). They tend to not sit on boards of firms that are horizontally related. Thus, output function background seems to be associated with sitting on boards of firms that are not closely related to one's own firm. We also found that CEOs with peripheral backgrounds tend to sit on boards of firms that are horizontally related to their own firms (F(1,487)=4.23, p<.05). This seems somewhat inconsistent with our idea that CEOs with peripheral function backgrounds have more ties to the investment banking and legal communities. Since interlock ties are only a subset of all possible ties among members of different firms, however, this result may reflect that investment banker and legal ties are not captured well in interlock data.

CEO tenure was measured as years as CEO. Uncertainty was operationalized as the variance of opinion regarding the value of the acquiring firm. While this cannot be directly observed, a proxy measure was obtained from the IIBIFJS database of Lynch, Jones and Ryan. Lynch, Jones and Ryan monitor the earnings per share (EPS) estimates produced by research analysts from leading brokerage firms for over 2000 companies. The variance of these estimates reflects the dispersion of opinion among analysts regarding the future performance of a company. If analyst estimates vary, then there is a lack of agreement or clarity about the underlying facts affecting the firm, which is likely to reflect the uncertainty experienced by their management. Similar to Haunschild (1994), we measured each firm's uncertainty as the coefficient of variation of the analyst's projected acquirer EPS estimates for that firm.2

Control variables Six variables were included to control for factors related to the likelihood of a firm completing acquisitions of various types. These variables were the size, profitability, and industry of the acquiring firm, CEO age, CEO education level, and the year the acquisition occurred. The size and profitability of the acquiring firm might affect either the likelihood of an acquisition or the type of acquisition undertaken. Size and profitability are important indicators of the existence of resources needed to undertake an acquisition. In addition, there is some evidence that size and profitability are related to the likelihood of conglomerate acquisitions (Davis et al. 1994; Haunschild 1993). Size was measured as a firm's annual sales. Profitability was measured as return on equity. Both size and profitability were obtained from the Business Week listing. Industry controls were included to control for the fact that patterns of acquisition vary by industry (Pfeffer and Salancik 1978), and

~ CD

Tab

le 1

. Des

crip

tive

stat

istic

s an

d co

rrel

atio

ns a

mon

g ke

y st

udy

vari

able

sa

Mea

n S

.D.

Min

M

ax

~Q

~22

P2

~42

~52

~62

~72

~82

~92

(102

~1

12

~122

~1

32

~142

~1

5) ~

16)

~17)

I.

Acq

uire

r (I =

yes,

.5

55

.497

0

I

0=00

)

2.

Con

glom

erat

e .2

19

.414

0

n1a

Acq

uisi

tion

3.

Ver

tical

.1

02

.303

0

n1a

-.179

A

cqui

sitio

n 4.

R

elat

ed

.670

.4

71

0 n1

a -.7

56

-.482

A

cqui

sitio

n 5.

D

omai

n-E

xpan

ding

.3

23

.468

0

n1a

.756

.4

90 -

.983

A

cqui

sitio

n 6.

M

BA

(no

oelit

e)

.112

.3

15

0 .1

84 -

.180

-.1

34

.164

-.1

64

7.

Elit

e G

rad.

Deg

ree

.290

.4

54

0 .0

33

.105

.1

28 -

.078

.0

78

-.227

(i

ncI.M

BA

) 8.

E

lite

Gra

d. D

egre

e .0

83

.276

0

I -.0

70

.129

.2

02 -

.222

.2

22

-.107

.4

71

(cxc

l.MB

A)

9.

Elit

e M

BA

.2

07

.405

0

I .0

85

-.068

.0

09

.044

-.0

44

-.18

1 .7

99 -

.154

' 10

. O

utpu

t Bac

kgro

und

.376

.4

85

0 I

-.15

1 .0

63

.164

-.1

68

.168

-.0

36

-.12

8 -.0

22

-.128

II

. T

hrou

ghpu

t Bkg

. .0

65

.248

0

.036

.0

88

-.002

-.0

72

.072

-.0

68

-.025

-.0

50

.007

-.2

06

12.

Perip

hera

l B

kg.

.359

.4

80

0 I

.141

-.1

39

-.114

.1

92 -

.192

.0

43

.136

.0

65

.108

-.5

82

-.198

13

. G

ener

al M

gmt B

kg.

.198

.3

99

0 I

.112

.0

35

-.047

.0

08 -

.008

.0

34

.007

-.0

21

.022

-.3

87

-.132

-.3

73

14.

CE

O T

enur

e 8.

14

6.80

0

38 -

.012

.1

25

.028

.1

24 -

.124

-.1

28

.103

.0

91

.053

-.0

66

-.032

-.0

57

.168

15

. U

ncer

tain

ty

10.6

6 27

.18

o 33

6.5

n1a

.018

.1

57

-.114

.1

14 -

.037

-.0

91

.145

-.1

05

.172

.0

35

-.144

-.0

45

.170

16

. C

EO

Age

55

.41

7.18

34

72

-.1

28

.047

.0

68 -

.101

.1

01

-.133

-.0

40

.077

-.0

97

.050

-.0

01

-.017

-.0

41

.298

.0

12

17.

Sale

s 34

34.4

6 58

85.0

7 19

.00

7160

0 .1

37

.184

.0

26 -

.172

.1

72 -

.568

.0

59 -

.012

.0

74

.090

.0

26 -

.068

-.0

43

-.007

-.0

39

.145

18

. Pr

ofits

21

9.59

42

1.38

-1

100

4600

.1

62

.105

.0

05

-.088

.0

88

-.019

.0

29 -

.060

.0

74

.108

.0

13

-.068

-.0

58

-.049

-.

101

.124

.8

30

a C

orre

latio

ns g

reat

er th

an .

11 a

re s

tati

stic

ally

sig

nifi

cant

at t

he .0

5 le

vel.

~ ~ g ~ ::r

n·

'" 8- ~ .0

5.

~.

a.

o :s '" !j

U\


CEO functional background may also vary by industry. Thus, in the absence of industry controls, an observed relationship between CEO background and patterns of acquisitions may be spurious. Industry was measured by a series of thirteen dummy variables for the 2-digit SIC codes described earlier. CEO age controls for the possibility that age is correlated with interorganizational networks or CEO tenure. CEO age was obtained from the BW listings. CEO education level (undergraduate degree, graduate degree, MBA) was included to control for correlations of CEO networks with education level. Year of acquisition controls for any macroeconomic factors, e.g., tax changes, GNP, that may affect the likelihood of acquisitions, or certain types of acquisitions.

RESULTS

Table 1 presents descriptive statistics and correlations for all study variables except year and the industry of the acquiring firm.

Table 2 presents the results of logistic regressions on whether a firm engaged in a domain-expanding acquisition during the study period. For these models, the dependent variable was coded 1 if the firm engaged in a domain-expanding (vertical or conglomerate) acquisition, 0 if they engaged in a related acquisition. Modell of Table 2 presents the effects for the control variables alone. As can be seen from this model, larger firms are more likely to do domain-expanding acquisitions, less profitable firms are more likely to do domain-expanding acquisitions, and there were fewer domain expanding acquisitions during 1987, 1989, 1990, 1991, and 1992 than 1986 (the excluded year). Model 2 adds the effects of functional background. There are no statistically significane relationships between CEO functional background and the likelihood of a firm engaging in a domain-expanding acquisition.

Model 3 reports the results of the analysis of whether firms whose CEOs have an elite graduate degree engage in more domain-expanding acquisitions than firms whose CEOs do not have an MBA (HI). As predicted by Hypothesis I, CEOs with elite graduate degrees were more likely to engage in domain-expanding acquisitions than CEOs without elite graduate degrees. Model 3 reports results for elite graduate degrees excluding elite MBAs. When we include elite MBAs with the other elite graduates, the results are the same-the effect of an elite graduate degree on domain-expanding acquisitions is positive (see Model 4). We thus combine elite graduate degree and elite MBA in all further analyses. Additionally, supplementary analyses reveal that an elite undergraduate degree produces the same effects-CEOs with elite undergraduate degrees were more likely to engage in domain-expanding acquisitions than CEOs with nonelite undergraduate degrees (details of these analyses available from the authors). It appears that something unique to obtaining an elite degree, which is not restricted to an elite graduate degree, has an effect on engaging in domain-expanding acquisitions. This 'something' is likely to be the social capital that resides in networks that come with elite education. Such networks may be more powerful in graduate schools, but should also come with undergraduate education.

While we had not hypothesized any effects for education level (undergraduate, MBA), Model 3 also shows that firms whose CEOs have a non-elite MBA engaged in fewer domain-expanding (and thus more related) acquisitions than firms whose CEOs do not have an MBA.4 While we report all three educational variable results


Table 2. Logistic regression analysis of determinants of domain-expanding acquisitionsa

Variable (I) (2) (3) (4) (5) (6)

Output Background -.00 -.12 -.12 -.11 -.19

Throughput Background .20 .\3 .09 .18 .01

Peripheral Background -.01 -.10 -.12 -.05 -.11

Elite Grad. Degree (exc!. MBA) 1.49**

Elite MBA .35

MBA (nonelite) -.98* -1.00* -1.32* -1.32*

Elite Grad. Degree (inc!. MBA) .81** .84 1.29*

Uncertainty -.00

Uncertainty x Elite Grad. -.03

Uncertainty x MBA -.06

CEO Tenure .02

Tenure x Elite Grad. -.08*

TenurexMBA JJ7**

Control Variables"

CEO Age .02 .02 .01 .01 .00 .01

Year-1987 -1.31* -1.30* -1.25* -1.32* -1.58* -1.42*

Year-1988 -.63 -.62 -.57 -.67 -.58 -.79

Year-1989 -1.24* -1.23* -1.22* -1.35* -1.31* -1.47*

Year-1990 -1.37* -1.38* -1.22 -1.33* -1.25 -1.30*

Year-I991 -1.46* -1.45* -1.23 -1.36* -1.57* -1.44*

Year-1992 -2.45** -2.46** -2.32** -2.49** -2.41** -2.56*

Year-I993 -1.01 -.99 -.64 -.75 -.65 -.84

Sales (thousands) .13* .13* .12* .11* .12* .11*

Profits (thousands) -1.70* -1.70* -1.60* -1.50* -1.40* -1.50*

Intercept .06 -2.36** .38 .50 .91 .27

N 273 273 273 273 262 273

Chi-squared 97.94** 97.87** 107.31** 105.52** 99.65** 118.29**

-2 Log Ukelihood 248.4 248.3 238.8 240.6 229.1 227.8

% Cases Correcdy Classified .83 .83 .85 .85 .84 .86

*p<.05;**p<.OI, I-tailed tests for independent variables, 2-tailed tests for control variables.

aDependent variable is whether the focal firm engaged in a related acquisition (l=yes, O=no)

bResults for industry control variables are not reported, but are available from the author

in the same model (Model 3), separate analyses show that the statistical significance of the results does not change when these variables are run separately. MBA is negatively related, elite degree is positively related, and elite MBA has no statistically significant effect on the likelihood of engaging in a domain-expanding acquisition. Thus, we find evidence supporting a network perspective on education (our elite graduate degree results). The elite MBA case, where cognitive and network predictions oppose each other (elite education dictates more domainexpanding acquisitions and MBA dictates more related acquisitions) produces no statistically significant effect. This may be due to the opposing effects cancelling


each other in this situation. For our purposes, it is important to note that the network effects are independent of the cognitive effects, supporting the validity of our network perspective.

Model 5 of Table 2 examines whether uncertainty moderates the effects of an MBA and an elite graduate degree on acquisitions (H3). Model 5 shows, contrary to Hypothesis 3, that the main effects of education on acquisitions are hardly moderated by uncertainty. In support of Hypothesis 4, however, Model 6 shows that while firms whose CEOs have an elite graduate degree are more likely to engage in domain-expanding acquisitions, this effect is weaker for long-tenured CEOs. Tenure also weakens the negative relationship between a nonelite MBA and domainexpanding acquisitions found in Model 3. Thus, it appears that tenure, but not uncertainty, moderates the relationship between education and domain-expanding acquisitions.

Models 1-3 of Table 3 present the results of analyses of vertical acquisitions. For these models, the dependent variable was coded 1 if the acquisition was vertical, and 0 if it was related or conglomerate. Model 1 shows, as predicted by Hypothesis 2b, firms having CEOs with output function backgrounds are more likely to engage in vertical acquisitions than firms having CEOs with other backgrounds. Model 1 also replicates the effect of elite educational background (HI) (elite graduate degrees are positively related to vertical acquisitions).

Models 2 and 3 examine whether uncertainty and CEO tenure moderate the effects of output background on vertical acquisitions (H3 and H4). Model 2 shows, in support of Hypothesis 3, the main effect of an output background on vertical acquisitions is strengthened by uncertainty. While CEOs whose social capital is shaped by output backgrounds are more likely to do vertical acquisitions, they are even more likely to do so under conditions of uncertainty. Model 3 shows, contrary to Hypothesis 4, that tenure has no statistically significant effect on the relationship between output function background and vertical acquisitions. Thus, functional background effects on vertical acquisitions are affected by uncertainty, but hardly by tenure.5

Models 4-6 of Table 3 present the results of analyses of conglomerate acquisitions. For these models, the dependent variable was coded 1 if the acquisition was conglomerate, and 0 if it was related or vertical. Model 4 tests whether firms whose CEOs have social capital that is shaped by peripheral function backgrounds engage in more conglomerate acquisitions than firms whose CEOs have other functional backgrounds. Model 4 shows, contrary to Hypothesis 2a, firms with CEOs that have peripheral function backgrounds are no more likely to engage in conglomerate acquisitions than firms with CEOs that have other backgrounds. Model 4 also replicates the effects of elite educational background (HI) (elite graduate degrees are positively related to unrelated acquisitions). Thus, the positive effects of an elite graduate degree apply to both vertical and conglomerate acquisitions. Models 5 and 6 of Table 3 show that uncertainty and tenure have no statistically significant effect on the relationship between peripheral background and conglomerate acquisitions. Thus, Hypotheses 3 and 4 get no support when looking at conglomerate acquisitions alone.

Overall, then, Hypothesis 3 gets mixed support from these results. Uncertainty does not appear to weaken the effects of educational background. It does not weaken


Table 3. Logistic regression analysis of determinants of vertical and conglomerate acquisitions

Vertical Acquisitions ConglOlrerate Acquisitions

(I) (2) (3) (4) (5) (6)

Output Background .19" .18" .17"

Peripheral Background .07 .34 .38

Elite Grad. Degree (incl. MBA) .91" 1.33" .n" .OS" .14 .08"

MBA (nonelite) -1.24 -1.25 -1.26 -.66"· -.66* -.67*

Uncertainty .01 .00

Uncertainty x Periph. Bkg. -.04

Uncertainty x Output Bkg. .05"

CEO Tenure .03 .01

Tenure x Output Bkg. -.01

Tenure x Peripheral Bkg. -.02

Control Variables'

CEO Age -.02 -.02 -.65 -.01 -.00 -.01

Year - 1987 -.52 -.85 -.67 -.66 -.92 -.71

Year- 1988 -.53 -.66 -.58 .09 .08 .04

Year- 1989 -.39 -.62 -.66 -.79 -.86 -.75

Year- 1990 -.45 -.51 -.66 -.66 -.80 -.69

Year- 1991 -.60 -1.04 -.70 -.61 -.72 -.70

Year- 1992 -1.27 -1.18 -1.29 1.05 -1.77 -1.04

Year - 1993 -.62 .82 .39 .63 -.86 -.63

Sales (thousands) .01 .02 .OJ .11* .12 .12*

Profits (thousands) 1.70 2.00 1.90 1.40 1.20 1.10

Intercept 2.59 .57 .84 .26 .25 .32

N 273 262 273 273 262 273

Chi-Squared 44.98"* 40.86** 45.94*" 62.10"* 58.42** 61.86·

-2 Log Ukelihood 135.6 123.2 134.6 225.4 216.0 225.7

% Cases Correctly Classified .83 .83 .84 .80 .80 .79

*p<.05, * *p<.O I, 2-tailed tests.

IResults for industry control variables are not reported, but are available from the authors.

the relationship between functional background and conglomerate acquisitions_ It does, however, weaken the relationship between functional background and vertical acquisitions. Hypothesis 4 also gets mixed support. Increasing CEO tenure weakens the effects of educational background on domain-expanding acquisitions. Tenure, however. does hardly seem to affect the relationship between functional background and vertical or conglomerate acquisitions.

DISCUSSION

Overall. the results provide support for the idea that the networks that come with educational and functional backgrounds affect acquisitions. These background


effects are net of many control variables. We found evidence that the networks that come with elite graduate degrees and certain functional backgrounds affect acquisitions. These network effects are independent of the cognitive effects of such backgrounds. We expected that an elite graduate degree would provide CEOs with access to financial resources and far-reaching contacts in the intercorporate network. In turn, we argued that these contacts would aid firms in completing domainexpanding acquisitions (conglomerate or vertical acquisitions that take firms into entirely new areas). Consistent with these arguments, we found that firms having CEOs with elite degrees were more likely to engage in domain-expanding acquisitions than those without such degrees. When transacting across industries with unfamiliar actors of unknown reputation, an elite degree may act as an important signal, both to target firms and potential investors, that the acquirer's CEO is competent (D' Aveni 1990). The social capital inherent in the networks that come with an elite degree may also provide information about acquisition opportunities in different industries, information not available to those without such networks. Alternatively, this finding may suggest that something about CEOs that obtain elite degrees is related to such acquisitions, e.g., an ability to think more globally about possible opportunities in other industries.

We argued that the social capital that resides in the networks of CEOs with different functional backgrounds, would result in different types of acquisitions. Indeed, we find support for the idea that output function work history is systematically related to external networks and acquisitions. We expected that output background would increase a CEOs awareness of and access to targets in distributor and also possibly supplier industries. Consistent with this idea, we found that CEOs with output backgrounds engage in more vertical acquisitions than CEOs with other backgrounds. Such CEOs also tend to sit on boards of firms that are unrelated to their own firms. We are not aware of any studies looking at the relationship between functional background and interorganizational networks. Our results suggest that such networks may have important effects on firm strategic actions.

We had also, however, predicted that CEOs with peripheral function backgrounds would be more likely to do conglomerate acquisitions than CEOs with other backgrounds. We had predicted this on the assumption that peripheral backgrounds would be associated with interorganizational networks that provide ties to firms very different from the CEO's own firm, including ties to others in the investment banking and legal communities. Our network analyses, however, showed that finance CEOs tend to have ties to firms similar to their own firms.

Many of the CEOs with peripheral function backgrounds come from finance. Taking a cognitive perspective on the relationship between functional background and strategy, Fligstein (1990) argued that a finance background produces a 'finance conception of control,' in which a firm is viewed as a bundle of assets to be bought and sold. Because of their experience with financial analysis, peripheral function CEOs prefer to evaluate and manage unrelated businesses, which requires running businesses as a portfolio of assets. Fligstein found that finance CEOs tended to complete more conglomerate acquisitions in the 1960s (Fligstein 1990). We, on the other hand, find no evidence that CEOs with peripheral function backgrounds completed more conglomerate acquisitions in our sample, even when we break out finance backgrounds from law and other peripheral background categories. So


peripheral backgrounds are not associated with networks to firms in different industries, and peripheral and/or finance backgrounds seem not to be associated with engaging in certain types of acquisitions. We may have obtained different results from Fligstein because the institutional pressures against conglomerate acquisitions in the 1980s overcame any tendencies toward acting in accordance with a finance conception of control, suggesting that perhaps the finance conception of control has declined in importance in recent years.

Moderators of Background Effects Our hypotheses for the effects of uncertainty on the relationship between background and acquisitions were supported for output functional background, but not for educational background. It may be that the reliance on cognitive shortcuts and heuristics proposed to result from uncertainty only applies to functional background because, when faced with uncertainty, functional background is more salient to decision makers. Education occurred in the more distant past than the experiences provided by working at a functional specialty. The social ties and norms decision-makers have been exposed to on the job, which have been reinforced over the years, are more salient than those gleaned from formal educational experience. When faced with uncertainty, decision makers may indeed fall back on cognitive shortcuts and social comparison processes (e.g., Festinger 1954; Cyert and March 1963; Haunschild 1994); however, recency and salience may dictate the set of referents chosen.

We found that tenure weakened the effects of educational background, but had no effect on functional background. The attenuation of educational effects over time is consistent with our earlier arguments about long-tenured CEOs becoming less sensitive to external cues (like external trends, norms, and ties with others), and more attuned to their own views of a particular firm and industry than short-tenured CEOs (see Hambrick and Mason 1984; and Davis, Tinsley, and Diekmann 1997 for similar arguments). The attenuation of educational effects may also be due to a weakening of the effects themselves-network contacts from one's elite education weaken over time. This does not appear to be the case, however, for the contacts from one's functional experience. Such contacts may be more frequently used, resulting in effects that are more stable over time. Future research investigating these different forms of ties and their stability and influence over time would be valuable.

Limitations and Future Research Some of our most striking findings concern the relationship between education and acquisitions. An alternative explanation for our education effects, however, is that firms that are likely to engage in acquisitions, or certain types of acquisitions (for whatever reason), hire or promote CEOs with MBAs or elite degrees. However, we think that this alternative explanation is unlikely because we controlled for several factors that affect both acquisitions and the selection of CEOs with certain backgrounds. For example, industry is likely to be correlated with the tendency to pursue acquisitions and also with the tendency to select CEOs with certain backgrounds. Additionally, if a firm were to hire (or promote) a CEO because the CEO's background matched the acquisition desired, there should be a relatively short period between the time the CEO was hired and the time the acquisition was


completed. This means that we should find effects only for short-tenured CEOs. Yet we control for the interaction of CEO tenure and background, and still find effects for background, making this alternative explanation less likely.

One limitation of our study is that top managers do not always have discretion to act in accordance with their preferences. Although we find evidence that top manager demographics affect acquisitions, there may be situations where these demographic factors are more or less important. For example, the agency theory literature shows that incentives can induce managers to act in accordance with shareholder interests. Thus there may be situations where, for example, CEO networks or normative preferences would result in a related acquisition, but the CEOs incentives overcome such preferences and result in a conglomerate acquisition. Boards of directors may likewise deter a CEO from acting in accordance with his or her preferences.6 Thus, the moderating effect of factors like CEO incentives and board power on the relationship between cognition, networks, and acquisitions is an important topic for future research.

A second limitation of this study is that we have somewhat indirect indicators of managerial networks. While we go further than many demographic studies in actually measuring the intervening process variables between demographics and firm outcomes, our interlock measures can only capture a subset of all managerial networks. Showing that output background is associated with more ties to firms in vertically related industries is important in supporting the validity of our networkfunctional background relationships. However, interlock ties are probably better indicators of the output background networks than they are of peripheral background networks. This is because the presence of investment bankers and attorneys on firm boards is relatively rare in our sample, so such ties are probably imperfectly captured in the interlock data. It would be useful, therefore, to be able to more directly measure the relationship between demographic backgrounds and their associated networks.

Our results show that the relationship between functional background and firm outcomes is strengthened with uncertainty. The relationship between educational background and firm outcomes, on the other hand, is unaffected by uncertainty and weakened by tenure. The moderators studied, uncertainty and tenure, have different effects on education and functional background, suggesting that studying additional moderators would be useful.

CONCLUSION

We sought to assess the effects of educational and functional background on corporate acquisitions. Educational and functional backgrounds are demographic variables, and demographic variables have been proposed to influence organizational outcomes by affecting managerial cognition (e.g., Dearborn and Simon 1958; Hambrick and Mason 1984; Fligstein 1990, 1991) and managerial power (Fligstein 1990, 1991). Overall, our data are consistent with an alternative view of the demographic variable-organizational outcome relationship: that demographic variables are indicators of interorganizational networks and access to information, which in turn affects outcomes like corporate acquisitions. Networks give rise to interorganizational social capital, which directs acquisition activities along certain channels. This is an important finding for both demography and


network research, demonstrating the link between demographic variables, networks, and firm outcomes.

Our results also add to the growing literature showing that social context affects acquisition activities (e.g., Fligstein 1990; Haunschild 1993; Davis et al. 1994). We find evidence that firm managers carry with them interorganizational social networks that affect their attention to information, and ability to engage in acquisitions. Returning to our original question of what affects a firms choice of strategic actions, we find evidence for the effect of both external (interorganizational networks) and internal (managerial preferences) factors. We thus link the internal and external perspectives on strategic action: external factors in the form of interorganizational networks are linked to internal factors (CEOs) and predict major strategic actions (acquisitions).

NOTES

I. Some of the finns in the sample completed multiple acquisitions. To test whether the nonindependence inherent in these transactions affected the results, all analyses were conducted twice: once with all acquisitions done by the multiple-acquisition finns included, and once where only one randomly-selected acquisition was included for these multiple acquisition finns. The results do not vary by whether one or all acquisitions for the multiple-acquisition finns was included, so results with all acquisitions are reported. 2. The measure's strengths lie in the fact that it is I) specific to the finn; 2) involves multiple respondents (analysts); and 3) allows uncertainty to be measured at the proper time prior to the acquisition. 3. Statements on statistical significance refer to the .05 level. 4. This result may be consistent with a cognitive argument for the effects of educational background on acquisitions. Graduate business schools tend to socialize students into accepted business practice (Schein 1968; Van Maanen 1983), resulting in cognitive preferences for some practices over others. Individuals with business degrees are likely to learn to be environmental scanners or monitors and thus prefer to adopt practices that are currently popular. Since related mergers and acquisitions were popular and nonnatively accepted during the period we study (Haspeslagh and Jemison 1991; Davis et al.l994), finns having CEOs with MBAs should be more likely to engage in related acquisitions than finns having CEOs without MBAs. 5. Uncertainty does not moderate the effect of an elite graduate education on vertical acquisitions, either (details of this analysis available from the authors). 6. See Walsh and Seward (1990) for a discussion of situations where boards might have more or less influence on management.

•

Public Service Organizations: Social Networks and Social Capital

ABSTRACT

15 Ewan Ferlie

There is evidence cited in the literature of a growing trend to network based styles of management, especially in high-technology or expert based organizations. For such network based organizations, the effective management of many external ties is a major invisible asset or form of social capital. It is unclear whether this social capital is owned by leading individuals or by organizations. The question of whether corporate forms of social capital can be created and sustained across the organization is in part an empirical one and not to be assumed through theory.

This chapter adds to the extant literature is two ways-first, by extending this form of analysis to public-service settings such as health care organizations and, second by providing recent qualitative empirical data. Public service organizations remain of substantial scale and significance in many OECD countries, and they need to be fully considered in the management literature.

Data are presented from U.K. health care organizations. An empirical study was undertaken of a set of health care purchasing organizations rated as 'leading edge' by peers. The data may not then reflect typical practice but rather as it is developing in an innovative form. A series of interviews and visits took place; full notes were taken and subject to contents analysis.

The study confirmed the importance of networks in health care management, so that the possession of rich ties was a key invisible asset in 'getting the business done.' The data does not confirm the conventional picture of total domination of professional networks as more managerial networks were also in evidence. Social capital belonged to individuals or autonomous groups at least as much as to the corporate organization, although there were instances cited of attempts to create more corporate forms of social capital.

Public Service Organizations: Social Networks and Social Capital - 285

However, the depth of the shift to network-based management can be questioned. The transition was only partial, recent in origin, and often mandated by the top. It is still premature to talk of a wholescale transition to network based management in health care, as managerial practice in health care in faddish and characterized by short life cycles.

INTRODUCTION

Markets, hierarchies, and networks are three ideal types of organizing (Thompson et al. 1991). As these are ideal types and not empirical descriptions, in practice mixed styles may be evident. There is evidence of a gathering trend toward more network based styles of management within private sector firms (Axelsson and Easton 1992; Nohria and Eccles 1992), especially in high-technology or expert-based organizations. Fenton's (1996) literature review suggested that the New Competition in the private sector may be resulting in widespread organizational transformations. Associated with this is a predicted emergence of the N form (or network form) organization with such 'signs and symptoms' as a flatter structure, self management, diffuse decision making, accelerated information flows, and an emphasis on learning.

For such network based organizations, the effective management of many external ties creates a major invisible asset or form of social capital. The quantity and quality of interorganizational ties between actors becomes a key factor of production (Coleman 1988). Of course, this social capital is often created and maintained by individuals rather than organizations. Effective organizations are those that create a degree of cohesive unity and corporate contribution out of these individually based networks. The depersonalization of network ties and their aligning with corporate objectives then becomes a form of corporate social capital creation (for a further discussion of the difference between social capital at the level of the individual level and the firm, see Pennings and Lee, this volume) resulting in a competitive advantage.

A reliance on interpersonal relationships may be supplemented by more formal links such as joint board membership. However, interpersonal exchanges may remain pervasive irrespective of formal structure, so we need to establish whether these are motivated by corporate or particularistic objectives (also see Smith-Doerr et al. this volume). Moving the focus of analysis from the individual to the organization poses not only theoretical difficulties but also management questions that need to be resolved empirically. Corporate social capital creation may be especially problematic within certain organizational forms; for example, those of very large size, those that display a range of occupational subgroups, where there is much conflict between management and other employees, or where front-line workers retain much autonomy over work practices, such as professional service firms (see Pennings and Lee' s discussion of social capital within audit firms in this volume). The organization may here be no more than a loose arena in which a large number of subunits and subcultures coexist. The question of whether such corporate forms of social capital can be created and sustained across the organization is in part an empirical one and not to be assumed through theory.


This chapter adds to the extant literature in two ways-first, by extending this form of analysis to public service settings such as health care organizations and, second, by providing recent qualitative empirical data.

The Public Services Setting First, the question arises as to whether similar shifts to network based forms of social capital are evident within public service organizations as in private firms. Public service organizations typically account for between 35 and 50 percent of GDP (this definition includes transfer payments) in many OECD countries. They discharge societally strategic functions such as the provision of health care, education, income support, and law and order. While these organizations represent important sites for analysis, they have not been fully explored in the managerial literature on networks, certainly in Anglo American contexts. The chapters in Nohria and Eccles (1992), for example, do not explore whether such shifts are specific to the private sector settings analyzed.

One argument is that public service organizations have long been operating along network based lines because of such factors as the historic lack of developed market mechanisms; the existence of a range of strong professional groupings and a relatively weak 'command and control' managerial spine; and the presence of parallel organizational mandates, jurisdictions, and agencies that have required interorganizational alliances. Corporate skills in managing these interorganizational ties effectively represent a core form of social capital management as external actors may be able to block key change objectives. larillo's (1993: 127) analysis noted somewhat sharply: 'all the first writings on networks as forms of coordinating stable relationships among different organizations were produced by sociologists working outside the field of business. The networks they described had to do with universities, health care centers and other non-profit making organizations. Business writers did not find it a useful concept for many years.'

Some key questions emerge: to what extent is the management of public service organizations network based? Are the issues and problems described in the private sector networks literature also applicable to public sector organizations? Is there evidence that the networking activity evident displays a corporate cohesiveness and is related to corporate performance objectives?

Qualitative Empirical Analysis This chapter also presents qualitative empirical evidence from one public service setting (health care). Sometimes the concept of the N-form organization remains at the level of a fashionable slogan, and detailed studies of how N-form organizations function are still rare. The concept remains faddish, with the risk of overgeneralization from restricted data. The focus is often on organization rather than management, so that the managerial problems associated with N-form status may be unexplored (Miles and Snow 1992).

Alongside a turn to theory (Salancik 1995), more finely grained empirical analyses of particular organizational subpopulations are also needed. Two approaches may be used in such empirical analyses. Quantitative sociometric techniques have been used to model the formal properties of networks (Leenders


1995b). Other scholars have used a mix of quantitative and qualitative techniques such as Provan and Milward (1995)'s close empirical analysis of network effectiveness in community mental health systems. They identified features of networks in that population associated with differential effectiveness, notably a higher degree of network centralization through the presence of a core agency, high direct fiscal control by the state, and high general system stability. Their results questioned the effectiveness of highly decentralized systems without sources of overall coordination.

We here focus on the subpopulation of health care purchasing organizations (HMOs in American parlance) and are aware that conclusions should not be extrapolated to other public service settings.

PUBLIC SERVICE SETTINGS: THE CASE OF HEALTH CARE

A rather separate literature on policy networks has developed within the public administration and political science literatures. This literature sees government as having a unique status in terms of its power resources (e.g., monopoly on the use of legitimate force) and societal mandate so that public sector networks may well be different in character (Kickert et al. 1997: 177).

Marsh and Rhodes (1992) explore the concept of policy networks (see also Laumann and Knoke 1987; Bueno de Mesquita and Stokman 1994) within the operation of government, drawing on pluralist and corporatist political theory. They propose a typology of network types, ranging from more corporatist and highly integrated policy communities to more pluralist and loosely integrated networks. The strength of this political science literature is that it relates variation in network structure to an analysis of the underlying distribution of power across the network. It is less concerned than the managerial literature with identifying tactics that facilitate effective action within the network.

An intermediate category of professionalised networks (Ham 1981) was seen as applicable to health care settings, where narrow professional networks were highly dominant. The implication was that they may serve professional rather than corporate objectives. Wistow (1992) argued that the integration of the health care network had historically been provided through dominant clinical professionals insulated from other networks such as the managerial.

This chapter argues that there have been significant recent changes to the network structure of U.K. health care purchasing organizations. Managerial networks must increasingly be taken into account as well as clinical networks. There is also preliminary evidence to suggest a modest opening up of health care networks to influence from nongovernmental actors and civil society. Similar broad trends toward cost containment, managerialization, and greater consumerism have been reported in other American and European health care systems, and it would be interesting to ascertain whether similar organizational shifts are evident there.

Recent Developments in Health Care: A Mixed Mode of Management The present evidence from U.K. health care suggests that a pure paradigmatic shift to a network-based ideal type of organizing is less likely than a period where mixed modes of management coexist. Diversity is more likely than consistency, and


ideological pluralism more likely than hegemony. This is not an unusual situation in management practice, but there are two schools of thought as to whether such mixed modes represent a sustainable compromise or a set of internal contradictions.

O'Neill and Quinn (1983) argue that different models should not necessarily be seen as contradictory and that apparent opposites could be reconciled. Rather they broaden the repertoire of approaches available to managers as they move from one situation to another, given that responses may be context specific. However, Miles and Snow (1992) argue that a patchwork move to a networks mode of organizing will fail to provide an organization with the coherent logic and set of skills needed to make this distinctive approach effective.

The managerial agendas and practices of local health care organizations in a publicly funded and accountable system such as the U.K. National Health Service (NHS) are to a large extent mandated by central government, although enacted locally. Recent legislatively driven changes (the NHS and Community Care Act 1990) introducing a quasi-market in health care were firmly driven from the top (Ferlie et al. 1996). In the early 1990s, many NHS managers adopted market-led, price-based and entrepreneurial modes of management. 'Macho purchasing' emerged in some localities (Flynn, Williams, and Pickard 1996), so that traditional local and informal networks dried up and transaction costs escalated. As the internal market matured in the mid-1990s, crude classical models of contracting increasingly gave way to more relational models of contracting.

This emergence of relational contracting began to highlight once again the importance of informal interorganizational networks in the NHS. In the mid 1990s, the center launched new policy initiatives consistent with the adoption of more network styles of management. This marked a retreat from more aggressive market based models of management and a turn back to a public health orientation. Recent policy (Cm 3807 1997) signals the formal abandonment of the quasi-market model and a stress on new core values of cooperation and the integration of care.

Network based initiatives have included the adoption of targets for improving the public health that require widespread interorganizational working, the development of enhanced primary care capacity in addition to hospital care, a growing awareness of the need of health and social care agencies to work together to provide long-term community care, and an increased stress on user involvement and a more outward-facing orientation.

There appears to be no long-term or unambiguous shift to a pure new mode of network based management. Rather, network, market, and line management principles coexist and compete for influence. Each of the three models waxes and wanes over time, with the networks approach finding most favor recently. However, the life cycle of each favoured management approach seems to last no more than about five years and is heavily dependent on central governmental sponsorship. Skeptics argue it is still much too early to conclude that the adoption of a more network-based model is an enduring trend, as it may yet tum out to be no more than yet another passing managerial fad (Abrahamson 1991).


NETWORK BASED MANAGEMENT: SOME RECENT EVIDENCE FROM HEALTH CARE

Methodology Informed by these important developments, the author and a colleague were commissioned by the National Health Service Executive (NHSE) to undertake an empirical study of the management by networks phenomenon (1994-1995) (Ferlie and Pettigrew 1996). An initial review of the literature was used to construct a semistructured interview pro forma to test arguments. Visits were undertaken to nine Health Authorities (purchasing organizations) across the country which were not selected randomly but were nominated by experts in the Department of Health as having the reputation for being at the 'leading edge' of development. The sample included a mix of inner-city, urban and rural localities.

A typical program consisted of four to five interviews at headquarters in the morning, combined with visits to one or two innovative localities or projects (e.g., local multiagency projects) in the afternoon, where further interviews took place. A spread of respondents from different functions and agencies was accessed, resulting in about seventy interviews in all. Both managers and clinicians were interviewed in each authority. Full notes were taken at the interviews. Categories were then developed from an analysis of the mass of interview notes, and documentation collected. All data are thus qualitative in nature.

As access was accorded to 'interesting people in innovative localities' as judged by expert reviewers, the data may not be typical of practice but rather reflect its leading edge. Note also that data were gathered solely within purchasing organizations, and it would be interesting to explore any differences with providers.

Empirical Evidence from Health Care Settings

The Networking Phenomenon Exists The first conclusion was that network based styles of management are of substantial and rising importance within the health care organizations studied. Specifically, we found that the health care managers and clinicians interviewed were spending much time in negotiating, persuading, and influencing partners in organizations outside health care to increase their contribution to joint goals. Successful implementation of a broad agenda for public health requires the support of many stakeholders far beyond the formal health care system to work together. Another set of network-rich settings has been uncovered empirically and in a different sector from that conventionally studied in the management literature (Nohria and Eccles 1992). Moreover, purchasing organizations were described as varying in their ability to construct interorganizational alliances, and this skill was an important requisite for social capital and organizational performance.

Network based management represented however a shift of emphasis rather than a total change, and respondents talked of the need to operate in different modes in different circumstances in order to create and maintain social capital. The caution expressed by Miles and Snow (1992) in relation to the difficult coexistence of different logics of managerial action needs to be recalled.


Most respondents agreed that networking with external stakeholders was now a key managerial skill within health care organizations. However, some felt that this had always been the case, citing traditional negotiations with other autonomous public service agencies about boundaries, or the historic need to persuade professionals (such as clinicians) or politicians before managerial action could take place. Influencing had always been a more functional tactic than the exercise of positional power. So constructing the 'right ties' in order to create social capital had always been important and had become part of the managerial repetoire.

Nevertheless, there was more emphasis on networking than in the recent past, as direct line management had fragmented. Reliance on the quasi-market and signals from prices was still seen as marginal, and even as declining in importance as major strategic change required an alliance between all key stakeholders to succeed.

When asked to characterize the interpersonal attributes and skills needed within network based forms of management, 'trust,' 'reciprocity,' 'understanding,' and 'credibility' all emerged as important basic concepts in use. It is clear that there is an important interpersonal component to network based forms of management, which can make it vulnerable to the turnover of key people. What is interesting is whether it is possible to institutionalize these characteristics at a collective level, so that they become a form of corporate social capital, as some organizations were trying to do.

The alignment of incentives at an agency level is also important so that a 'win-win' situation can be established. Achieving such interorganizational trust may be more difficult than building inter personal trust, yet vital if alliances are to survive the departure of key individuals.

Changing Internal Management Structures and Styles The networking phenomenon has had internal as well as external consequences. Specifically, we found that a number of respondents reported moves internally away from reliance on established functional hierarchies and toward looser and network based or (cautiously) matrix management styles of working. Active organizational development (OD) function was at work in some of the sites, trying to reshape organizational style in this direction. A move from organizing along uni functional or professional lines to more team-based and mUltidisciplinary approaches was widely reported. This had important implications for clinicians who were working in a more corporate and managerially orientated mode than previously. They were increasingly immersed in the corporate core of the organization as well as undertaking their traditional negotiating role with clinical subgroups.

As a result, respondents reported that they were being operating with a more mixed portfolio of tasks (,matrix hoppers') and ad hoc teams. Some managers reported problems with the lack of a clear structure, particularly at middle management level where the pace and complexity of work has been 'upped.' Such matrix management approaches depend on strong channels of informal communication to make them work and a movement away from the 'meetings culture' that had previously been dominant in these organizations. The introduction of e-mail is seen as having a helpful decentralizing effect. Critics argued that sometimes these organizations were changing their internal management styles less in practice than they proclaimed in theory.


A growing number of organizational mergers were providing an opportunity to move away from the old organizations, based on strong chairs and CEOs and displaying a 'command and control' culture. It was commented that the new organizations felt more 'diffuse' and 'looser' than the old. Nevertheless, these mergers were generally imposed from above rather than growing out of locally based ties and alliances.

Evolving CEO Role and Style What is the role of the CEO in a more network based form of organization? Clearly, the role is different from the role played in a 'command and control' organization as the apex of all reporting lines and sole crossover point for a number of vertically organised functions. However, the distribution of power and lines of upward accountability are not entirely diffuse, and so CEOs were grappling with how to retain a shaping, influencing, and indeed accountable role while also 'letting go.'

One role is to build and maintain networks through making external contacts and then retain an overview across the multiplicity of networks that emerge. It is the CEO who above all attempts to transform social networks into corporate social capital. The CEO might be expected to take a view about the value added from each network, deciding to boost involvement in one network, while withdrawing from another. The consequence may be less CEO time spent in the office and a lessened ability to react immediately to the short-term crises that constantly flare up in health care. The CEOs interviewed felt themselves to be 'outward facing'-one estimated that she spent only 25 percent of her time at HQ.

A second key role of the CEO is to give the collection of networks strategic direction, to reinforce core purposes and tasks, to ensure corporacy and to assess performance against key objectives. The CEO may then embody the performance management orientation within the network. There was a concern expressed by respondents that network based forms of management might degenerate into the social liability of perpetual 'talking shops' and that key purposes could be lost.

A third role of the CEO could be to institutionalize strategic alliances by moving on from a reliance on interpersonal trust to a deeper level of interorganizational trust. CEOs may act to construct win-win situations, to broaden channels of communication and joint working, to extend and deepen the group of personnel crossing boundaries, and to build an internal culture more receptive to alliances. At present, however, many of the alliances studied remained at the interpersonal level and have not been progressed onto the interinstitutional level. Clinical networks and alliances still remain distant in many cases from the corporate agenda.

In essence, the CEO plays a key role in corporate social capital creation and aligning sources of human capital with the corporate agenda. Can clinicians be drawn into the corporate agenda? Can cooperative relations be established with other care agencies, particularly their senior management? Can research and education alliances be created with senior personnel at local universities? Such alliances are so complex and disparate that they may require committed group leadership from within the health care organization to be progressed effectively. Creating and maintaining this networking team is then a strategic role for the CEO.


Clinical Networks: Less Dominant But Still Important Contemporary health care networks are more polycentric than the traditional political science literature suggests. A variety of networks were found to be in operation, some of them clinically based and difficult for management to penetrate. Nevertheless, the networks contained more of a managerial component than assumed in the earlier literature (Wistow 1992), with a (lay) CEO sometimes assuming a metamanagement role.

Within the surviving clinical networks, the public health function displayed an important coordinating role, although its strength at the core of the organization varied. It often emerged as the natural focus for maintaining relationships with clinicians in the provider units. Directors of Public Health (DPHs) also reported the need to operate as 'persuaders,' communicating effectively with a variety of different groups. They need to manage and secure change in practice, not only within the public health function but more broadly. DPHs were seen in this respect as needing generic skills, flexibility and an ability to switch networks and tasks in a 'reedlike' way as well as narrow technical skills. They themselves wore two hats, as clinicians and as executive directors of a corporate board.

These findings indicate a substantial change in network composition in health care when set against earlier work, and in particular the emergence of newly created managerial networks needs to be set alongside traditional clinical networks.

Some Opening Up: Networks from Below One of the criticisms of network based management is that it operates in a closed manner, excluding potential stakeholders, and privileging powerful groups such as clinicians as the historically dominant elite, perhaps now joined by management as a rising corporate elite. Community interests may, however, still remain absent from the decision-making networks (to use Alford's 1975 categories). Closed networks may militate against social change, especially of a discontinuous nature (Gargiulo and Benassi, this volume, term this 'the dark side' of social capital).

Interestingly, there was-against this proposition-evidence of a revival of community development ideas, projects and models and an attempt to open up networks to influence from below. Influential in the 1970s, community development work lost ground in the 1980s but now seems to be climbing back up the agenda. These experiments were often well resourced and backed by the top even if primarily dependent on local product champions. Initiatives were concentrated in localities where there was evidence of deprivation, social isolation, and lack of access to health care. Initiatives were associated with the presence of a strong public health agenda or champion and could generate considerable conflict between very different stakeholders. The proposition that a network based mode of management would result in a bland or consensual management style was not sustained in these case examples, as the social liability of sharp conflict between the governmental and non governmental sectors could arise.

This very early evidence suggests some modest broadening of networks may be occurring, as health care purchasing organizations increasingly face outward to local civil society. This broadening was of a partial nature, and its force should not be overstated.


A Web of Interagency Alliances The diffuse, shifting, and sometimes transient nature of many of the health care networks uncovered was striking. This contrasts with some studies of the private sector that sometimes stress a narrower range of strategic alliances, often driven by resource dependency (see Barley et al.'s 1992 analysis of strategic alliances in commercial biotechnology).

The health care organizations studied were actively engaged-with varying degrees of success-in establishing alliances with many other organizations, including public agencies, private firms, and civil society. These ties took the form of formal interorganizational linkages as well as interpersonal networks (as in the analysis of the biotechnology sector by Smith-Doerr et aI., this volume). These linkages were not so much driven by a search for financial flows as a desire to influence policy implementation. We are not so much dealing with a single strategic alliance directed from the top but rather a dense web of interagency alliances, with different leaders in different relationships.

Managerial Problems and Challenges of Network Based Organizations

Networks and Performance A major criticism encountered in the field was that network based forms of management were highly time consuming and did not lead to tangible output. What, in other words, are these networks for? There was a perceived danger that the means of managerial process could drive out the ends of delivering better services. Some alliances (for example, joint planning with social care agencies) are mandated by legislation, and therefore there is a debate as to whether participation may be ritualistic or whether such imposed structures may nevertheless reflect objective requirements of work processes such as inter agency referral flows (Leenders 1995b).

Managers in the study wanted to be clear about the social capital that the network was intended to deliver. The combination of network based forms of management and a strong performance management orientation-with its focus on quantification and short-term target setting-now evident in health care organizations could represent an uneasy and volatile combination. Networks could easily proliferate, soaking up a lot of time while delivering very little.

There was also a concern about lack of clarity in present network forms. While senior management often welcomed moves to network based forms of management, middle management was reported as feeling under threat and finding it difficult to cope without a structured framework. A high level of turnover at middle management level was apparent in some of the health care organizations visited.

Network Sustainability Another question relates to the sustainability of some of these networks that were time consuming to create and maintain but could collapse if they were not seen as delivering by participants. Participation was usually instrumental rather than altruistic and depended on what kind of social capital was on offer to each player.

Networks could also rapidly disintegrate if known players were pulled out,


destroying the social capital that was contained within them: 'a face that you know and recognize is key' said one respondent. So networks may be seen as a volatile mode of organization, with short life-cycle effects, especially where there is no foundation of early successes or 'quick wins' to institutionalize the level of commitment.

Many of the networks studied needed a focus or a broker to animate and sustain them (Thorelli 1990), given the continuing need for network mobilization. This is an influencing rather than a hierarchical role, so that the question of who is filling these broker roles is critical. Are people with the networking skills coming forward, or is the pack of existing personnel simply being reshuffled without thought being given to selection against job and skill specifications?

Networks and Change A heavy reliance on preexisting or closed networks is often associated with continuity and maintenance management rather than innovation or change. Network based management may here tend toward conservatism (see Garguilo and Benassi this volume), while organizations are now judged in performance terms on their ability to deliver an ambitious change agenda. The formally declared purpose-and in at least some cases informal intent--of many of the health care networks studied was to create service and indeed behavioral change, so a reliance on closed and traditional networks would be dysfunctional.

So players may need to engage far more openly than hitherto in network building activity to open up networks and bring in new players. Initially, they may need to sell participation in the network to partners not previously involved. Achieving behavioral change in local communities may be dependent on an ability to 'script in' the nongovernmental sector. As reciprocity is a fundamental norm of network life (Leenders 1995b: 193-197), official organizations may here need to give before they can expect to receive from an initially suspicious nongovernmental sector.

HRM Implications and Issues The HRM implications of the move to network based forms of management have not been fully addressed. Brass and Labianca (this volume) have recently argued that the HRM function may have a strategic role as human and social capital brokers and reconfiguring human resources in the light of rapidly changing demands. So what sort of skills and attributes are needed to make these forms of organization work? When asked, respondents were able empirically to identify key attributes and skills, including strong interpersonal, communication and listening skills, an ability to persuade, a readiness to trade and to engage in reciprocal rather than manipulative behavior, and an ability to construct long-term relationships.

A mutual orientation and reciprocity emerge as key concepts in network based organizations. The exercise of positional power gives way to influencing skills, such as an understanding of how best to manage group decision making (Kanter 1989). It is important to foster relational sentiments (e.g., indebtedness) so that exchanges are not seen as one off transactions (Powell 1990). Such norms of obligation and cooperation imply the continued existence of a community of shared values across


the health care field. Within such clannish organizational fields, contractual opportunism is unlikely.

In addition, respondents pointed to the need for successful networkers to possess the ability to cross a variety of occupational, organizational, social, and political boundaries, an ability to speak different languages, and an ability to act as an interpreter between different groups and be credible with a range of different groups.

A single master culture is not apparent within the health care sector, but rather it continues to display very different subcultures. If anything there is a trend away from a dominant professional network to a (modestly) broader and inclusive set of networks. In Hofstede's terms (1994), participants in such polycentric organizations need to cross conventional boundaries to manage intercultural encounters. It also implies a need for highly developed analytic and diagnostic skills so that networkers are aware when they are crossing subcultural boundaries and are able to switch language and behavior patterns as appropriate.

A third cluster of desired attributes was identified in the following areas: tolerance of high levels of ambiguity and uncertainty, a long-term as well as a short-term view, a good strategic sense, an ability to reflect on experience and conceptualise, a capacity to learn quickly and to adapt in new situations.

Managers in such network based organizations will need to cope with high levels of uncertainty and ambiguity. They will face a greater pressure to innovate and hence have to acquire an accelerated capacity to learn (Powell and Brantley 1992 examine the relationship between network forms and organizational learning in relation to biotechnology).

Is accelerated learning explicitly recognized as an objective in network based organizations? Learning patterns may remain at an individual level, but where there is a corporate effort to develop individuals so that they are able to learn and adapt more effectively, then it is legitimate to talk of a corporate social capital creation initiative. Extensive corporate development programmes were evident in some of the sites visited in an attempt to remould the working styles of key personnel. However, effective learning may also depend on the capacity of individuals to reflect on their experience, to conceptualize experiential knowledge, and to adopt new skills and behaviors: an ability to impart knowledge to others, to act as a teacher or a mentor, an ability to transfer knowledge from one setting to another, and an ability to convey requisite standards and attitudes to others inside and outside the organization (norm setting).

CONCLUDING DISCUSSION

Whose Networks? Whose Social Capital? This study confirmed the importance of networks in health care management, so that the possession of rich ties with actors in parallel organizations was seen as a key invisible asset in 'getting the business done.' Markets and hierarchies both remained weak modes of organizing, and so network based assets were of key importance. Our data did not confirm the conventional picture of total domination by professional networks (Wistow 1992). Clinical networks retain great importance and centrality but have been somewhat broadened by the creation of other networks.


One view is that the introduction of the internal market in health care was significant in beginning a process of deinstitutionalisation where the old professional networks no longer have quite the same power as hitherto (Morgan and Maddock 1997). Managerial networks emerged as of greater importance than predicted. In addition, some community-based networks were also emerging in a bottom-up fashion.

The question is: who owns this social capital? Key individuals were seen as particularly effective as boundary spanners so that in these cases capital accrued to particular people who could well be head hunted elsewhere, on the grounds that they had a track record in creating effective alliances. Other networks were contained within particular subgroups (e.g., clinical networks) that it was difficult for management to penetrate. Such social capital belonged to a group rather than an individual, but hardly to the organisation.

However, there were also instances cited of attempts (to put it no higher) to create forms of social capital at the level of the organization. Senior management played a strong role in creating and managing a number of the networks uncovered, including relationship building with managerial personnel in other organizations, in the interests of the corporate business agenda. Clinical networks could be managed through the public health function, which stood at the interface between managerial and clinical systems. The fusion of managerial and clinical networks is a difficult area that has been handled through the creation of special clinicaVmanagerial hybrid roles. The Director of Public Health thus had a key role in taking the corporate agenda to clinical networks. Many of the board level personnel had been subjected to extensive management development and team-building interventions in an attempt to shift them into more corporate modes of thinking and doing.

How Significant is this Shift? The depth of the shift to network based modes of management in health care purchasing settings can be questioned. The transition to network based forms of management was found to be only partial, recent in origin, and often mandated by the top. It is not yet clear what its long-term significance may be, whether there is substantial ownership at local level, whether it is no more than a passing trend, and whether the distinctive skill requirements of network based management are being seriously addressed. It is premature to talk of a wholescale transition to a N-form of organization in health care as managerial practice in health care is faddish and characterized by short life cycles.

Currently the health care system is in mixed management mode. The problems of such mixed modes of management have been commented on by Miles and Snow (1992) as they are seen as creating an idiosyncratic system, highly dependent on a few key individuals or units to perform. Rhodes (1997) also argues that mixing different governance structures may be like trying to mix oil and water. The competing values framework, by contrast (O'Neill and Quinn 1983), takes a more optimistic and pluralist perspective. Further longitudinal studies are required to establish whether the present mixed pattern continues and what the associated management challenges may be.


The breadth of the shift across public service settings should also be examined further. Developments in one set of organizations (health care purchasers) may be specific, and others elsewhere may not be experiencing the same shifts. Further detailed empirical work is required across different public service organizations.

Network Based Management: Not an Alien Practice Can managerial theories and practices developed in one sector be usefully transferred across to another sector, or are radically new concepts required? Public administrative theory argues that the nature of the tasks undertaken in the two sectors is fundamentally different (Pollitt 1990; Hood 1991; Ransom and Stewart 1994). Some have seen the rise of the New Public Management as a attempt to move the public services 'downgroup' (Dunleavy and Hood 1994): that is, making the public sector less distinctive as a family grouping of organizations. Other organizations in essence converge on the model of the private corporation, reducing intersectoral variety.

Our position (Pettigrew et aI. 1992; Ferlie et aI. 1996) is that there are both differences and similarities between the two sectors that need to be disaggregated. Up to the mid-1980s, the dominant problem was one of isolation: there was a too ready presumption of difference. Since the mid 1980s, the problem has been the overmechanistic transfer of practice from private to public sector. There is here too ready a presumption of similarity. Useful theories are those that can take account of intersectoral differences as well as similarities. Indeed, those that do not are harmful, resulting in imposed change that fails to root and that has to be unpicked.

So are we to see network based management as no more than the latest private sector managerial fad, inappropriately exported into the public sector? It is rather a very special case of a managerial practice originally evident in public service settings (Jarillo 1993), later diffused into the private sector, only to be reexported back. It can therefore 'fit' public service work to a greater extent than other contemporary managerial 'products' (e.g., BPR). Some of the forces favoring the emergence of the network form identified by Castells (1996) in the private sectorincluding but going beyond the rapid proliferation of informational technologiesare also sector neutral and apply equally to public service organizations.

Our data confirmed that many health care managers saw a networks based approach as always strongly developed within their sector. Flynn et aI.'s (1996) analysis of community health services analyzed them as naturally organised through multidisciplinary networks, a form of coordination eroded by the importing of alien concepts of contracting and the market. Morgan and Maddock (1997) similarly point to the erosion of historic professional networks and the introduction of 'hard' contracting styles, which may however fail to institutionalize themselves.

So a network based form of management could be appropriate within public service as well as private sector settings: in a sense, it was invented there. This would predict rapid diffusion of the networks approach within public services as it is seen as 'fitting.' This is an empirical proposition that requires testing. We also need to know more about who owns these invisible forms of social capital and in particular, the degree to which they adhere to corporate tearns rather than individuals.

The Dark Side of Social Capital

•

ABSTRACT

16 Martin Gargiulo

Mario Benassi

Research on social capital has stressed the advantages that networks can bring to managers and other economic actors. The enthusiasm with this 'bright side' of social capital, however, neglects the fact that social bonds may at times have detrimental effects for a manager and produce social liability, rather than social capital. This chapter tries to correct the optimistic bias by looking at the 'dark side' of social capital. Continuing benefits from social capital require that managers can adapt the composition of their social networks to the shifting demands of their task environment. This often implies the ability to create new ties while lessening the salience of some of the old bonds-if not severing them altogether. Available evidence, however, suggests that this ability may be encumbered by the same relationships purportedly responsible for the prior success of the manager. When and how this may happen is the central question we address in this chapter. We argue that strong ties to cohesive contacts limit a manager's ability keep control on the composition of his network and jeopardize his adaptability to changing task environments, which damage the corporate social capital of the organization. We test our ideas with data on managers working for a special unit of a high-technology firm operating in Europe.

INTRODUCTION

Social capital has become a ubiquitous notion in the study of organizations. Despite differences in emphasis adapted to specific research agendas, most treatments coincide with the general definition of social capital as ' ... the sum of resources, actual or virtual, that accrue to an individual or a group by virtue of possessing a

The Dark Side of Social Capital - 299

durable network of more or less institutionalized relationships of mutual acquaintance and recognition' (Bourdieu and Wacquant 1992: 119; see also Bourdieu 1980, for an early formulation). Behind this compelling metaphor, a rich stream of. research highlights how social networks can enhance the ability of individuals and organizations to attain their goals. Scholars have identified two main mechanisms that account for the positive impact of social networks on individual and organizational action. First, networks facilitate access to information, resources, and opportunities (Grano vetter 1995; Campbell, Marsden, and Hulbert 1986; Flap and De Graaf 1989; Coleman 1990; Burt 1992, 1997; Podolny and Baron 1997).

Second, social networks help actors to coordinate critical task interdependencies and to overcome the dilemmas of collective action (Blau 1955; Pfeffer and Salancik 1978; Kotter 1982; Gargiulo 1993; Gulati 1995a). The resources and information embedded in networks of interpersonal and interorganizational relationships can help managers and organizations to attain their goals, and hence they can be discussed as 'corporate' social capital (Gabbay and Leenders, this volume).

Faithful to an agenda that vindicates the importance of social ties in modern economic transactions (Granovetter 1985), sociologists have largely stressed the advantages that embeddedness in social networks can bring to actors engaged in such transactions (e.g., Coleman 1990). This enthusiasm with the 'bright side' of social ties has led sociologists to disregard the possibility that social bonds may at times hinder, rather than help, an actor's ability to pursuit his interests (Portes and Sensenbrenner 1993). The instrumental value of social ties rests on the good match between the resources needed by an actor and the resources he could access through his contacts. To maintain this match, and thus maintain his stock of social capital, the actor should be able to adapt the composition of his social network to fit his changing needs. Available evidence, however, suggests that an actor's ability to maintain his social capital may be encumbered by the particularistic demands posed by the same relationships purportedly responsible for the initial success of this same actor.

This chapter tries to correct the optimistic bias of the existing research by looking at the 'dark side' of social capital. Building on existing theory, we argue that the network that provided a manager with social capital might also limit his ability to change the composition of this network as required by his task environment. In these circumstances, the network that provided the manager with social capital would no longer be an asset but rather a liability that impedes successful action. When and how this may happen is the central question addressed in this chapter. Our answer focuses on the obstacles that strong, cohesive social bonds may pose on the adaptation of social capital to changing task environments. Evidence comes from managers operating in a special unit of a European subsidiary of a high-technology firm. The analysis emphasizes how the structural characteristics of the managers' social networks thwarted the adaptation of their networks to a fluid organizational context.

THE TWO SIDES OF SOCIAL CAPITAL

Managers bring to their job more than the skills they have accumulated through years of education and experience. They also bring the assets they can procure


through their social networks. Economists refer to the assets embodied in the manager's skills as his 'human capital' (Becker 1964). Sociologists coined the term 'social capital' to designate the assets tied to the manager's social network (Bourdieu 1980; Coleman 1990).

The compelling metaphor embodied in the notion of social capital has brought together a number of research streams that focus on the beneficial effects of social networks in economic exchanges. Different researchers have emphasized different aspects of social capital, albeit not always making explicit reference to the concept as such. In his comprehensive review of the literature, Gabbay (1995, 1997) distinguishes between approaches that focus on the resources controlled by the alters from those that stress the relationship between ego and the alters. Resource-based approaches are dominant in studies of individual success in labor markets (Lin, Ensel, and Vaughn 1981; Lin and Durnil 1986; Flap and De Graaf 1989), but these studies have been also approached from a relational perspective (Granovetter 1973). The focus on the structure of the relationships dominates analysis of managerial careers (Burt 1992, 1997; Podolny and Baron 1997). The relational approach is also dominant in studies of inter-organizational cooperation. Networks can help organizations seeking to form partnerships by providing efficient access to information on the availability, competencies, and reliability of potential partners, thus lowering searching costs and alleviating the risk of opportunism (Granovetter 1985; Powell and Brantley 1992; Gulati 1995a). Closely related to this last point, scholars have also demonstrated how social networks have a crucial role in helping managers and organizations to reduce the uncertainty that results from task interdependence (Blau 1955; Pfeffer and Salancik 1978; Kotter 1982; Gargiulo 1993).

The interest in the role of social capital as relationships in overcoming coordination problems has been recently fueled by the growing importance of flatter structures, teamwork, entrepreneurial initiative, and decreasing reliance on authority relations to handle interdependence (Kanter 1983a; Ditcher 1991; Baker I 992a; Sheppard and Tuchinsky 1996). This new environment poses an important challenge for managers. While the new organizational blueprint gives managers considerable more latitude to perform their role, it also increases the uncertainty surrounding their task and careers. Managers have to handle a growing number of task interdependencies that link them to people outside their line of authority. Since the formal structure offers little assistance in dealing efficiently with these 'lateral' interdependencies, managers have to find alternative ways of coordination to accomplish their task effectively. Social capital becomes a strategic tool to add value in this new organizational environment (Nohria and Eccles 1992).

The distinction between 'resources' and 'relationships' helps clarify a key aspect of social capital, often ignored in the literature (see Araujo and Easton, this volume). In fact, social capital encompasses both the resources controlled by the actors and the relationships linking ego to those actors. As Gabbay (1997) adequately points out, resource-based approaches implicitly assume the alter's motivation to assist ego, while relational-based approaches assume that ego would choose the 'right' alter to include in his network (e.g., Burt 1992: 13), restricting the causal force to the strength and the structure of the relationships. Yet, effective social capital requires

The Dark Side of Social Capital- 301

both alters with the resources needed by ego and a social structure that facilitates ego's access to those alters. Although analytically distinguishable, both resources and relationships are essential to the notion of social capital. Indeed, the main theme of this chapter is that the interplay between the inertial tendencies of some social structures and the dynamic character of the resources needed by a manager may turn social capital into a liability that hinders, rather than help, managerial action.

By demonstrating the positive role of social relationships in modern economic life, sociologists have contributed to a time-honored research program revitalized by Granovetter (1985) with his seminal essay on the social embeddedness of economic transactions. While economists have typically ignored social relations or have treated them as an obstacle to attain economic rationality, sociologists have shown that economic rationality can be actually enhanced by embedding transactions in social networks that facilitate trust and diminish the risk of opportunism. In pursuing this research program, however, sociologists have often disregarded a well-known duality also stressed by Granovetter (1985) in his essay: social structures can be both a source of opportunities and a source of constraint for individual behavior. As one might expect from the positive tones of the metaphor, the bias towards the favorable effects of social structures is especially apparent in the discussions of social capital (e.g., Coleman 1990: 300). This enthusiasm with the 'bright side' of social capital has led sociologists to disregard the 'dark side' of social relationships economists like to emphasize-that is, that social bonds could be an obstacle to pursuit economic interests.

Despite this pervading optimism, the evidence on the constraining effects of social bonds on individual action is not missing. While most of this evidence may come from studies that portray the supporting and yet oppressive effects of dense social networks in small communities (e.g., Fischer 1982), the dark side of social capital has been also stressed by sociologists analyzing economic behavior. Thus, Portes and Sensenbrenner (1993) review various studies describing ethnic entrepreneurs suffocated by the particularistic demands posed by the same strong social ties purportedly responsible for facilitating their initial access to essential resources. Summarizing their analysis of these findings, Portes and Sensenbrenner (1993: 1341) propose a direct relationship between the amount of social capital initially available to successful entrepreneurs on one hand, and the level of particularistic demands and restrictions to freedom of action placed on those entrepreneurs, on the other.

The theme behind these warnings is one familiar to economists that emphasize the negative effects of social bonds: the maintenance of social capital entails honoring obligations that may conflict with the pursuit of self-interest. Confronted with this problem, some sociologists have come to accept that actors may at times be better off with the impersonal, 'arm's-length' transactions glorified by economists than with the embedded exchanges advocated by their discipline (e.g., Uzzi 1997a, this volume). Yet, the critique to the one-sided view of social capital goes beyond the dichotomy between 'embedded' and 'arm's-length' transactions that typically confronts sociologists and economists. The restrictions that strong social bonds may pose on an actor may hinder not only his attempts to substitute arm's-length transactions by previously embedded ones, but also attempts to initiate new


embedded transactions. In other words, the network that provides the actor with social capital may also encumber his ability to change the composition of this network as required by changes in his task environment. The tension here is not between embedded and arm's-length transactions, but rather between the currently available social capital and the social capital required to cope with the new demands or opportunities perceived by the actor.

Perhaps one of the reasons for which scholars have largely ignored the dark side of social capital is related to the specific conditions in which its negative effects may become apparent. The dark side of social capital is not always consequential; rather, its effects will be noticed only after the actor reaches a point beyond which the resources available to him through his current contacts are no longer adequate. The entrepreneurs discussed by Portes and Sensenbrenner (1993) did benefit from the social capital initially available to them, but the obligations that resulted from those benefits curtailed their subsequent ability to pursue other business opportunities. It was the change in the entrepreneurs' opportunities-which may have resulted from their initial success-that made the constraining effects of social bonds consequential. Since the instrumental value of social capital lies on the match between the resources needed by an actor and the resources provided by the actor's contacts, changes in the actor's task environment may require changes in the composition of his social network. At least in this respect, social capital is similar to human capital. Changes in task environments may render obsolete current skills and capabilities, forcing people to make further investments in education.

Investments in social capital, however, are substantially more complex than investments in human capital (Coleman 1990). While a person can typically acquire new skills without having to discard previous ones, the same is not always true for social capital. The maintenance of social capital requires investing time and energy in one's contacts. Since people-and managers are not an exception-have a limited amount of time and energy, pressures to maintain relationships that are no longer advantageous may hinder the ability to cultivate other relationships necessary to renew the managers' social capital. The ability to withdraw from business relationships that are no longer advantageous has been often recognized as an important factor in the adaptability of managers and organizations to changes in their environments (Miles and Snow 1992). Paradoxically, the better the fit between an actor's social network and his current environment, the harder will be to adapt this network to a new environment (Uzzi 1997a).\ Changing the composition of social capital often implies creating new ties while lessening the salience of old bonds-if not severing them altogether. The more intense and productive the ties with the old contacts were, the more difficult will it be to part with those relationships. Such is the paradox of social capital: the brighter its bright side, the darker the potential effects of its dark side.

The mechanism behind this paradox is rooted in the very logic of reciprocity that turns relationships into the assets that form social capital (Coleman 1990). Reciprocity is more than repayment or the last favor received: it is a general norm that prescribes a certain type of behavior towards relevant others: takers ought to be givers. The norm of reciprocity may oblige a manager to assist a contact, even if he expects few benefits from future exchanges with this particular contact. This is


especially the case if this contact has been rather helpful to the manager, or to somebody he is indebted to. Failure to reciprocate may result in strong sanctions and in a serious damage to his reputation as a trustful contact, a damage that can be consequential for the manager's ability to create new ties. It is precisely this failure to reciprocate what we often disapprove of the 'instrumental' individual, the person who cuts the ties of obligation to the group once he sees no further benefit from exchanges with its members. In a culture of 'generalized exchange' (Ekeh 1974; Bearman 1997), reciprocity is not restricted to localized exchanges within cohesive clusters: favors can be repaid to people other than the ones in the group (see Lazega, this volume). Although most organizations probably aspire to have a culture of generalized exchange, the majority of them operate within a system of restricted exchange: favors ought to be repaid to the givers, or to the giver's friends. In these organizational contexts, the norm of reciprocity, combined with the capacity to impose sanctions to defectors, may force a manager to attend demands even when he expects no further benefit from the contacts posing those demands.2

The effects of reciprocity are likely to be compounded by a second mechanism that may keep managers tied to contacts that have lost their value as social capital, even without the managers being aware of the problem. This second mechanism is relational inertia. People get used to dealing with their long-term partners. This familiarity breeds strong bonds of mutual understanding and trust that greatly facilitates cooperation (Gulati 1995b). Unfortunately, the same strong bonds may also serve as a filter for the information and the perspectives reaching the actors, generating a 'cognitive lock-in' that isolates them from the outer world (Grabher 1993). In addition, the easiness of cooperation with familiar partners raises the cost of making the investments that are necessary to initiate and to consolidate new relationships. This cost is even higher once the uncertainty associated with the formation and with the probability of maintenance of new ties is taken into account. These inertial factors can make established relationships extremely resilient to losses in their instrumental value due to changing task environments. If the change in task does not come with tangible signs such as a physical relocation or new a new reporting relationship, the manager may fail to recognize the need for adapting the composition of his social capital to fit the new task. Since failure to develop the adequate relationships is likely to have an impact on performance, the manager may eventually learn the hard way about the inadequacy of his network. Late adaptation, however, comes at a cost. In some cases, the cost may be failure in the new task, with potentially serious consequences for the manager and for the organization.

The tension between the forces of reciprocity and relational inertia on one hand, and the changing nature of the resources needed by a manager on the other is intrinsic to social capital. One simply cannot have the benefits of social capital without the corresponding obligations and the risk of relational inertia. This tension is particularly apparent for managers facing transitions that affect their task environment without simultaneously introducing clear-cut changes in the conditions under which they have to perform their jobs. This situation is rather frequent in the dynamic contexts that characterize today's flat, entrepreneurial organizations. Social capital is key to coordinate task interdependencies in these organizations. At the same time, the fluid character of the managerial task makes task interdependencies


shift, following the changes in the resources and expertise managers have to coordinate to add value to the organization. Indeed, the fluidity of the interdependencies is one of the main arguments against attempts to coordinate them through formal mechanisms and standard procedures. In these contexts, the ability to adapt the composition of his social capital is perhaps one of the most valuable attributes of a good manager.

For a manager, the problem is how to have the benefits of the 'bright side' of social capital while minimizing the potentially deleterious effects of its 'dark side.' How to enjoy the advantages of resourceful social networks while maintaining some degree of control over the composition of this network? Our approach to this question is to look at factors that may boost the impact of the forces of reciprocity and relational inertia, hence hindering the manager's ability to adapt his network. We argue that the structure of the network that defines social capital is one such factor. More specifically, we argue that a manager's ability to adapt the composition of the network that carries his social capital is a function of the structure of that network. The issue is to specify how network structure may curtail the manager's ability to decide how to allocate his time and energy across a set of actual or potential contacts. For this purpose, we turn to the control implications of structural hole theory.

THE EFFECTS OF NETWORK STRUCTURE

Structural hole theory (Burt 1992) asserts that the ability of an actor to have control over his environment is a function of the structure of this actor's network. The more a manager depends on parties who are difficult to substitute and who can coordinate their behavior, the less his ability to negotiate his role in the network. Conversely, lack of contact among those parties creates 'structural holes' that enhance the manager's ability to control his environment.3 Players who occupy brokerage positions between disconnected alters have comparative advantages in negotiating relationships, which allows them to secure more favorable terms in the opportunities they chose to pursue. According to Burt (1997: 343), 'managers with contact networks rich in structural holes know about, have a hand in, and exercise control over more rewarding opportunities. They monitor information more effectively than it can be monitored bureaucratically. They move information faster, and to more people, than memos.'

The focus of the research inspired by structural hole theory has been on the outcomes of social structure. The information and control advantages that accrue to managers occupying brokerage positions in networks are deemed responsible for the managers' early promotions or comparatively higher bonuses, which are seen as a recognition to their ability to add value to their organizations (Burt 1992, 1997). Structural hole theory, however, may be also used to explain the process through which social capital is created and renewed by the managers. According to structural hole theory, a manager strongly tied to contacts that are in turn connected to one another is expected to have little autonomy to negotiate his role vis-a-vis his contacts. Likewise, we can hypothesize that such a manager should have difficulty in renovating the composition of his network when the pressures of the environment may require him to do so. More specifically, we expect this manager to be less able


to develop new relationships or to have control over the strength of the ties with existing ones.

The reasons behind our hypothesis lie on the interplay between social structures on one side, and the forces of reciprocity and relational inertia on the other. As Granovetter (1985) has argued and empirical studies have confirmed (e.g., Gulati 1995a) the probability of cooperation between two actors is enhanced by the presence of third parties that bring 'closure' to the social structure (Coleman 1990). Burt and Knez (1995) dissect the mechanism behind this effect in their analysis of the impact of third parties on trust, which they define as 'anticipated cooperation' (cf. Coleman 1990). Indirect connections create a reputational lock-in that ensures cooperation. Non-cooperative behavior by either partner may be reported to-or worse, observed by~ommon partners. This would typically have serious negative effects on future relationships entered by the defecting partner, hence serving as an effective deterrent to defection (Raub and Weessie 1990; Burt and Knez 1995; Nooteboom, this volume). Consequently, managers linked through ties embedded in third-party relationships are more likely to conform to pressures of reciprocity. The strong ties to his contacts, and these contacts' ability to coordinate their demands on the manager, magnify the impact of the norm of reciprocity on the manager's behavior as well as the effects of relational inertia, hence affecting the manager's ability to invest in the development of new ties.

In line with our argument on the two sides of social capital, these potentially harmful effects may remain latent until the manager actually needs to develop those new ties due to changes in his task environment. Moreover, there is evidence that at least in some situations the short-term effects of strong ties to coordinated contacts may be actually beneficial to the manager.4 As Podolny and Baron (1997) have recently argued, dense ties among the key people a manager depends upon may actually facilitate performance, especially for low and middle-level managers. The manager with such a network is likely to face a well-defined and consistent normative framework within which to perform his role, thus avoiding the tensions of having to respond to conflicting demands. Our argument, however, suggests that this consistent framework may turn against the manager when he, prompted by changes in his task environment, needs to change the composition of the network. Although the peril is perhaps negligible when the new task is accompanied by changes in reporting line or physical location, this may not be always the case. If such tangible changes in the conditions under which the manager performs the new job are absent, the dense network that was hitherto a major source of support for the manager may now curtail his ability to develop the social capital needed to succeed in the new task. The old network is no longer a resource for the manager, but a liability that encumbers his performance.

From a structural perspective, the negative effects of lack of structural holes on the manager's autonomy to adapt his social capital can operate in two different ways. First, restricted autonomy may result from the combined pressure posed by all contacts in the manager's network. The stronger the ties to those contacts, and the stronger the connections among the contacts, the more they will be in the position to collectively enforce reciprocity from the manager towards the members of the network (Lazega, this volume). Group cohesiveness is also likely to drive


members-including our focal manager-to a passive acceptance of the ongoing patterns of interaction as the normal way to carry out their business, hence promoting relational inertia. Second, restricted autonomy may also result from the dominant role of one or a few contacts that monopolize a large proportion of the time and energy the manager devotes to his network. The 'hierarchical' role of these contacts in the manager's network puts them in the position to force their demands upon the manager, leaving him little time to invest in new relationships. Alternatively, the manager may perceive these prominent contacts as all the social capital he needs, thus discouraging further investments in developing this social capital (Higgins and Nohria, this volume).

To summarize our argument: we expect managers whose contacts networks lack structural holes to face more difficulties in adapting their networks according to the changing demands of their task environment, thus leading to a higher rate of failure in the renewal of their social capital. Lack of structural holes may result from strong ties to the various members of a cohesive cluster or from very strong ties to few players that are well connected within the rest of the manager's network. Managers lacking structural holes should be more likely to fail in adapting the content of their social capital in line with changing environmental demands. To test this proposition, we examine managers operating in a special unit of a European subsidiary of a hightechnology firm. We focus on their failures to adapt the composition of their social network to the changing demands placed on them by a flat organizational environment. We describe this organizational environment, then explain the data and methods used in our analysis.

THE ORGANIZATION

Our study focused on a unit within the Italian subsidiary of a leading multinational computer company in the early nineties (see Benassi 1993, for detailed description). Like most firms in the industry, the company was dealing with difficult market conditions. Impressive price-cutting and accelerating competitive dynamics driven by dramatic advances in chip technology made profit margins plummet, forcing firms, and especially large firms, to reshape their activities. In this context, the search for more effective organizational configurations was a major endeavor for large computer manufacturers (The Economist 1993). Our firm was not an exception. At the time of our study, several initiatives of organizational change were under way. Headquarters explicitly initiated some of these initiatives, while others were emerging out of the everyday practice of organizational transformation. Our study focused on one of these emergent strategies. A small unit operating in one of the Italian plants was promoting alternative forms of voluntary cooperation among business units within and outside the organization. Although our unit was not originally conceived as a change agent, its style of work resulted in an emergent strategy that could lead towards a radical change in the processes and the culture of the Italian organization. A good part of that change was already observable inside the unit.

The Direzione Processi Industriali (Direction of Industrial Processes, or DPI) was a small unit formally created in January 1991 and staffed by 19 members, all but one of them male. Its origins can be related both to a process of organizational


change and to individual initiatives. Early in 1991, the existing functional structure was substituted by a business unit organization and new centers of competence were created to deal with the emerging issues the company was facing. At the same time, an autonomous research initiative undertaken by a group of people working at the Italian plant focused on new technological strategies and on alternative models of conceptualizing the overall activity of the firm, both at the national and the international levels. The creation of DPI gave an institutional form to the new, disperse expertise that resulted from these previous initiatives. Out of the 19 members of DPI, 10 came directly from a unit that launched some of these initiatives and continued to work on their projects until these were completed or terminated. Another group of 6 people also came from a single unit within the firm. Thus, from the inception, the largest majority of DPI managers had worked before with some of their current colleagues. At least initially, several of them even continued with the same projects that they have started in their previous job, although new colleagues might have joined the teams.

DPI's scope of activity was very broad, but is overarching mission was to promote horizontal linkages within the firm and between the firm and its suppliers and customers. These linkages were essential to implement the new business unit structure. To this end, the unit collaborated with-and promoted collaboration among-actors inside and outside the company, providing solutions to internal business units, top management, and international functional managers, as well as to external clients. Its competencies included devising manufacturing strategies for the Italian plants, developing a market-driven quality approach, promoting marketingmanufacturing cooperation, and creating tools and methods to implement the different initiatives. It also coordinated activities of managers in charge of setting long-term strategies and represented the Italian plant in international company hearings. To an important extent, these tasks resulted from autonomous initiatives that became institutionalized in the new unit and were then formally recognized by the company. In this sense, DPI clearly resembles the entrepreneurial image associated with flat organization. This image also corresponded with the internal structure of the unit. Although formal hierarchy and task differentiation did exist, barriers among task groups and individuals were insignificant. Personal initiative was not only strongly encouraged within DPI: it was a prerequisite to do the job. The managers knew that solutions were driven by multiple contributions and that external and internal relationships were crucial for getting things done. They believed that traditional organizational mechanisms were no longer valid to accomplish the complex task they had at hand. Instead of relying on the formal hierarchical structure, the managers sought to leverage horizontal relationships and to promote cooperation among internal departments. These initiatives often led to new business opportunities for the firm, both inside and outside the formal boundaries of the organization. Although DPI had the support of top management, its location in the formal structure made it a peripheral part of the organization. The head of DPI reports to the plant manager, who is two steps below the top management.

In line with the organizational blueprints adapted to its style of work, DPI favored working through project teams by which the unit pooled resources from


within and outside the organization to provide specific solutions to both internal and external customers. Between January and October 1991, DPI was directly involved in 73 of such projects. At the time of our survey, 70 percent of the projects were still under way, 20 percent were recently completed, and 10 percent abandoned. One third of these projects was a continuation of those carried out by previous units. The remaining ones were either a direct initiative of DPI (43 percent) or were launched upon internal or external customers' demands (57 percent). In more than 60 percent of the projects, DPI acted as leading unit. Only 8 of the 73 project teams were formed exclusively with people from DPI. On average, project teams had 6.6 people representing 3.7 different units. DPI contributed an average of 2.8 team members per project. Typically, one of these people was the team leader, often paired with a leader from another unit. On average, each of the 19 DPI managers was the team leader in 3.8 projects, ranging from 1 to 7. In addition, he would participate in other project teams. The average DPI manager participated in 10.2 different projects, with a range from 2 to 21. Although managers did have initiative regarding their participation in the projects, the composition of the teams was largely driven by reasons of technical expertise and experience, and was ultimately the responsibility of the head of the unit. In this sense, DPI managers were not truly self-selected into the project teams they participated.

The managerial approach and the internal structure of DPI makes the unit a good example of the fast pace, relationship-driven environment where adaptability of ones' social capital is key to individual and corporate success. The managers developed and leveraged relationships both inside and outside the firm, actively seeking to promote cooperation ties that cut across the formal structure and the boundaries of the organization. To attain that goal, DPI managers must also coordinate the task interdependencies created by their joint participation in teams and by their combined attempts to mobilize resources from independent areas of the organization. This setting makes the unit a fruitful laboratory to analyze the factors that may thwart the renewal of the content of social capital.

DATA AND METHODS

We argue that managers lacking structural holes should be more likely to fail to adapt the composition of their social capital in line with changing environmental demands. To test this hypothesis, we examined mismatches between the level of task interdependence and the level of work-related consultation between the nineteen managers working for DPI. Specifically, we focused on failures to maintain consultation ties at levels deemed adequate for the intensity of the respective task interdependence between the managers. Drawing on resource dependence theory (Pfeffer and Salancik 1978), we assume that effective managers should shape their social network to maximize their capacity to handle the task interdependencies affecting their jobs. In an extreme case, this would mean a perfect mapping of the consultation network on the interdependence network. Individual managers, however, are likely to deviate from this extreme case in two ways. First, a manager may keep ties with people he does not depend upon-Qr people with whom he has only weak task interdependence. We may refer to these as cases of 'surplus' consultation. Second. a manager may have a low level of consultation (or no


consultation at all) with people he strongly depends upon. We call these situations of missing consultation 'coordination failures.'

From the standpoint of resource-dependence theory, the impact of each type of deviation on managerial performance should be different. Although ties with people a manager does not depend upon to carry his task may be seen as an inefficient way to allocate effort, these ties do not necessarily have a negative impact on the manager' s performance. Moreover, such ties may be instrumental in two ways that go beyond the immediate task interdependencies affecting the manager. First, managers may have to keep weak ties with people that were key for their task in the past and who may be important again in the future. Indeed, Granovetter's (1973) analysis of labor markets vindicated the instrumental value of those ties. Second, managers may use those ties as bridges to access resources or information controlled by people the manager has been unable to reach directly, or to effect indirect political influence on people they depend upon but with whom they cannot establish a good working relationship (Gargiulo 1993). Thus, consultative ties not coupled with task interdependence may still perform an instrumental role for the manager. The consequences of the second type of deviation are different. Failure to consult with a colleague a manager clearly depends upon should make coordination more difficult and time-consuming, which in turn should affect the manager's performance. Our focus on coordination failure captures the theoretical difference between the two types of deviation.

The data analyzed in this chapter comes from a self-administered questionnaire distributed to all members of DPI in October 1991. This questionnaire was tailored using the extensive information gathered through ethnographic observation. From June 1991 to December 1991 , one of the authors was allowed to engage in daily observation of the DPI operations. He also had easy access to the unit members for interviews, could consult written communication and formal documents, and attended team and unit meetings. All 19 managers working for DPI responded to the survey. Our questionnaire comprised two parts. First, a general part covered information on the manager's involvement in projects, as well as an evaluation of the functioning of the unit. Each member was also presented with a roster of his or her colleagues within DPI and asked to rank his or her level of consultation on problems relevant to his or her work with each of these colleagues, in a scale from 0 (no consultation) to 3 (strong consultation). Second, those managers who were responsible for the coordination of at least one of the 73 projects implemented by DPI were asked to fill out a booklet with information about each project under his supervision. The booklet surveyed detailed information on the project and on the people involved. It also included a matrix in which the manager responsible for the project had to list all the people that participated in the project team, including himself or herself, and to rate the level of cooperation between all pairs of members. Based on this information, we constructed independent measures of task interdependence and consultation among the nineteen managers working for DPI.

Task Interdependence We measure task interdependence among DPI managers as a linear function of their joint involvement in the projects implemented by the unit. Two characteristics of our


measure are worth noting. First, the measure is an objective, rather than a perceptual indicator of task interdependence among our managers. It is also exhaustive, since the managers' task was largely confined to their participation in the projects initiated or facilitated by the unit. Second, the measure captures both direct and indirect task interdependencies between the managers. By working with one another in a project, the managers' tasks become directly interdependent. Managers working on the same projects are also indirectly interdependent, since they are demanding resources from the same units in the firm, that is, from the ones that jointly participated in these projects. For any two managers {i, j}, we can define task interdependence tij as the number of projects in which managers i and j were jointly involved during the period covered by our research. This raw measure is affected by the number of colleagues with whom a manager jointly participated in project teams. To control for this factor and to focus purely on the manager's pattern of interdependence-rather than on its volume-we use a proportional measure of task interdependence. Manager i's task interdependence with manager j can be thus represented in proportional terms, as the ratio between the number of joint projects with manager j, tij' and the sum of all joint projects across all managers q, including j:

i"# j .

Structural Holes We expect that lack of structural holes in the managers' consultation networks will restrict their autonomy to adapt this network to the requirements of their task interdependencies, hence making them more prone to coordination failures. To measure consultation, all 19 DPI managers were presented with a complete list of their unit's colleagues and asked for the extent to which they routinely consult with each colleague regarding matters that concerned their work in the unit. Managers could rate their answers from 0 (no consultation) to 3 (strong consultation). The raw response data generated a 19-by-19 matrix of consultative ties, with zeros along the main diagonal. The consultation network reveals a relatively flat structure, with little variation in the consultation centrality of the managers and a dense web of ties: Any member of the unit could reach any other member in a maximum of three steps.5 For a variety of reasons not related to their task interdependencies, managers may vary in their tendency to be the source or the target of consultation regarding workrelated matters. To control for this idiosyncratic variation, we use a proportional measure of consultation rather than the raw scores; this allows us to capture the pattern of how a manager has to allocate time to the different people he actively consults or to the people who consult him for work-related matters. This allocation is a function of both the attention the manager seeks from others and the demands other colleagues pose on him. Thus, we measure manager i's consultation with manager j, Sij' as the proportion of attention manager i has to allocate to manager j, both as a result of his seeking out j and of being sought out by j (being sij the selfreported raw measure and 0 ~ sij ~ 3):


and Li Sij = 1 for all managers. The inclusion of the demands posed on the manager by his colleagues, even when the manager does not seek these colleagues out, is relevant in this context. Indeed, our reasoning assumes that the manager will have to pay some attention to these contacts, even if he does not seek them out to discuss his own task. The measure is sensitive to this difference. If the manager simply responds to demands on his time (Sji > 0), but does not seek out the particular contact (sij = 0), his proportional consultative tie with the contact (Sij) would be weaker. As we shall see, such weak tie need not be detrimental. The tie, however, would become stronger if the manager also seek this particular contact out for consultation (Sij> 0).6 A manager's autonomy may be restricted by the combined pressure posed by all his contacts in the consultation network, or from the dominant role of one or a few contacts that monopolize a large proportion of the time and energy he devotes to this network. We define two structural variables that respectively capture these two conditions: network constraint and network hierarchy. Following Burt (1992), we measure network constraint as the sum of the constraint posed by each of the contacts in the network. This constraint is in turn a function of the direct joint consultation between i and j and of the extent to which j consults with the other contacts q in i's network (see Burt 1992: 50-71, for detailed discussion of these measures):

where Sij is the proportional measure of consultation ties discussed before and Ljcij, or network constraint, captures the lack of structural holes in manager i's network resulting from having strong consultative ties to members of a cohesive cluster.

Network hierarchy, in turn, captures the fact that restricted managerial autonomy can also result from a consultation network built around strong ties to one or few players who occupy a central position in the manager's network. We measure this informal hierarchy in the managers' consultation networks as a function of the distribution of the constraint across the manager's contacts. Concentration of constraint in dominant contacts is captured by the Coleman-Theil disorder index.7

The closer the index gets to 1.0, the more a single contactj dominates the manager's consultation network, and the higher his network hierarchy score. Contact j's dominance is a function of the strength of the consultation between the manager i and j (Sij), and of the extent to which j is also consulted by the other contacts q with whom i also consults (SjtPqi). Network hierarchy captures the lack of structural holes in manager i's consultation network resulting the dominant role of one or a few contacts in this network.

Coordination Failures The number 0/ coordination/ai/ures committed by each of the 19 DPI managers is the dependent variable in the analysis. We define as coordination failures those


cases in which high (above average) task interdependence Tij between two DPI managers is coupled with low (below average) consultation Sij between these managers.8 As we noted before, DPI managers had a large number of consultation ties, which in all but one case exceeded the number of their task interdependencies. In this setting, the relative strength of the consultative tie is relevant. By focusing on the joint occurrence of high interdependence and low consultation, we capture cases of strong task interdependence that are coordinated by a weak consultative tie-that is, a tie weaker than the expected level of consultation among any two managers in the unit.

Following this criterion, each of the 342 dyads in the network was coded as either failing (1) or not failing to coordinate (0). On the basis of this coding, we computed an individual failure measure for each manager, defined as the sum of coordination failures across all the people with whom he jointly participated in at least one project team. This is the dependent variable in our study. The measure may vary from 0 when the manager had successfully allocated above average consultation ties to all his strong interdependencies to a maximum equal to the number of strong (above average) task interdependencies if all these strong interdependencies are coupled with weak (below average) consultation. For example, one of the managers worked with 13 different colleagues and he was strongly (above average) task-interdependent with 7 of those colleagues. His level of consultation with those 7 colleagues, however, was below average in 4 of the cases, hence our manager committed 4 coordination failures. On average, DPI managers committed 3.7 failures in coordinating 10.6 task interdependencies.

Coordination failures were detrimental for DPI's corporate social capital and had a notieeable negative impact on the level of cooperation attained in the project teams, hence affecting DPI's ability to attain its goals. The main goal of the DPI was to promote cooperation among the other members of the project teams, who often came from different areas of the organization. Yet, coordination failures between DPI managers jointly participating in a team prevented them from attaining satisfactory levels of cooperation within that team. The proportion of coordination failures within DPI dyads in a project team was negatively correlated with the average level of attained cooperation among all team members (r = -.308; p < .01). The negative impact of coordination failures was also statistically significant for both the average cooperation between DPI managers and other members of the team (r = -.494; p < .01), as well as for the cooperation among non-DPI team members (r=-.396; p < .01).9

Control Variables We have controlled for factors that may affect a manager's tendency to incur in coordination failures. Specifically, we looked at the number of weak consultative ties in a manager's network and to his workload. High failure scores may result from a manager's tendency to have a large number of 'weak' consultative ties-that is, ties that are below the average strength of consultation in the unit. A larger number of weak consultative ties can result from having a large network of contacts, which in turn may translate into a larger number of coordination failures. Failure, however, was not statistically associated with the size of the manager's consultation network


(r = .200; P = .206), nor with the number of ties to colleagues the manager did not work with in a project team (r = .197; P = .209). Thus, failures were not the simple effect of having a large consultation network or from a tendency to consult with colleagues not working in their project teams, but rather from how managers allocate consultative ties to their task interdependencies. The effect of weak ties may be more troublesome for managers who score high in network hierarchy. By definition, such managers have a large number of weak ties and a few very strong ones. Unless the distribution of task interdependence is equally skewed, this profile may automatically result in a large number of coordination failures. The mechanism at work, however, would not be related to the dominant role of the strong-tie contacts, as proposed in this chapter. A statistically significant association between consultation network hierarchy and the number of coordination failures could be simply an artifact of the number of weak ties in the consultation network, rather than a test of the impact of network structure on the manager's ability to freely allocate his consultative ties.

To rule out the possibility that the correlation between coordination failures and network structure was simply an artifact of the number of weak ties in the consultation network, we removed the effect of the number of weak ties from the original variables and re-estimated the models using the adjusted scores. We computed adjusted measures of number of coordination failures (our dependent variable), network constraint, and network hierarchy. The adjusted measures are the standardized residuals obtained by regressing coordination failures, constraint, and hierarchy on the number of weak ties in the manager's consultation network. Consultation ties whose proportional strength (Sij) was below the criterion level

used to define coordination failures were coded as 'weak.'10 The adjusted measures are thus free from the spurious influence of the number of weak ties in a manager's consultation network. If our theory on the encumbering effects of network structure on the ability to match consultative ties to task interdependence is correct, both network constraint and network hierarchy should still have an impact on coordination failures once the spurious effect of weak ties is removed. On the contrary, if the alleged impact of the structural variables on coordination failures is simply an artifact of the presence of weak ties, the impact of these structural variables should disappear after the effect of weak ties is removed from the measures.

A manager's failure to maintain adequate consultation ties with people he depends upon can also be affected by the particular workload conditions in which this manager has to perform his task. The larger a manager's workload, the higher the probability of failing to keep an adequate level of coordination for all the task interdependencies defining his role, and thus the higher the number of failures. Two workload dimensions are particularly relevant here: the number of project teams in which the manager participated and the number of different colleagues with whom the manager jointly participated in at least one project team, which is the number of task interdependencies. As one might expect, these two variables are strongly correlated (r = .744; P < .(01), since the number of task interdependencies typically increases with the number of project teams in which the manager participates. This

g. 0' -

Tab

le 1

. Mea

ns,

stan

dard

dev

iati

ons,

and

zer

o-or

der

corr

elat

ion

mat

rix

Var

iabl

es

Mea

n (J

F

ailu

re

2 3

4 5

6 7

O.

Num

ber o

f coo

rdin

atio

n fa

ilur

es

3.68

4 2.

136

1.

Lev

el o

f net

wor

k co

nstr

aint

.2

19

.006

.5

16

2.

Net

wor

k hi

erar

chy

.043

.0

15

.661

.6

30

3.

Num

ber o

f tas

k in

terd

epen

denc

ies

10.6

32

3.8

62

.436

-.

136

.088

4.

Num

ber o

f pro

ject

team

s 10

.158

5.

786

.566

.0

14

.315

.7

44

5.

Adj

uste

d co

ordi

nati

on f

ailu

res

• .0

00

1.80

7 .8

46

.302

.3

52

.490

.5

29

6.

Adj

uste

d le

vel o

f con

stra

int'

.000

.0

05

.292

.8

73

.340

-.

179

-.10

9 .3

46

7.

Adj

uste

d ne

twor

k hi

erar

chy'

.0

00

.Oll

.4

06

.406

.7

32

.081

.2

23

.481

.4

65

8.

Num

ber o

f wea

k co

nsul

tati

ve ti

es

8.89

5 1.

823

.533

.4

88

.682

.0

42

.223

.0

00

.000

.0

00

a A

djus

ted

scor

es a

re th

e st

anda

rdiz

ed r

esid

uals

of r

egre

ssin

g th

e re

spec

tive

raw

mea

sure

on

the

num

ber

of w

eak

ties

in

the

man

ager

's n

etw

ork.

w - """

~

-tl c a ~ ~

~

is' - ~ '1:

:i [ ~ t-<

is' ~

~


Table 2. Structural hole effects on the number of individual coordination failures (Standardized OLS coefficients)

Structural Hole Effects Beta t-value R:l F

Network Constraint

Controlled by project participation .508*** 3.128 .578 10.978***

Controlled by task interdependencies .586*** 3.337 .527 8.927***

Removing weak tie effects:

Controlled by project participation .408** 2.176 .444 6.394**

Controlled by task interdependencies .448** 2.343 .434 6.136**

Network Hierarchy

Controlled by project participation .536*** 3.135 .579 11.007***

Controlled by task interdependencies .627*** 3.863 .581 11.098***

Removing weak tie effects:

Controlled by project partiCipation .382* 1.950 .418 5.745**

Controlled by task interdependencies .444** 2.353 It 1.435 6.167**

*p < .10; **p< .05; ***p<.OI; N=19

correlation creates multicollinearity problems that make it impossible to have both controls in the equation simultaneously. To circumvent that problem, we estimated two separated models. We expect both variables to have a statistically significant positive impact on failure rates. We are not, however, interested in this impact, but only on the changes in the effect of the structural variables after controlling for workload conditions. Table 1 presents descriptive statistics and zero-order correlations between all these variables.

RESULTS

Table 2 presents standardized regression coefficients measuring the impact of the structure of the managers' networks on the number of coordination failures incurred by this manager. The results furnish evidence supporting our predictions. The effects are statistically significant for both network constraint and network hierarchy. Network constraint increases the number of coordination failures committed by managers. The impact of network constraint is still strong after removing the effect of weak ties, which suggests that the number of weak ties in the managers' consultation network play a marginal role in accounting for coordination failures. Rather, these failures are the result of how those weak ties are allocated to the different task interdependencies. Managers with a network lacking structural holes were more likely to allocate weak consultation ties to strong interdependencies and strong consultation ties to weak or non-existent interdependencies, which prompted coordination failures. Similar results are obtained using network hierarchy. The tendency to have a consultation network built around strong ties to one or few players makes managers more likely to allocate weak consultation ties to strong task interdependencies. Workload conditions were an important factor contributing to managerial failures. The effects of the structural variables, however, were not substantially affected by the introduction of the control variables. This suggests that those effects are largely independent of the managers' workload. II


Both network constraint and network hierarchy have comparable effects increasing coordination failures. The high correlation between these two independent variables (r = .661 for the standard measures and r = .465 for the adjusted measures; p < .01), coupled with the small number of observations, create multicollinearity problems that made it difficult to evaluate the relative impact of each of these structural properties on coordination failures . Attempting an approximation to such an evaluation, we obtained instrumental measures of constraint and hierarchy by removing the effect of the network constraint from network hierarchy, and vice-versa. Regressing the number of coordination failures on network constraint and on the instrumental hierarchy measure, we obtained estimates for the impact of each variable on coordination failures. A similar estimation was done using the instrumental measure of constraint and network hierarchy. The results, however, do not allow for an empirical evaluation of differences in the impact of each structural property on coordination failures. A comparison of the corresponding standardized regression coefficients shows that the impact of the instrumental measure of hierarchy is similar to the impact of the instrumental measure of constraint. 12

Going one step further in the analysis, we explored the sources of variation in the shape of the manager's consultation networks. More specifically, we investigated individual and organizational sources of constraining relationships. To this end, we examined constraint scores at the dyad level and sought to identify correlates of highly constrained relationships. We used the marginal constraint posed by each contact relatively to the average constraint in the manager's network. Contacts posing high constraint are strongly tied to the focal manager and to this manager's other contacts. 13

We found no statistically significant association between relative constraint and similarity in background variables such as educational level, major, seniority, or age. The average DPI manager was neither more nor less likely to be constrained by a colleague like him or herself, when similarity is defined along these 'human capital' attributes. Relative constraint was also unrelated to formal reporting relationship. There was, however, a clear association between relative constraint and organizational background, with average relative constraint higher among people who were together in the same unit before being assigned to DPI (2.66 t-test; p < .01). Similar results were obtained using raw constraint scores (cij) or the relative strength of the tie (Sij). The dominant players in the managers' networks were more likely to be the people the managers used to work with in their previous assignment in the organization. As we mentioned before, ten of the nineteen DPI managers came directly from a unit that launched some of these initiatives and continued to work on their projects until these were completed or terminated. Another group of six people also came from a single unit within the firm. The relatively smooth transition between those previous assignments and DPI, which in some cases even implied a continuation of the same project work, with the same people, did not always signal a change in the task environment of these managers. As DPI started its own projects, the task interdependencies of the managers shifted, but the change went unnoticed for a number of them, still faithful to consultative relations built over years in the firm.


This finding confirms some well-known ideas about the origin of strong relationships. Granovetter (1973: 1361) has pointed out that the strength of a tie' ... is a (probably linear) combination of the amount of time, the emotional intensity, the intimacy (mutual confiding), and the reciprocal services that characterize the tie.' Looking at the origin of ties, Feld (1981) stressed joint participation in similar organizational contexts as one of the main sources of relationships. Common organizational history puts people in contact, prompts the exchanges of advice and services, and allows for repeated exchanges that are the basis for a strong relationship. Our analysis suggests that these relationships are likely to outlive the specific context in which they arise. If the change of context is not drastic enoughsuch as physical relocation-the old relationship may remain strong. This was particularly so in the case of DPI. A large number of the observed managers were previously working together in the unit that preceded DPI and initially continued to do so in the new unit, often in similar projects, even if new colleagues often joined the teams. The smooth transition did not signal a clear change in the task environment for these managers, who continued to consult with their old colleagues even when the new projects created interdependencies with managers coming from other parts of the organization.

Our analysis of the DPI managers furnished evidence supporting the propositions discussed in this chapter. Lack of structural holes in the network that defines social capital makes it difficult for the manager to renew this capital as required in a changing task environment. The results are independent of the potentially spurious influence of the number of weak ties in the consultation networks and from differences in workload conditions across managers. The results, however, did not shed light on the relative importance of the two ways in which a constraining network structure may affect a manager' s ability to renew his social capital. Both strong ties to cohesive cliques and the presence of dominant contacts in the manager's network equally jeopardize his ability to adapt his social capital. An examination of the sources of the constraining relationships responsible for the lack of structural holes in the managers' networks revealed that those relationships typically corresponded to ties forged through years of work in similar organizational environments. The strength of those bonds, and the lack of tangible signs of change in the managers' task environment, galvanized the mechanisms of reciprocity and relational inertia and increase the risk of coordination failures.

DISCUSSION AND CONCLUSION

This chapter draws upon the control implications of structural hole theory to explore the dark side of social capital. We explain how the structure of a manager's contact network may affect his ability to adapt the content of his social capital. Faced with a growing number of task interdependencies that are mostly beyond his administrative control, a manager must develop informal mechanisms to effectively coordinate the uncertainties surrounding these tasks. In this context, a manager's social network is a critical resource to amass the necessary corporate social capital. Not any kind of social capital, however, will be equally useful in such an environment. Effective managers should be able to strategically adapt the composition of their social network to help coordinate shifting interdependencies. To do so, however, they must


have the autonomy to develop, maintain, and leverage relationships, thus keeping control on how they allocate time and energy to their contacts. Factors that reduce such autonomy increase the likelihood of failures in coordinating critical task interdependencies, which in turn decrease the available corporate social capital and may have a negative impact on performance. Our research illustrates the impact of two such factors. A consultation network formed by strong ties with a set of cohesive contacts, or a network dominated by a small set of contacts, constrain the manager's ability to adapt the composition of his social capital according to the changes in his task environment. Conditions that facilitate the creation of strong ties with densely connected clusters, such as a common organizational history, increase the probability that managers will build constraining networks, hence affecting their subsequent ability to renew their social capital.

Our research has consequences for the debate on social capital, as well as for its effects on managerial and organizational performance. By focusing on factors that may hinder the renewal of social capital in dynamic task environments, we showed how the network structure that carries social capital could have an effect on the manager's ability to renew the composition of this social network. Network structure, network composition, and outcomes are interdependent, but the relationship between these three variables is problematic. Early studies of social capital showed how the composition of an individual's contact network had an impact on the benefits accruing to this individual (Lin, Ensel, and Vaughn 1981; Lin and Durnil 1986; Flap and de Graaf 1989).

More recently, scholars have focused on network structure as the motor behind those benefits, assuming implicitly or explicitly that network composition is a simple correlate of network structure (Burt 1992; Podolny and Baron 1997). By focusing on the renewal of social capital, our research clarifies the relationship between network structure and network composition, elucidating the mechanisms through which this relationship operates. Our results suggest that both ties to a cohesive cluster and strong, exclusive ties to few contacts are conditions may turn the assets of social capital into social liability that damage a manager's adaptability and performance in changing task environments.

The significance of our findings is enhanced by the fact that the relationship between network structure and favorable outcomes is not stable across contexts. If lack of structural holes was always detrimental for the performance of a manager, our results simply would add another spin to this deleterious effect. Yet, this is not always the case. Existing research suggests that the effects of structural holes on managerial performance may depend on the particular situation of the manager. In his analysis of managers in a U.S. high-tech firm, Burt (1992: 147-153) found that strong ties to prominent contacts in a manager's network was the best route to fast promotion for women and entry-level men. He suggests that such ties may be a necessity for a manager in need of a legitimating sponsor. This positive effect of what amounts to an informal 'network hierarchy' on promotion captures the weIldocumented influence of mentoring in managerial careers (e.g., Kram 1988). With a different emphasis, Podolny and Baron (1997) argued that strong ties to cohesive contacts linked to a manager by authority relations may also be beneficial in early stages of their career, when defining an identity and obtaining support from


powerful players is critical for the manager. A manager with strong ties to a cohesive group may attain the necessary legitimacy to be recognized as a player. He may also profit from the additional social capital he can 'borrow' from his sponsors. These benefits, however, may come at a price. Our evidence suggests that strong ties to a cohesive group may eventually curtail a manager's ability to adapt the composition of his social capital, even if those ties were initially beneficial for the manager. This result is congruent with Higgins and Nohria's research (this volume) on the negative effects of early mentorship on the ability to develop social capital at a later career stage. The difficulty in renewing the composition of his social capital affects the managers' ability to coordinate task interdependencies with and among other members of the organization, hence having a deleterious effect on his performance.

Our findings, therefore, pose an interesting dilemma for managers as well as for organizations. On the one hand, managers at the early stages of their career may need to obtain decisive informal sponsorship to become legitimate players (Burt 1992) or to assert their identity in the organization (Podolny and Baron 1997). This legitimacy requirement is compounded by the fact that the lack of structural holes is less detrimental for entry level managers, where the benefits of a diverse contact network may be negligible (Burt 1997). A small, cohesive core of supportive contacts may be the best form of social capital for these managers. On the other hand, the managerial career eventually requires autonomy to allocate time and energy to different contacts to successfully cope with the shifting task interdependencies associated to the role. Yet, the same cohesive social network that was instrumental in asserting the manager's position in the organization may be an obstacle to develop the type of social capital required to further his professional growth and his capacity to add value within this organization. Our argument suggest that this peril may be more consequential when the new task is not accompanied by tangible changes in reporting line or physical location that naturally weaken the manager's previous relationships.

Ideally, managers should be able to go through a smooth transition between these two stages in their careers, adapting both the structure and the composition of their social capital to the changes in their task environment. The human resource practices and the culture of the organization should also support this transition. Although the observed differences in network structure between senior and junior managers suggest that a good number of people does succeed in adapting their social capital to the growing complexity of their task environments, the transition may not be always easy. Moreover, the focus on 'success' stories may hide the fact that a considerable amount of managerial talent could be wasted in the process. The initially supportive, legitimating sponsorship of a cohesive cluster or a strong mentor may translate into a liability that hinders the manager's ability to adapt his social capital in later stages of his career (Higgins and Nohria, this volume). For the young manager, there is no easy way out of this dilemma. For the organization, the solution lies in efficiently combining the benefits of sponsorship with the benefits of autonomy. Factors that make the transition more noticeable, such as physical or departmental relocation may help avoiding the perils of the dark side of social capital discussed in this chapter. Yet, these systemic solutions may not be always


feasible or even desirable. Another, perhaps complementary way to deal with the problem is to clarify the role of organizational sponsors and to build in the ability to promote young 'entrepreneurial' managers into their performance evaluation. Like a wise father, a wise sponsor must know not only how to help and defend his protege, but also to allow him the necessary autonomy to grow independently.

Finally, our research has implications for the study of the managerial role in the dynamic, flat organizational environments that increasingly penetrate today's firms. We suggest that the attenuation of authority embedded in the idea of 'flat' organizational structures may not suffice to promote entrepreneurship within a traditional organization and to create the type of corporate social capital that is key to its success. This is especially so in context of rapid organizational change. Like the bird that remains inside the cage despite the open door, managers adapted to a command-and-control structure may fail to use the freedom created by flatter structures, even when they are convinced of the need to do so. Scholars have recognized the role of previous socialization in this inability to take advantage of the new freedom. Our argument suggests that the freedom itself may be sometimes a mirage. Powerful control mechanisms embedded in informal structure and rooted in the organizational history can create subtle bonds of interpersonal dependence that may severely curtail the organization's ability to move away from a command-andcontrol managerial style and to create the corporate social capital required in an entrepreneurial firm. Although necessary, the attenuation of authority is not a sufficient condition to promote managerial entrepreneurship. Hierarchies ingrained in the informal organization may still create an effective obstacle to flexibility. The unobtrusive nature of these ties may make their effect less apparent than the conspicuous impact of the formal hierarchy. Such an effect, however, may be equally consequential.

This warning is especially relevant in contexts of the transformation into a flat, entrepreneurial organization. Moving away from the well-established commandand-control structures is a task that requires more than administrative decisions. Top management decisions of reshaping the overall organizational through a general reduction of managerial levels are important. However, such decisions may not be enough to promote entrepreneurship. Past organizational characteristics can survive, even if formal structures have been revised or the chain of command reduced. A manager cannot act as a flexible network entrepreneur if the informal organizational structure does not allow it. Yet, the emergence of this structure is largely the result of effective managerial initiative. Informal structures that result from established managerial practices can have a direct impact on the behavior of those managers, which in turn may affect the organization's ability to effectively enact the change.

The research reported in this chapter shed light on the 'dark side' of social capital by discussing when and how social networks may become an obstacle to managerial and organizational performance. We show how strong, cohesive social bonds may hinder the necessary renewal of social ties to fit changes in the manager's task environment, hence diminishing the level of corporate social capital available to the firm. We suggest that the risk can be higher if those strong ties have initially produced substantial advantages for the manager, a condition that defines the effectiveness of social capital. This paradox reminds us that social capital, like

The Dark Side of Social Capital- 321

other properties linked to social structures, may have both positive and negative consequences for individual and organizational action. Like most human things, social capital has a bright side, but also a dark one.

This paper was written while the first author was a Visiting Scholar at Columbia University's Graduate School of Business (1997-98 academic year). We would like to thank helpful comments by Ron Burt, Ranjay Gulati, David Gibson, Roger Leenders, Andrej Rus, Harrison White, and the participants in the Network Analysis Workshop at the Paul Lazarsfeld Center for the Social Sciences at Columbia University.

NOTES

I. This phenomenon has been documented in studies of organizational adaptation, that show how the same processes that help organizations to be well adapted to their current environment can curtail their ability to adapt to a new environment (Grabher 1993). A similar idea plays a central role in ecological models of organizations (Hannan and Freeman 1989). These models argue that 'specialists' firms-that is, organizations highly adapted to specific environments-are less likely to survive drastic environmental changes than 'generalists' organizations, since the latter are less dependent on specific resources for their survival. 2. Leifer (1988) and Bearman (1997) note how the norm of reciprocity, when applied to dyadic exchanges, may lock the players into endless exchanges due to ambiguity on the valuation of the 'gifts' exchanged. A similar argument can be made regarding exchanges with a cohesive group. In this case, the presence of third parties reinforces compliance, thus increasing the likelihood of people entering such endless exchanges. 3. Structural hole theory builds upon a key intuition of exchange theory, which states that control in a relationship is a positive function of the availability of alternatives (Emerson 1962; Blau 1964). A series of simulation and experimental studies confirm the adequacy of this basic intuition. Players with access to several exchange partners who themselves lack such alternatives enjoy competitive advantage (Cook and Emerson 1978; Marsden 1982, 1983; Cook, Emerson, Gillmore, and Yarnagishi 1983). 4. It is worth noting that our treatment of social capital differs from that of structural hole theory, which makes social capital a sole function of the structure of the network: the more structural holes, the larger the social capital. Burt (1992: 13) defends his analytical choice by assuming that actors who know how to structure a network to provide access to opportunities would also know whom to include in that network. This assumption provides an elegant solution to the tension between the constraint and opportunity aspects of social capital, but it also restricts the applicability of the concept. An individual strongly tied to a dense cluster of resourceful people has little social capital in Burt's sense. Yet, in some situations, strong connections with few strategic players may be actually beneficial for a manager (Burt 1992: 147-153; Podolny and Baron 1997). This effect is predicted by the more traditional definition of social capital adopted here, but it does not fit easily within structural hole theory. More importantly for this chapter, Burt's assumption dissolves the problem of adaptation of social capital that is central to our discussion. Paradoxically, this drives the attention away from the fruitful control implications of structural hole theory developed here. 5. In the consultation network, centrality scores varied from 1.00 to .594, with an average of .791 and a standard deviation of .118. Centrality scores reflect a manager's tendency to be consulted by colleagues who are also preferred targets for consultation (Bonacich 1987). The average strength of the ties in the network is 1.32, with an effective range from 0 to 3. The average DPI manager reported some level of consultation with II different colleagues. Out of the 342 possible consultation ties, 62.6 percent are present; among these, 23.1 percent are strong, 23.7 percent medium and 15.8 percent weak. 6. The inclusion of both types of consultation is relevant in this context, since a manager may still have to allocate time to colleagues who consult him, even if he does not seek these people out. Granted, when two managers meet informally to discuss work-related matters at the initiative of one of them, the other may still bring up issues that are of his specific concern. Yet, if this second manager does not report the interaction-or reports a weaker interaction-he is probably less likely to initiate these exchanges, which in tum should lead to a weaker overall relationship.


7. The Coleman-Theil index is calculated as follows:

L --.9L [In --.9L ] Ai= ICfN CfN

NIn(N) where C is the lack of structural holes (aggregate constraint) on actor i (LjCij) and N is the number of colleagues i consults with andfor is consulted by. Note that the quotient CIN is the average constraint posed by i' s alters, and thus ci/(CIN) expresses the constraint posed by actor j in terms of the average constraint in i' s network. 8. The mean strength of task interdependence between managers is .056, with a median of .038 and a skewness coefficient of 2.035. Sixty percent of the interdependence ties fall below the mean and hence could not prompt a coordination failure as defined in this chapter. Thus, a low level of consultation with a sporadic project partner need not be a coordination failure. Consultation ties are normally distributed (.056 mean, .057 median; .033 skewness). The proportional measures of task interdependence (Tij) and consultation (Sij) used in this chapter removes differences in the volume of interaction of individual managers, making average figures a good indicator of the socially expected levels of interdependence and consultation between any two managers in the unit. Given the distribution of interdependence and consultation ties, managers clearly had enough 'strong' consultation ties to adequately coordinate all their strong task interdependencies. Failures resulted from the way they allocated their strong consultation ties. 9. At the project level. we defined a coordination failure index as the ratio between the number of DPI dyads in the project team who failed to coordinate and the total number of DPI dyads in the project. For this analysis, projects with a single DPI participant were coded as having zero failure rate. There were 65 projects in which at least one member of a unit other than DPI participated and 58 with 2 or more other units. Cooperation was measured as the average level of cooperation between team members as reported by the manager coordinating the project, on a scale from 0 (no cooperation among the team members) to 3 (strong cooperation). Reported values have a mean of 1.7 and a standard deviation of 0.8. As one might expect, average cooperation in project teams was highest between DPI managers (2.1), intermediate between DPI and other team members (1.8) and lowest between members of other units (1.3). 10. On average, DPI managers had 8.9 weak ties, ranging from 4 to 12. As expected, the number of weak ties in the consultation network was significantly correlated with the three variables, that is, number of failures (r = .533; P < .05), network constraint (r = .488; p < .05), and the hierarchy (r = .682; p < .01). These high correlations ruled out an alternative approach to control for the effect of weak consultative ties on coordination failures by simply including the number of weak ties into the regression. II. We also estimated models where the dependent variable is the rate of coordination failures-that is, the number of failures divided by the total number of interdependencies-and obtained similar, albeit slighty weaker results. Using the rate of coordination failures as a dependent variable, standardized beta coefficients using the instrumental variables that control for the number of weak ties in the network are .366 for constraint (p = .06) and .451 for hierachy (p = .03). Note that the rate of failure automatically control for workload effects, which thus become statistically insignificant. 12. These results were computed using the adjusted measures of our variables and thus are independent of the number of weak ties in the manager's consultation network. None of the coefficients for the instrumental measures were statistically significant at the .10 level. 13. Relative constraint, given by the ratio ci/(CIN), is largely responsible for the level of hierarchy measured by the Coleman-Theil index (see note 7). Thus, consultation networks were often built around co-workers who had similar careers in the firm.

•

Social Capital, Social Liabilities, and Social Resources Management 17

Daniel J. Brass Giuseppe Labianca

ABSTRACT

This chapter explores the role of social capital in human resources management. We suggest that the recent interest in social capital has neglected the possibility that social networks may contain negative ties, and that attention to these social liabilities may provide additional insights into relationships and social networks in organizations. Research focusing on the antecedents and consequences of social networks in organizations is reviewed. We consider the effects of social capital and social liabilities on 'social' resources management outcomes such as recruitment, selection, socialization, training, performance, career development, turnover, job satisfaction, power, and conflict.

INTRODUCTION

Human resources management, as the name implies, has persistently defined its task and focused its energy on developing methods of measuring individuals' human capital. To focus on the individual in isolation is. at best, failing to see the entire picture. Rather, people are embedded within a network of interrelationships with other people. These networks of relationships provide opportunities and constraints that make up the social capital of the individuals and the larger system. Social capital inheres in the social relationships that can potentially confer benefits to individuals and groups (i.e., Bourdieu 1972; Burt 1992; Coleman 1988, 1990; Fukuyama 1995a; Gabbay 1995, 1997; Putnam 1995b). Social capital can be contrasted with human capital (one's knowledge, skills, and abilities), financial capital (money), and physical capital (physical property such as land, buildings, machinery). We do not mean to suggest that individuals do not differ in their human


capital, their skills and abilities and their willingness to use them. Nevertheless, we will try to nudge the study of human resources management toward the study of social resources management; from human capital toward social capital.

Coleman (1988) suggests three forms of social capital. First, obligations and expectations, and the trust that facilitates them, arise when actors are willing to do something for other actors because they expect and trust that the recipients will honor the obligation to reciprocate in the future. The second form of social capital refers to actors obtaining information that may be useful to them from others, indirectly taking advantage of others' knowledge, skills and other forms of human capital. The third form of social capital, norms and sanctions, allows for the reduction of transaction costs.We expand the notion of social capital by adding to it the role of negative relationships-relationships in which at least one person has a negative affective judgment of the other. Although relationships create opportunities and benefits, the current focus on social capital emphasizes only the positive aspects of social networks while neglecting the potential liabilities that may be associated with negative relationships. Although the early social exchange theorists and network researchers considered both the positive and negative aspects of relationships (e.g., Homans 1961; Tagiuri 1958; Thibaut and Kelley 1959), recent network research has focused almost exclusively on the positive aspects of social structure, with little exploration of any possible negative aspects. The question of where and when the potential benefits outweigh the potential liabilities of expanding one's network has been left largely unanswered. Rather than solely investigating social capital, we attempt to consider the 'social ledger'-both the potential benefits as well as the potential liabilities of social relationships (Labianca and Brass 1997). Just as a financial ledger records financial assets and liabilities, the social ledger is an accounting of social assets (social capital) and social liabilities (negative relationships). We also suggest that the liability side of the social ledger, negative relationships, may have greater explanatory power than positive relationships.

THE SOCIAL LIABILITIES OF NEGATIVE RELATIONSHIPS

In suggesting that relationships have benefits, researchers have found that one's social network relates to such outcomes as job attainment, job satisfaction, power, and promotions in organizations (e.g., Brass 1984; Burt 1992). The consideration of social liability may add to our understanding of these outcomes. For example, the underlying assumption is that an employee's friends and acquaintances help the employee obtain promotions by providing such forms of social capital as critical information, mentorship, and good references. However, it is also likely that an employee's negative relationships with others in an organization might prevent promotion, particularly if those with whom the employee has negative ties are in influential positions. Those people may withhold critical information or provide bad references in order to prevent promotion. Thus, it may be equally important to consider the negative side of the social ledger: social liabilities as well as social capital.

We define negative relationships as relationships in which at least one person has a negative affective judgment of another person. Prior research has viewed the negative aspects of personal relationships in two ways. One perspective assumes that

Social Capital, Social Liabilities, and Social Resources Management - 325

every relationship contains both positive and negative aspects and that these aspects are independent (e.g., Bradburn 1969; Diener and Emmons 1985; Russell 1979; Watson and Tellegen 1985). We adopt a second perspective that acknowledges the first, but further assumes that people form a global judgment of others that can be captured by such terms as 'like' and 'dislike,' that are opposite ends of a continuum (e.g., Berscheid and Walster 1969; Newcomb 1961; Tagiuri 1958).

We distinguish between social liabilities (negative relationships) and the opportunity costs of building social capital. We note that the process of developing social capital may include opportunity costs as well as benefits. For example, developing strong, trusting relationships may interfere with the opportunities for bridging diverse groups via weak ties (see Gargiulo and Benassi in this volume). As Granovetter (1985) noted, the obligations and expectations of strong, long-lasting relationships may prevent a person from realizing greater economic opportunities by constraining the search for, and development of new trading partners. Gabbay (1995, 1997) and Gabbay and Leenders (this volume) have noted that some actors may be constrained by the same social structure that benefits other actors.

In their chapters in this volume, Gabbay and Leenders the term 'social liability' to denote the constraining effects of social structure. Social liability is the opposite of social capital. In the current chapter, we use the term to explore the potential liabilities that can result from negative relationships.

NEGATIVE ASYMMETRY

The importance of considering the social liabilities of negative relationships is based on our hypothesis of negative asymmetry: negative ties may have greater explanatory power than positive ties in organizations. This hypothesis is based on indirect evidence from research in the area of social support in health care (see Labianca and Brass 1998 for a review), and a diverse psychological literature indicating that negative events elicit greater physiological, affective, cognitive, and behavioral activity and lead to more cognitive analysis than neutral or positive events (see Taylor 1991 for a review). In addition, Burt and Knez (1995) found that third parties amplified the effects of gossip, and that the amplification effect was stronger for negative gossip than positive gossip. In an organizational field study, Labianca, Brass, and Gray (1997) found that negative relationships increased perceptions of intergroup conflict, but strong positive relationships had no counterbalancing effect. Why do negative events and relationships have more impact than positive events and relationships? Evolutionary psychologists explain the negative asymmetry by noting that it is adaptive to respond quickly to negative events in order to enhance survivability (cf. Cannon 1932). Developmental psychologists suggest that negative events are discriminated and evaluated earlier by children than are positive events because negative events are more likely to interrupt action. Children learn the rules governing negative behavior before those governing positive behavior; children are punishment oriented (cf. Piaget 1932). Nature and nurture combine to make humans risk-averse (Kahneman and Tversky 1984).

In seeking to theoreticaIIy explain negative asymmetry, Skowronski and Carlston (1989) summarize a number of theories. For example, negative events dominate social judgment because of the contrast effects with positive and moderate


events that people typically experience and expect. Since people expect positive, moderate information, negative, extreme information is weighted more heavily in impression formation. Negative information is also attended to because it is more unambiguous than positive information; it allows people to make social judgments more easily.

In addition to direct social information, indirect social information can also result in negative relationships. Second-hand information is filtered and simplified so as to be unambiguous, and amplification of negative aspects is often more pronounced that positive information (Burt and Knez 1995).

ANTECEDENTS OF SOCIAL CAPITAL AND SOCIAL LIABILITIES

Although previous discussions of social capital have focused on outcomes, In

particular, the beneficial outcomes of social relationships, it is important to understand the antecedents of social relationships and how they might affect social capital and social liabilities. We now consider the factors that affect social relationships, and social resource management in organizations.

Organizational Structure Positions in organizations are formally differentiated both horizontally (by technology, workflow, task design) and vertically (by administrative hierarchy), and means for coordinating among differentiated positions are specified. Because communication is a fundamental means of coordination, it follows that social relationships are influenced by the prescribed vertical and horizontal differentiation and the resulting need for coordination. Formally differentiated positions locate individuals and groups in physical and temporal space (workers assigned to different time shifts), and at particular points in the workflow and hierarchy of authority. Organizational structure therefore restricts opportunities to interact with some people, and facilitating interaction with others. As Festinger, Schacter, and Back (1950) found, physical proximity is related to amount of interaction. Despite the use of telephones and electronic mail, proximate physical and temporal ties are easier to maintain and more likely to be strong, stable links (Monge and Eisenberg 1987).

The restrictions on social interaction imposed by the formal organizational hierarchy and workflow requirements are particularly important when considering negative relationships. In everyday, non-work activities, actors can easily avoid or decrease interactions with others whom they dislike. However, the formally prescribed, required interactions in organizations create the possibility of required negative relationships; relationships with disliked others that cannot be avoided. Thus, the effects of negative relationships on social capital and the social ledger may be particularly relevant in organizational settings.

Size The size of an organization (number of employees) may also affect the social relationships and social capital. As the size of the organization increases, network density naturally decreases (assuming that actors can maintain only a limited number of ties), and the possibility of fragmentation (individuals forming sub-groups) increases (Shaw 1971). Similarity and increased interaction result in


strong ties forming among sub-group members, and decrease the probability of strong, positive connections across groups. Thus, decreased density may make it more difficult to maintain the 'closure' needed for effective norms and sanctions (Coleman 1988), and the trust necessary for obligations and expectations may be more difficult to establish in larger systems.

Actor Similarity Although social networks likely shadow the formal structure of the organization, employees may informally modify the prescribed interaction. That is, we do not ignore human agency. We do not assume that organization structure completely constrains individual action. Instead, we adopt a bi-directional perspective (Giddens 1976; Leenders 1995b)-structure and behavior are mutually causal. Interactions that occur within the constraints of structure can gradually modify that structure. Individuals break relationships and build new ones. Thus, it is important to understand why individuals choose some relationships and not others.

A good deal of research has noted a tendency for similar people to interact and it is a basic assumption in many theories (Homans 1950; Davis 1966; Granovetter 1973; Blau 1977). Homophily (interaction with similar others) results from ease of communication and increased predictability of behavior, and it fosters trust and reciprocity. Brass (1985a), Ibarra (1992), and Mehra, Kilduff, and Brass (1998) have found evidence of homophily by gender in organizations. For example, Ibarra (1992) found that men had homophilous ties (with other men) across multiple networks, whereas women had social support and friendship network ties with other women, but they had instrumental network ties (e.g., communication, advice, influence) with men. In addition, perceived similarity (religion, age, ethnic and racial background, and professional affiliation) among executives has been shown to influence interorganizationallinkages (Schermerhorn 1977; Galaskiewicz 1979).

In combination with age, sex, status, etc., we would expect similarity of human capital characteristics such as personality and ability to be related to the interpersonal network patterns of interaction and social capital. We also would expect the characteristics of the links between actors to be related to the degree of actor similarity. Communication between two dissimilar actors is likely to be infrequent, not reciprocated, less salient to either, asymmetric, unstable, uniplex rather than multiplex, weak, or even negative. Similarity of actors also may be positively related to the density or connectedness of the network.

Similarity may be a necessary precondition to social capital that results from all three forms noted by Coleman (1988). Individuals are more likely to trust similar others in reciprocating obligations and expectations, similar others are more likely to share information, and dense networks of similar others are more likely to develop and maintain norms and sanctions. For example, Coleman has argued that 'closure' of the network is essential for the effectiveness of norms. We emphasize actor similarity, and later, attitude similarity, because both are key to such human resources practices as recruitment and selection.

It is important to note that similarity is a relational concept; an individual can only be similar with respect to another individual, and in relation to dissimilar others. That is, interaction is influenced by the degree to which an individual is


similar to other individuals relative to how similar he or she is with everyone else. Due to culture, selection and socialization processes, and reward systems, an organization may exhibit a modal demographic or personality pattern. Kanter (1977) has referred to this process as 'homosocial reproduction.' Thus, an individual's similarity in relation to the modal attributes of the organization (or the group) may determine the extent to which he or she is central or integrated in the interpersonal network, and the extent to which he or she shares in the social capital of the organization.

Conversely, actor dissimilarity may be a source of negative relationships and exclusion from the benefits of social capital. Actors are less likely to trust dissimilar others, share information with them, or include them in the norms of the system. For example, research has shown that relational differences on such demographic variables as age and tenure are related to commitment and turnover (Tsui, Egan, and O'Reilly 1992; Tsui and O'Reilly 1989; Wagner, Pfeffer, and O'Reilly 1984; Zenger and Lawrence 1989). Although all people possess some similarity, and some capacity to identify with others, research on 'moral exclusion' suggests that dissimilar others may be excluded from social relationships and the targets of unethical behavior (Brass, Butterfield, and Skaggs 1998; Opotow 1990; Smith 1966).

The exclusion of individuals from the social network can be detrimental to the overall social capital of the system. The diverse, non-redundant information that may come from relationships with dissimilar others is lost (Burt 1992), and the overall creativity and productivity of the collective may be diminished (Brass 1995a). Thus, exclusion of dissimilar actors, or negative relationships with dissimilar others, decreases the amount of system-level social capital that is available to anyone.

Attitude Similarity In addition to noting the propensity for similar actors to interact, theory and research have also noted that those who interact become more similar in their attitudes (see Krackbardt and Brass 1994 and Leenders 1995b for a detailed discussion). The relational basis of attitudes (Erickson 1988) assumes that people are not born with their attitudes, nor do they develop them in isolation. Attitude formation and change occur primarily through social processes, such as communication and comparison (Leenders 1995b). As people attempt to make sense of reality, they talk with others and compare their own perceptions with those of others, in particular, similar others. For example, Kilduff (1990) found that M.B.A. students made decisions similar to their friends' regarding job interviews with organizations.

We build upon Erickson's ideas by suggesting that negative relationships may foster attitude dissimilarity. The dislike of another actor may prompt one to take opposing positions on issues. Disagreements can be attributed to dissimilarity, and may even be used to reinforce one's own attitudes. Thus, our understanding of attitude similarity and social capital may be enhanced by accounting for negative relationships as well as positive relationships. Negative relationships may interfere with attitude similarity and disrupt social norms, expectations, and obligations, especially when several negative relationships exist in a network. The presence of a


relative few negative relationships and dissimilar attitudes may actually strengthen group norms, just as the overt intolerance of deviant behavior reinforces norms.

Although attitude similarity has typically been considered an outcome of social interaction (Erickson 1988), we view it as an antecedent to social interaction as well. That is, actors may seek out others with similar attitudes, and avoid those with dissimilar attitudes, thereby reinforcing their own perceptions of reality. As Leenders (1995b) notes, it may be as easy for actors with similar attitudes to seek and find each other as it is for one actor to influence the attitudes of another. For example, it is likely that actors with similar demographics, attitudes, and behaviors will meet in similar settings. Activities are organized around 'social foci' (Feld 1981). Football fans meet in the stadium to watch a football game, school children of similar age meet at school to be educated, accountants meet in the accounting departments of organizations. Actors who gather around a particular social focus are likely to be similar, and interaction with each other will likely enhance that similarity (Leenders 1995b).

Just as actor similarity (or dissimilarity) may affect social capital, attitude similarity may have similar effects. As with actor similarity, the trustworthiness of similar expectations and obligations, as well as norms and sanctions, may depend upon similar attitudes among organizational members.

SOCIAL CAPITAL AND SOCIAL LIABILITIES: THE OUTCOMES OF SOCIAL RELATIONSHIPS

We now turn our attention to the outcomes of social relationships. We consider both social capital and social liabilities in relation to typical human resource management outcomes.

Job Satisfaction The early laboratory studies of small groups found that central actors were more satisfied than peripheral actors (see Shaw 1964 for a review). Since that time, research in organizations has found that relative isolates (zero or one link) in the communication network were less satisfied than participants (two or more links) (Roberts and O'Reilly 1979), that job characteristics (such as autonomy and variety) mediated the relationship between workflow network centrality measures and job satisfaction (Brass 1981), and that betweenness centrality in the friendship network was negatively related to job satisfaction (Kilduff and Krackhardt 1993). Kilduff and Krackhardt reasoned that conflicting expectations and stress may result from mediating the relationships between actors who are not themselves friends. Thus, when linking actors who are not connected themselves, central actors may not enjoy the common expectations and obligations of social capital.

Kilduff and Krackhardt's (1993) findings remind us that interaction is not always positive. Since Durkheim (1897) argued that social integration promotes mental health, there has been a long history of equating social interaction with the positive aspects of social support (Wellman 1992). Yet we have all experienced negative interactions with others in the workplace. We try to avoid these negative relationships whenever possible, thereby producing a positive relationship between interaction and friendship. However, prescribed interactions are not always positive


or avoidable. As the previously cited literature on negative asymmetry indicates, negative relationships may overwhelm the job satisfaction resulting from positive social relationships.

Power It is also possible that the effects of centrality on job satisfaction are indirectly related to power. Network studies of power have centered on the social capital of information. The finding that central network positions are associated with power has been reported in a variety of settings (see Brass 1995b for a review). Theoretically, actors in central network positions have greater access to, and potential control over relevant resources, in particular the social capital of information. Just as collective information represents a type of social capital for the entire system, individual actors can acquire social capital via access to information. In addition, simple degree centrality measures of the size of one's network (number of direct links) have been found to positively relate to power (Brass and Burkhardt 1992, 1993; Burkhardt and Brass 1990).

Knoke and Burt (1983) have argued that being the object of relations (chosen by others) rather than the source (choosing others) is an indication of prestige, and a possible source of social capital. For example, Burkhardt and Brass (1990) found that all employees increased their closeness centrality (symmetric measure) following the introduction of new technology, but the early adopters of the new technology increased their (in-degree) prestige and their power significantly more than the later adopters.

The social capital represented by access to information may be related to who an actor knows as well as how central the actor is. For example, Brass (1984) and Blau and Alba (1982) found that relationships beyond the prescribed workgroup and workflow requirements were related to influence. In particular, links to the dominant coalition in the organization were strongly related to power and promotions. The dominant coalition was represented by a clique of the top executives in the company. These top executives likely had more social capital in the form of more information, and more relevant information, to share with those connected to them. Brass (1985a) also found that men were more closely linked to the dominant coalition (composed of four men) and were perceived as more influential than women. Assuming that power positions in most organizations are dominated by men, women may be forced to forgo any preference for homophily in order to build connections with the dominant coalition and share in the social capital. As previously noted, actor dissimilarity (in this case based on gender) may affect interaction patterns and consequently exclude some people from sharing in the social capital.

In addition, Brass (1985a) found evidence of the constraints of organizational structure on social relationships and power. Women who were part of integrated formal workgroups (at least two men and two women) were more closely linked to the men's network (only male employees considered) and were perceived as more powerful than women who were not in integrated workgroups. Similarly, men who were closely linked to the women's network (only women employees considered) were also perceived as more influential than men who were not. These findings

Social Capital, Social Liabilities, and Social Resources Management - 33 t

suggest that social capital in the form of information can be enhanced, or constrained by formal organizational structure, and links to dissimilar others.

None of these studies considered the possible effects of negative relationships on power. It is likely that an actor's enemies, friends of those enemies, and the enemies of that actor' s friends, determine the amount of power an actor holds in an organization. These enemies might actively attempt to thwart an actor's efforts by withholding, or distorting important information in the hope of diminishing that actor's power. The negative relationships represented by enemies are examples of social liabilities.

Recruitment In order to be recruited for a job, both parties, the individual and the organization, must know of each other. The use of social networks, as contrasted with employment agencies or job listings, can be very valuable in both searching for jobs and recruiting workers. In the classic example of the strength of weak ties, people were able to find jobs more effectively through weak ties (acquaintances) than strong ties or formal listings (Granovetter 1982). Granovetter argued that an actor' s acquaintances (weak ties) are less likely to be linked to one another than are an actor's close friends (strong ties). An actor's set of weak ties will form a low density, high diversity network, one rich in non-redundant information. A set of strong ties will be densely interconnected and will likely represent a high degree of redundant information. Thus, individuals may have greater access to more and different job opportunities when relying on weak ties. Later findings (Flap and Boxman, this volume; Granovetter 1995; Lin, Ensel, and Vaughn 1981) have both modified and emphasized the strength of weak ties.

Although weak ties may be effectively used to increase the diversity and nonredundancy of recruiting information, strong ties may be the mechanism behind homosocial reproduction in organizations (Kanter 1977). As previously noted, actor similarity is an important antecedent in building strong relationships, and repeated interaction can lead to attitude similarity and, in tum, the further strengthening of relationships. Organizations have for many years used referrals by current employees to aid in recruiting (Granovetter 1995; Neckerman and Fernandez 1997). Using the social networks of previously-hired employees represents a relatively easy, low-cost mechanism for linking the organization to potential employees (Marsden and Gorman, this volume). While these links may facilitate recruiting and decrease the turnover of newly hired applicants (Neckerman and Fernandez 1997), they may also promote homosocial reproduction. That is, current employees and recruiters seek out those whom they believe will 'fit in' well in the organization. Fit is often based on actor similarity. Actor similarity may ease the development of social capital (sharing information, establishing common expectations, obligations, and norms), but it may also decrease the diversity of information available for sharing. Thus, the benefits of social capital in recruiting may also involve costs.

Selection As Pfeffer (1989) and Ferris and Judge (1991) have noted, selection in organizations is not entirely the result of human capital (abilities and competences). Pfeffer (1989)


argued that credentials and hiring standards are often the result of political contests within organizations, rather than purely rational decision making. Those in power seek to perpetuate their power by setting criteria and selecting those applicants most like themselves. As in the case of recruiting via the use of networks, selection may also depend on network ties, and indirectly, actor similarity. In particular, we expect these effects when the qualified applicant pool is large, or when hiring standards are ambiguous. In such cases, similarity between applicant and recruiter, or indirect similarity through a common friend , may be important, but unstated, selection criteria.

Just as positive relationships may prove helpful in recruitment and selection, negative relationships may be particularly harmful. When applicant pools are large, any negative information, or any dislike of an applicant may eliminate him or her from further consideration. Thus, applicants are well advised to be risk averse with interviewing for jobs or acquiring references. Negative asymmetry is particularly apparent in letters of reference. Because typical letters are positive, any negative information is attended to and weighted more heavily in diagnosis and decision making.

Socialization Once an employee is selected for a job, socialization of that employee may be dependent on network involvement. Jablin and Krone (1987) and Sherman, Smith, and Mansfield (1986) both found that network involvement was a key process in the socialization of new employees. Similarly, Roberts and O'Reilly (1979) and Eisenberg, Monge, and Miller (1984) found that network participation was positively related to organization commitment.

However, due to the cross-sectional nature of these studies, it is impossible to know whether integration into the network leads to socialization and commitment, or vice versa. It is likely that early connections in the organizational network lead to socialization of expectations, obligations, and norms, and enhanced social capital, and that commitment leads to further integration in the network. It may be especially important that new employees do not form negative relationships early in their employment tenure, because such relationships may deter socialization and lead to turnover.

Training In traditional human resources management, the focus of training is typically on human capital-acquiring new and innovative ideas and skills. A social capital perspective on training might focus on the role of social networks in acquisition and diffusion of these skills, thus paralleling the research on adoption and diffusion of innovations (cf. Rogers 1971; Tushman 1977; Tushman and Anderson 1986; Burt 1982). Innovation is generally thought to require diverse and novel information. As previously noted, weak ties may provide more diverse and novel information than strong ties. A member of a closely knit, dense clique of strong ties is less likely to be exposed to diverse, novel perspectives than an actor with weak ties to a number of different social groups. Thus, Burt (1992) argued that the diversity of one's contacts is more important than the size of one's network. The findings that cosmopolitans


(i.e., actors with external ties which cross social boundaries) are more likely to introduce innovations than are locals (Rogers 1971) supports this diversity argument. Conversely, central actors, sometimes identified as 'opinion leaders,' are unlikely to be the source of innovations that are not consistent with the established norms of the group (Rogers 1971).

Once training is introduced, the diffusion of the training (or the spread of new ideas and skills) can be predicted by social network relationships. Some controversy exists over whether diffusion is best predicted by a cohesion (communication through direct interaction) approach or a structural equivalence perspective (observation and comparison of others in similar roles in the network). As Leenders (1995b) notes, these two theoretical approaches are practically impossible to sort out empirically. Theoretically, the direct communication, cohesion apprOl1ch suggests that the most central actors should be the first to experience training. Having the most direct and indirect contacts, central actors will quickly diffuse the skills.

In their investigation of the introduction, training, and diffusion of a technological change in an organization, Burkhardt and Brass (1990) found that early adopters were not highly central prior to the introduction of the system. However, the early adopters increased both their centrality and power in the organization as the technology was implemented. The diffusion process closely followed the new network patterns following the change, with structurally equivalent employees adopting at similar times.

In a similar study of the introduction of a new computer technology, Papa (1990) concluded that training programs can provide basic operating information, but that much of the learning about a new technology occurs after training as employees attempt to apply the training. He found that productivity following the change, and the speed at which the new technology was learned, were positively related to interaction frequency, network size, and network diversity (i.e., number of different departments and hierarchical levels contacted). Both studies (Burkhardt and Brass 1990; Papa 1990) support the idea that learning, whether through training or networks, is an active process of information exchange, and an important outcome of social capital.

Training can also be viewed as a social focus (Feld 1981) that provides an opportunity to build social connections among participants. For example, for organizations such as police forces and military units, training is intentionally designed to develop strong ties among participants (Van Maanen 1975). Likewise, corporations have recently emphasized teamwork and building strong, trusting relationships among team members in intensive training programs where executives spend prolonged periods of time in survival settings. Deep and lasting relationships can develop as people go through the rites of transition together (Trice and Morand 1989). For example, relationships established as cohorts proceed through college together can provide social capital in later life. Even when training experiences are brief, network connections are formed and may develop further. When viewed from this network perspective, training can be effectively used to build connections across diverse, heterogeneous groups in anticipation of the future formation of cross-functional teams (Krackhardt and Hanson 1993). In the context of training,


connections may be based on similar training experiences, rather than actor similarity on attributes such as race, sex, or age.

Performance The human resources management literature has focused on methods of increasing the accuracy and reliability of appraisal of individual performance. However, performance evaluations may also depend on actual and perceived similarity in attitudes, values, and demographics between the supervisor (evaluator) and the subordinate (See Ferris and Judge 1991 for a review). While actor similarity may positively bias performance evaluations, negative relationships may result in lower evaluations. Due to negative asymmetry, negative events may be more easily recalled, perceived as more diagnostic, and receive greater weights in evaluations than positive events (Taylor 1991).

The social capital perspective on performance invites us to analyze the pattern of relationships rather than view individuals' performance in isolation. As is the case with interdependent tasks in organizations, relationships with others affect performance (Brass 1981; Roberts and O'Reilly 1979). For example, Brass (1985b) found that performance varied according to combinations of technological uncertainty, job characteristics, and interaction patterns. The results are consistent with small group laboratory network studies of the early 1950's (see Shaw 1964 for a review). These studies consistently found that performance was better when the communication network matched the information processing requirements of the task. Centralized communication networks resulted in more efficient performance when tasks were simple, and decentralized networks were better for performing complex, uncertain tasks.

As Lazega (this volume) concludes, social structure can result in social capital and increased performance. Lazega found that the existence of a multiplex, generalized exchange system of sharing information resulted in better performance in a law firm. In addition, Uzzi (1997a) found that the social capital inherent in a system of strong ties can provide economic benefits for small firms in the garment industry (cf. Uzzi, this volume).

Career Development From a social capital perspective, getting ahead in organizations may be a function of 'who you know, not just what you know.' This perspective is contrary to the individualistic values that drive much of the research in human resources management: achievement and rewards should be contingent on individual effort and abilities (human capital). Yet, most managers' careers are contingent on what they can effectively accomplish in connection' with others. The myth of managerial work is that it occurs in isolation (Mintzberg 1973). Most of a manager's roles involve social relationships.

The popular press has noted the importance of 'networking' as well as the advantages of having mentors in organizations. The advantages of building a large network have been extolled and seldom questioned, although little systematic research has actually addressed this prescription. But large networks require a large amount of time and effort in maintaining relationships. Rather than simply building


relationships randomly, some strategy may be useful. For example, should actors develop close personal relationships with mentors or highly connected others (i.e., a strong tie strategy) or attempt to develop many weaker relationships with disconnected others (i.e., a weak tie strategy)?

Assuming a limit to the number of direct links that an employee can maintain, being linked to central others is more efficient than being linked to peripheral others. This strategy allows an employee to be central by virtue of a few direct links to others who have many direct links. The employee has access to social capital (resources such as information) via the indirect links of the highly-connected other, such as a mentor. However, the reliance on indirect links creates a dependency on the highly- connected other (mentor) to mediate the flow of resources. Thus it is important to form a strong, trusting tie to the highly connected other. We refer to this as a strong tie strategy.

Alternatively, an employee may choose a strategy of creating weak ties to diverse groups. Burt (1992) has extolled the advantages of linking groups that are not themselves linked-taking advantage of 'structural holes.' A structural hole is defined as the absence of link between two others who are both linked to an actor. Because the others are not themselves linked, the actor gains non-redundant information from the others (Le., the strength of weak ties argument), and is in a position to control the information flow between the two (Le., broker the relationship), or play the two off against each other.

Burt (1992) found that structural holes were associated with early promotions for a sample of managers in a high-technology firm, except in the case of women and newly-hired managers. For women, connections to highly-central others (the strong tie strategy) worked best. However, because the network data were not longitudinal, it is difficult to assess whether the networks were the result of early promotions or the cause of early promotions. Kilduff and Krackhardt (1994) found that strong ties also carry advantages. They found that a friendship link to a prominent person in an organization tended to boost an individual's performance reputation. In addition, they found that the perceived network, rather than the actual network, statistically significantly predicted reputations. Thus, it appears that individuals may improve their reputations by 'basking in the reflected glory' of prominent others. Similarly, Brass (1984, 1985a) found that connections to supervisors and the dominant coalition in an organization were related to reputational measures of power and actual promotions over a three-year time period. To the extent that acquiring power and influence is related to upward mobility and success, much of the previous discussion of power applies.

Connections to powerful others may result in 'basking in the reflected glory,' but they may also result in being perceived as 'second fiddle.' In the latter case, one's own talents are diminished in the presence of a powerful other (i.e., one is perceived as 'riding the coattails' or as a 'side kick'). These different perceptions may be contingent on the stage of one's career, boundaries to entry, and/or the type of organization. Early in one's career, strong connections to a powerful other may be perceived as an indication of potential success. However, later in one's career, one is expected to successfully perform on one's own, and to mentor others. Continued


attachment to a powerful other may not be beneficial to one's career (but see Higgins and Nohria, this volume, for contrary evidence).

As the previous discussion of actor similarity suggests, women, minorities, or newly hired managers may face barriers to entry in established networks, and may be excluded from the social capital of such networks. A strong connection to a powerful, well-connected mentor may overcome such barriers, as Burt's (1992) analyses suggest.

Avoiding or resolving negative relationships may be equally, if not more important to career development. As in the case of selection, negative asymmetry may overwhelm any of the social capital of either strong or weak tie strategies. It is likely that an actor's negative ties within an organization will prevent promotion, particularly if those negative relationships are with influential others. Others may withhold critical information that worsens an actor's performance or they may provide bad references in order to prevent a promotion. Employees adopting a strong-tie, mentor strategy must also be aware of any negative relationships between the mentor and others. Being perceived as the 'side-kick' of a highly-disliked other may be detrimental to career success. Conversely, forming the many diverse connections involved in a weak-tie strategy may increase the possibility of negative relationships. Avoiding negative relationships may be particularly important for women and minorities in organizations. Any evidence of negative relationships may confirm negative stereotypes and quickly interrupt career development.

Turnover Krackhardt and Porter (1985, 1986) found that turnover was a function of an actor's position in the social network. Turnover did not occur randomly, but in structurally equivalent clusters in the perceived interpersonal communication network. In this longitudinal study, the closer the employee was to those who left, the more satisfied and committed the remaining employee became, cognitively justifying their own decision to stay.

When turnover occurs in large numbers, such as layoffs connected with downsizing in organizations, social networks, and any of the benefits of social capital, may be disrupted or destroyed. Shah (1996) examined survivor's networks and reactions following downsizing in a firm where 42% of the workers were discharged. Although the advice network was restored in six months, the friendship network remained depleted. Survivors responded negatively to the loss of the social capital provided by friendships, but responded positively to the promotional opportunities provided by vacancies due to the layoffs of structurally equivalent others.

Actor similarity may also affect turnover. Research has shown that similarity in age and tenure among group members is negatively related to turnover (Tsui et al. 1992; Tsui and O'Reilly 1989; Wagner et al. 1984; Zenger and Lawrence 1989). As McPherson, Popielarz, and Drobnic (1992) found, similarity leads to increased communication which, in tum, is negatively related to turnover in voluntary organizations. They found that network ties within a group were associated with reduced turnover, while ties outside the group (weak ties) increased turnover.


These findings suggest that employees who share in social capital may be less likely to quit. We can further speculate that negative relationships may be a primary cause of turnover, especially when these social liabilities involve supervisors.

Conflict In a study of intergroup networks in twenty organizations, Nelson (1989) argued that the interaction networks were significantly different for high and low conflict organizations. He found that low-conflict organizations were characterized by a high number of strong ties (measured as frequency of communication) between members of different groups. Similarly, Krackhardt and Stern (1988) found that during a crisis, strong ties across groups (friendships) provided the links necessary for effective coordination in a simulated organization.

However, Labianca, Brass, and Gray (1997) found that strong friendship ties between members of different groups in an organization had no statistically significant effect on perceptions of intergroup conflict. Rather, they found that negative interpersonal relationships between members of different groups strongly predicted perceptions of intergroup conflict. The findings illustrated the effects of negative asymmetry; strong positive relationships did not dampen, or counterbalance the effects of negative relationships. They also found evidence of indirect effects. When an actor's friend had a negative relationship with a member of another group, the actor perceived increased conflict between groups.

CONCLUSIONS

Much of the progress in human resources management research and application has been achieved via the traditional emphasis on the identification and measurement of individual attributes. Yet, it has been estimated that the average adult maintains more than 1000 informal ties (mutually recognizable others). Research on the 'small world' phenomenon (Travers and Milgram 1969) has shown that two randomly selected people can 'reach' each other through a path of a surprising few number of links. We are a network of social interdependencies. In attempting to move toward a focus on 'social resources management,' we have outlined some of the important antecedents of social networks, and tried to show how social networks may affect such human resource practices as recruitment, selection, training, socialization, performance, careers, and turnover.

We have also attempted to expand the focus on social capital by resurrecting attention to the negative relationships noted by earlier social network theorists such as Homans (1961) and White (1961). It appears that the negative asymmetry of negative relationship may destroy the possible benefits of social structure for individuals. These social liabilities may also deter the system-wide benefits of social capital. Negative relationships within a system may destroy the trust necessary for common expectations and obligations, and the closure necessary for norms and sanctions. In addition, negative relationships may prevent information sharing and even lead to intentional distortions of information in a system. Negative relationships cannot always be avoided in organizations where workflow and hierarchy require interactions. Thus, it is important that we expand our research on


social capital to include negative relationships and a consideration of the entire social ledger.

What does the future hold for organizations and 'social' resources management? Many see increased acceleration of change in the environment, increased uncertainty, and increased information processing requirements. They have suggested the emergence of a new organizational form-the 'network' organization (Miles and Snow 1986; Baker 1992a; Nohria and Eccles 1992). Managing social resources in this new network organization may be as important as managing human resources. Managers will likely be involved in identifying, locating, and organizing employees across organizational and international boundaries. Social resource managers will likely become human capital 'brokers,' bringing together the right mix of people to successfully offer a product or service. And that mix may be used only temporarily as environments and technologies rapidly change. Social resource management will require identifying and nurturing potential relationships that may change quickly. Given the diversity of people and perspectives, managing social liabilities will be particularly important. Conflict management skills may help prevent or resolve negative relationships. Overall, the network organization will place additional importance on relationships, and the social structure needed for social capital. As one executive said, 'Our business is one of relationships' (Business Week 1986: 64).

• SECTION IV

Structure at the Firm Level social capital in collaboration and alliances

The Triangle: Roles of the Go-Between

18 •

Bart Nooteboom

ABSTRACT

Social capital is seen here as part of the overall resource base of a frrm. One part of social capital is positional advantage in a network. It is established on the basis of relational competence. Third parties can help in the development of social capital by offering their relational competence, in playing one or more of six roles: the roles of the go-between. Transaction cost economics recognizes that in an inter-firm relation the inclusion of a third party can economize on the setting up and operation of a governance mechanism ('trilateral governance'). The third party acts as a go-between in monitoring and controlling compliance to agreements, thus eliminating the need of intricate and costly forms of 'bilateral governance.' A second role of the go-between is to serve as a repository of hostages. A third role is to help in the judgement of the value that partners have for each other, i.e. to solve the 'revelation problem.' A fourth is to serve as a filter against spill-over. A fifth is to mediate in the building of trust. A sixth is to act as a boundary spanner: to offer a link between an established network and outside sources of innovation, while maintaining the integrity of the network. These roles are especially important in relations that are aimed at innovation. By performing these roles, the third party also increases the flexibility of networks of frrms. In sum, third parties may form an important part of the social capital that supports networks of frrms. The analysis opens opportunities, or new perspectives for fulfilment of their roles, to governmental agencies, such as innovation transfer agencies, municipalities or provinces, and market agencies, such as banks, and suggests that a new market is opening up for professional go-betweens.


INTRODUCTION

In studies of inter-organizational relations (lOR's), there is a tendency to look at dyads of flrms, and to consider networks as aggregates of such dyads. But there are several roles for a third party; a go-between. This chapter looks at a go-between not in the sense of a middleman who intermediates in existing production or trade, such as an agent, wholesaler, retailer, and not in the sense of an entrepreneur who intermediates in the realization of new potential in connecting supply and demand. It looks at a gobetween in the sense of a relationship counsellor for the development and maintenance of social capital; providing support in setting up, adapting and ending cooperative relations between others. Or, in yet different terms: to help in the embedding of relations, in Granovetter's (1985) sense (Uzzi 1997a). Such roles may be performed by middlemen or entrepreneurs, but also by specialized agents who do not playa direct role in linking stages in a chain of production and distribution, as middlemen and entrepreneurs typically do. Indeed, some of the roles require an independence that is served by not having a direct stake in the relations that need to be developed.

I propose that the analysis yields a perspective for looking at the roles of trade and industry associations in European business systems, and of banks and trading houses in Japanese enterprise groups (Kigyo Shudan). Some of the roles are played by the Innovation Centres that were instituted in the Netherlands to act as intermediaries in the transfer of technology to small ftrms, without actually serving as suppliers of technology. More importantly, perhaps, I propose that there is an emerging market for the specialized services of the go-between. Some of the most important roles of the gobetween are associated with problems of information, knowledge and learning. These are particularly important in lOR's that are aimed at innovation; at the development of products, processes and competencies.

One theoretical perspective employed in this chapter is transaction cost economics (TCE). I grant that TCE lacks a perspective on innovation and learning and fails to deal with trust next to opportunism, and more generally fails to deal with embedded ness of transactions. But in my view a more encompassing theory can be developed which retains what is valid and useful in TCE and incorporates notions from social exchange theory. I have attempted to contribute to such development (Berger et al. 1995; Nooteboom 1992, 1996; Nooteboom et al. 1997). Especially the 'resource' or 'competence' view in present non-mainstream economics (Foss 1993; Foss and Knudsen 1996) is well suited for such integration of perspectives. In the spirit of this endeavor I will try to position the notion of social capital in the framework of the resource perspective.

TCE retains the assumption from mainstream economics that technology is accessible to all, and this conflicts with the competence view that competencies are embodied in people, teams, procedures, organization and organizational culture, need to be built up, and enable as well as restrict the cognitive and technical repertoire of the flrm. TCE also does not incorporate issues of spill-over. When competence is associated with tacit knowledge and is embedded in the flrm, risk of spill-over may be limited. Nevertheless such risk often remains to a greater or lesser extent, and spillover control is another aspect of the governance of lOR's that must be taken into account.

The Triangle: Roles of the Go-Between - 343

One might say that by introducing the competence perspective, innovationllearning, spill-over, trust and the social embedding of transaction relations TCE is transformed beyond recognition, and that its core assumptions are cut out, so that it is misleading to call it transaction cost theory. But the point is that although trust is made part of the theory, opportunism also remains part of the theory, next to trust: trustworthiness restricts opportunism, but opportunism also puts a limit to trustworthiness. Even the most loyal partners may succumb to pressures of survival or golden opportunities. Repeated cheating quite rightly leads to a loss of trust (see also Uzzi 1997a). The core idea from TCE that dedicated investments, combined with uncertainty concerning environmental conditions (market and technology) create dependence and a potential hazard of hold-up, is retained. When it works, trust can serve as a substitute for legal contracting, and thereby yields a form of governance with lower costs, greater speed, more flexibility, less of a straightjacket for innovation and a better basis for exchange of ideas and learning. But arms-length legal contracting still is an instrument of governance that may fit the conditions and the type of exchange. Other instruments from TCE also retain their potential use: relational contracting on the basis of balanced ownership of dedicated investments, exchange of hostages or mutual dependence; trilateral governance. Indeed: the notion of trilateral governance yields the frrst role for the go-between.

SOCIAL CAPITAL AS A RESOURCE

Social capital is defined as the resources inherent in social structural arrangements, facilitating the attainment of goals (Coleman 1990; the chapters of Gabbay and Leenders in this volume). I would like to elaborate on this notion, and to embed it in the wider notion of 'resources' available to frrms.

The 'resource' or 'competence' view of the frrm (Prahalad and Hamel 1990; Foss 1993; Foss and Knudsen 1996) goes back to the work of Penrose (1959). According to this view the main cause of a frrm's profit is its unique competencies, embedded in the firm. This view complements the 'positioning' view from Industrial Organization that a frrm's profit is due to market conditions: market concentration, oligopoly, collusion, entry barriers (Porter 1985). The two views meet in their joint recognition of the importance of product differentiation to evade pure price competition, and the need to protect one's differentiated product by basing it on competencies that are at least temporarily unique and inimitable. The competence view implies the rejection of any notion of the 'representative frrm': the essence of markets is precisely the heterogeneity of frrms; frrms seek to be different, as the main source of profit. Competencies are cumulative and path dependent, and that is why they cannot always be identically copied (cf. Nooteboom 1992). In particular, the focus of the present chapter is on the development of competencies: on learning and innovation, and the role thereby of lOR's.

Following Stoelhorst (1997), I take resources as a general concept that can be classified into assets, competencies and positional advantages. Assets are characterized by the fact that they are subject to legal ownership and contracts. Competencies and positional advantages are not easily subject to property rights. Resources cannot all be instantly copied by other frrms because they are to some extent inscrutable or subject to 'causal ambiguity' (Lippman and Rumelt 1982): even if a would-be imitator can


observe activities, that does not yet imply that he can understand them, let alone implement them. Resources, and especially competencies, can be difficult to understand and imitate, because the knowledge involved is to a greater or lesser extent tacit (not documented) and embodied in the heads and hands of people, and embedded in teams, organizational structure and procedures, and organizational culture.

Competencies can reside on the personal level, in the form of knowledge, skill and relational competence, but related to the latter I would also include motivation and morality. 'Morality' includes norms and values of conduct that the individual holds, his degree of commitment to them and susceptibility to ethical appeals (concerning loyalty, justice, truthfulness). On the aggregate, interpersonal level of 'communities of practice' (Brown and Duguid 1991) within an organization, entire organizations and even networks of organizations there are assets and positional advantages, but also competencies. Competencies on the level of an organization or network would include institutions and patterns of knowledge exchange and transformation. Institutions are defined as environments and arrangements which limit and guide conduct (North 1990; North and Thomas 1973). They include practices, procedures, rules, technical standards as well as cultural entities such as prevailing norms and values of conduct, goals, role models, rituals. Among other things, they may serve to guide relations with other organizations, and are then part of fmn-Ievel relational competence. Organizational competencies in the form of 'patterns of knowledge exchange and conversion' refer to the way in which knowledge is converted from tacit to documented knowledge, absorbed from documented into tacit knowledge, transmitted, pooled, shared and recombined in novel combinations (cf. Nonaka and Takeuchi 1995). Relational competencies on the individual level enhance learning. Relational competencies on the individual level, supported and guided by relational competencies on the organizational level in the form of guiding institutions, yield positional advantage in the form of efficient access to resources of other organizations. Positional advantages further include product-technology-market combinations, access to materials, distribution channels, political acceptance, brand loyalty and reputation.

In this framework, social capital, according to the wide definition of 'social structural arrangements which facilitate the attainment of goals,' would include positional advantages achieved in networks, as well as structures of communication, knowledge conversion and learning within the fmn. It this framework it is seen as the product of relational competencies on both the individual and the fmn level. The central point of this chapter now is that third parties can contribute relational competence in the building of social capital in the form of positional advantages.

SYSTEMS OF COMPETENCE

From the perspectives of competence and dynamics, a basic hypothesis underlying this chapter is that under present conditions fmns need other fmns in order to cope with the need and opportunity to innovate and differentiate products. The need follows from increased competition, due to globalization, and the opportunity follows from increased prosperity, yielding more differentiated needs and wishes of consumers, technologies of flexible production and information- and communication technology (lCT). But this differentiation, innovation and technical development yield great complexity and variability of opportunities and threats. For a single firm, a full grasp


of the detailed and fast changing stock of knowledge concerning market and technological opportunities and full command of all relevant competencies is not 'sustainable,' as Zuscovitch (1994) put it. One needs partner fIrms that are close to specifIc markets, and partner fInns close to specifIc sources, who specialize in the required knowledge and make it profItable by sharing it with partners, in networks of partial cooperation.

Elsewhere, this has been called 'cross-fInn economy of learning' and 'external economy of cognitive scope' (Nooteboom 1992). In a constructivist theory of knowledge, knowledge, including perception, understanding and evaluation, is based on categories that have been fonned in interaction with the physical and social environment. Having different categories, different people are able to see, interpret and evaluate different things. The underlying categories make knowledge tacit to a greater or lesser extent, and the fact that categories are fonned in past experience makes knowledge cumulative and path-dependent. While such theories of cognition apply to individuals rather than fIrms, fIrms have similar properties because of the alignment of perception, interpretation and evaluation between people in the fIrm, established in the fIrm's procedures and its culture. The implications for transaction costs theory, going beyond 'classical' TCE (Williamson 1985), are as follows. First, an additional reason for cooperating with others is that one lacks the requisite categories, and cannot simply buy and install them. You cannot buy understanding, for example. Second, to perceive and understand opportunities and threats, one may need a diversity of external sources with suffIcient variety of categories to grasp them, to the extent that conditions are complex and changing.

This cognitive view stands in contrast with the view of Teece (1986, 1988) and Chesbrough and Teece (1996) that, particularly in the case of 'systemic' innovations, integration into a single fIrm is required to deal with connections between elements of the system, to control spill-over and to ensure appropriability of profIts from the innovation. That perspective does not deal with cognitive issues of understanding and learning. I have argued that there is an argument for integration only under the following limited conditions: the innovation is indeed systemic, innovation is not radical but incremental, and it has not yet progressed so far as to generate standards across the interfaces between elements of the system (Nooteboom 1999). In all other conditions, but particularly when innovation is radical, a disintegrated structure is better, to ensure suffIcient flexibility of novel combinations and suffIcient variety of elements to make novel combinations of.

The view of fIrms as not only rivals, but also potential collaborators in development, distinguishes European and Japanese enterprise systems from AngloAmerican ones. The fonner have been called 'voice' systems, with corporate social capital in the fonn of networks or other groups of frrms, versus the Anglo-American 'exit' system, with more autonomous and more integrated frrms (Gelauff en den Broeder 1996; Groenewegen 1997; Nooteboom 1997). The characteristics of the two types of capitalist systems are summarized in Table 1 (adapted from the comparison between the Anglo-American and the German system proposed by Gelauff and den Broeder 1996).


Table 1. Exit and voice systems (adapted from Gelauff en den BrtJeder 1996)

General characteristics

Culture/institutions

Corporate governance

Important shareholders Corporate control

Shares managers Creditor control Regulation

Contractual governance

Relations Contracts Contract execution Organization of work organizational form labour market contracts motivation labour participation

TRUST

Exit system

market orientation short-term relations rivalry autonomy of fIrms individualistic legalistic

individuals exit (sell shares) take-over many withdraw loan supports markets banks not owners (US) no cross-participation

market, arm's length formal courts

vertical integration competitive formal wage, profIt sharing low

Voice system

network orientation long-term relations cooperation embeddedness of fIrms group oriented community ethics

fIrms, banks voice mutual interest few co-owner, monitoring obstacles to trade in shares banks co-owners cross-participation

networks, close relational reputation

networks protection relational wage, job security worker councils

There is an important difference between competence trust and intentional trust. The first pertains to someone's ability to perform, and the second to his intentions to do so. Their breach has different implications for action. When trust in ability is breached an appropriate response is to help to improve the partner's performance. When trust in intentions is breached repeatedly, the response is more likely to be retaliation or the breaking of the relation. But it is not easy to determine which of the two is the case. When the breach is intentional, one can claim force majeure, i.e. a freak break of competence, to prevent retaliation. Most discussions in the literature focus on intentional trust, because that is the most difficult to deal with.

Intentional trust is a slippery notion, and this is the cause of dangerous misunderstandings. A key question is whether trust is based on coercion and self-interest or goes beyond it. A wide definition of trust is that one expects that no harm will occur. But this may be due to the fact that contractually the partner has no opportunity to do harm, or that it is not in his material interest to do so. According to a much narrower definition, trust means that one expects that the partner will do no harm even though


he has both the opportunity and the incentive to do so (Nooteboom 1996, 1999b). If someone says that you can trust him, which is meant? If people say they trust each other and one means the one and the other means the other, unpleasant surprises are likely to occur.

I agree with Williamson (1993) that if the notion of (intentional) trust is to yield any added value in theory, it must go beyond the calculative self-interest that is already covered in established economic theory. For instance, it must go beyond fear of punishment by the law. It must go beyond loyalty based on the fact that it is in one's material interest to maintain the relation, taking into account the 'shadow of the future': the discounted value of future rewards in an ongoing relationship that one would obtain by remaining loyal. It would need to go beyond reputation if that is based on calculation of potential rewards in future relationships with others than the present partner. I also agree with Williamson that trust should not be and rarely is blind. But I disagree with Williamson that if trust is not to be blind it must be calculative. Trust can be based on routine, or habituation: after a while, if things go well, actions and procedures may be taken for granted; behavioral risks may no longer be part of 'focal knowledge' (Michael Polanyi), as long as conditions and observed actions stay within some range of tolerance. Only unexpected events trigger awareness and attention to possible opportunism. As explained by Herbert Simon, such routinization is rational, given bounded ness of rationality. When a certain pattern of action goes well, it is conducive to survival to routinize that behaviour, and focus limited capacity of rationality on new challenges. Thus routine can be non-calculative and yet have a rational basis, in the sense of being conducive to survival, and in that sense it is not blind. Uzzi (1997a) made the interesting proposal of using the notion of a 'heuristic' in this context: some quick, efficient method of assessment that is neither calculative nor blind. I suspect that it comes close to my notion of routine: the heuristic has arisen from its survival value in the past.

Furthermore, trust can go beyond self-interest, in the realm of decency, loyalty, solidarity or self-respect. Here we are dealing with trust in the narrow sense: the expectation that the partner will not utilize opportunities for opportunism, even though it is in his material interest to do so (Nooteboom 1996, 1999b). Trustworthiness in that sense can be based on an intrinsic valuation of reputation, i.e. not as an instrument of utility, or on self-respect, or on moral or social obligation. These can be related to friend- or kinship. When related to membership of a social group it is in place prior to the transaction. When it is not already present, it has to be built up in a relationship. Habituation then plays a double role: in developing shared values and norms, and in the sheer decay of awareness of what might go wrong, when things have consistently been going right (Nooteboom 1996; Nooteboom et al. 1997).

But I accept that trust should not be blind, because trustworthiness has its limits: it may be breached by a 'golden opportunity,' and some people are more resistant to such temptation than others. Here I retain some of the perspective of TCE. But the conclusion of the analysis is that no governance is needed when the advantage of defection, or opportunism, does not exceed the resistance to temptation. And here I go beyond TCE.

Trust between organizations can be based on inter-personal relations between staff of those fIrms (see Pennings and Lee, this volume). In the resource view set up before:


relational competencies and competencies of motivation and morality. Motivation may be based on personal friendship, kinship, empathy, habituation, shared norms and values. Firm-level relational competencies can be found in institutions (rules, habits, role models) that guide contact with other firms. Inter-organizational trust may further be based on reputation and on survival conditions. When an organization is under great pressure to survive, its loyalty may come under stress.

TRILA TERAL GOVERNANCE

Already in classical TCE the possibility was indicated of using a third party, in 'trilateral governance,' as an alternative to integration of the activity within the fIrm or bilateral governance between fIrms, for solving the hold-up problem that arises in the case of dedicated investments. This is recommended when the transactions involved are infrequent, so that the volume of transactions does not warrant the cost of setting up bilateral governance, while there are significant advantages of not integrating the activity within the user firm. Thus the argument for trilateral governance is one of efficiency: it is cheaper. Only limited explicit agreements between the partners are made, and the most important one is procedural: If disagreement arises, the third party will be called on to arbitrate. An important part of this role is to help the partners to set realistic goals with balanced advantages and risks. A classic example is the role of an architect as third party in transactions between a builder and a supplier of building materials.

Clearly, a condition is that the third party must be trusted by both partners, in both the competence and the intentional sense. He must be knowledgeable on the technologies, markets and strategies involved, and he must be fair in judgement and adjudication. The basis for this trust may be self-interest: the third party has an interest in ongoing relations with both partners, and therefore has an interest in doing his best and being fair, in order to maintain his reputation. The basis may also be more intrinsic, stemming from desire, self-esteem or moral or social obligation. The question then arises: if there is such a basis for trust in the go-between, why does it not also arise between the partners themselves, so that the go-between is superfluous? The answer is that between the partners there is the temptation to defect, because it yields advantage, while for the go-between there is no such temptation. But note that the gobetween is not needed if between the partners the advantage of defection does not exceed their resistance to temptation.

HOSTAGE KEEPING

Another instrument of governance, recognized by TCE, is the exchange of hostages, in order to guarantee that agreements are kept. The characteristic of a hostage is its asymmetry in value: it has value for the giver but not for the taker. This condition serves to keep the taker from the temptation of keeping the hostage permanently, even if the other side has kept his end of the bargain.

In relations between fIrms, people could be hostages: in intermarriage between family businesses or the detachment to each other of key executives. It can also take the form of cross-participation between fIrms, or pieces of technology or knowledge that are sensitive to competition. The use of cross-participation as hostage taking is


admitted by the Japanese Agency for Economic Planning in a document in 1992 (Scher 1996: 17).

In hostage taking there is a well-known riddle. Suppose a king must supply a hostage to another king with whom he is concluding a treaty, and must choose between a charming, attractive daughter and a nasty, unattractive one, whom he both loves dearly. The standard answer is: the latter, because that better satisfies the condition of asymmetric value. The more attractive daughter could more easily assume value for the hostage taker, and thereby tempt him never to return her. But what if the first king does not really intend to honour the treaty? Then it would be better to give the attractive daughter, because then he has the greatest chance that the hostage taker will not sacrifice her, in spite of the breaking of the treaty. But the second king might see through this, and become suspicious when the attractive daughter is offered. But the first king might trump this by signalling that he does not really love his unattractive daughter, and the second king might therefore demand the attractive daughter, because the unattractive one would not be an effective hostage, and thereby he would fall into the trap of the first king.

Such problems can be reduced by having a third party keep the hostages. This solution is antique; it was used in treaties between medieval rulers (De Laat 1996). It works only if the third party is less tempted to retain the hostages after completion of the agreement than the party in whose interest the hostage is taken. But note that the third party would jeopardize his relation with both partners by not keeping the agreement, and is thus likely to be disciplined by this double jeopardy. There would no longer be the temptation to play the game discussed above. If the hostage giver supplied his attractive daughter with the aim of breaking the treaty, and then did break the treaty, his partner in the treaty would demand from the go-between to sacrifice her. The ploy would no longer work.

REVELATION

Particularly when frrms cooperate for innovation, the knowledge they are pooling will tend to be tacit, and to exchange tacit knowledge one must set up intensive interaction, in which mutual understanding must be built up. Note also that the setting up of such interaction and mutual understanding constitutes a dedicated investment. How do you judge whether such investment is worth while before you commit it? This is the revelation problem: how can value of knowledge or competence be revealed before it is transmitted? Here we are dealing with Arrow's paradox: if you give information that is sufficient to judge the value of the information, then little may be left to trade or hold back. Here lies a third role for the go-between. Having ongoing relations with both partners, it has already made investments in ability to understand and judge competence. Thus it can act as judge of value and relevance for both partners before they invest in their relationship. Having the trust of both partners, and knowing them well enough to assess both on the value and relevance to each other, it can inform them on that without actually supplying the underlying information.


SPILL· OVER CONTROL

Closely related to this is the role of the go-between in spill-over control. Partners do not only want to have an indication of value before they commit dedicated investments. They also face the problem that by giving information about their knowledge they reveal at least part of that knowledge, and that may spill over to competitors. The fourth role of the go-between is to serve as a screening device: the go-between does not make the knowledge available to the partner, but assesses it for him. Next, when the partners proceed to pool and exchange knowledge the go-between can perform or monitor spill-over control. Spill-over control might otherwise require mutual in-house monitoring of streams of information by the partners themselves, to check that it does not leak to competitors, but such mutual monitoring may increase rather than control information exchange, and hence aggravate rather than relieve the spill-over problem. Especially at the beginning of the relationship such in-house monitoring is likely to be unacceptable to the partners. Both the risk of spill-over and the unwillingness to grant monitoring to guard against it will be greater to the extent that the partners are actual or potential competitors, or have intimate linkages with competitors.

Note that the whole issue of spill-over is relevant only under certain conditions. The first is that the nature of the knowledge involved indeed enables spill-over. This is not the case to the extent that the knowledge is tacit or embedded in the organization. When knowledge is tacit, its transfer requires intensive mutual interaction, or a buyout of the people in whom the knowledge is embodied, and even that may not be effective, to the extent that the performance of those people is contingent upon supporting knowledge, procedures or organization that stays behind. The second condition for the problem of spill-over is that the partner is himself a potential competitor, or has connections with competitors. The third is that development is not so fast that spill-over no longer matters. If change is so fast that by the time a competitor has imitated it, the knowledge or technology has already changed, then the problem of spill-over simply disappears (Nooteboom 1998a).

BUILDING AND MAINTAINING TRUST

Related to the previous roles, there is a fifth role in the building of trust. This was recognized also by Uzzi (1997a), who described the role of a go-between in creating the embedded ness of partnerships. It was also recognized in mechanisms in the transmission of trust discussed by Deutsch (1973). One aspect of this is the transitivity of trust: If X trusts Y and Y trusts Z, then X can be disposed to trust Z. Here Y is the go-between. One often initiates relations with partners who are trusted partners of one's own trusted partners. If X and Z both trust Y, then they may be disposed to also trust each other. Here Y again is the go-between. Such bringing together of partners in a disposition towards trust is only a beginning. Next, intermediation may help to set up initial small steps of cooperation, and to ensure that they are successful and do not raise misunderstanding that might antagonize or raise suspicion. This is important in the building of both competence trust and intentional trust. It was noted before that breach of competence trust calls for very different action from breach of intentional trust. In the first case one might help to improve competence, but that would not help


in the second case, where it might be more appropriate to signal a warning or even retaliate. But it can be difficult to tell which of the two is the case. The third party can act as an independent observer, with access to inside information. He can eliminate misunderstandings that might otherwise destabilize the relation. This role is particularly important where trust is not pre-existent, as part of a group culture or set of norms and values and has to be built up in the relationship.

The beginning of relationships tends to get most attention, but the ending of relationships is at least as important, and often more difficult (Nooteboom 1996). When in an existing relationship one partner runs into a more attractive alternative, there is substantial risk of fierce antagonism, whereby the other partner tries to evade loss by keeping the first partner from cutting loose, and the first partner retaliates with nasty behaviour to badger the second partner to let go. In expectation of this, the fust partner may keep his wish to exit secret, and prepare his defection in silence. But this makes the problem worse for the second partner, because he has less time to adapt, and may react all the more fiercely in trying to tie the fust partner down. One can observe this in both fum relationships and marriages. The better road may be for the fust partner to announce his intentions at an early stage, but stay on for the time being to help the second partner to find an acceptable way out. But such an announcement breaks trust, and the destructive spiral of mutual antagonism is difficult to prevent. Careful counselling by a trusted third party may be needed to guide and control this process.

An alternative is for the fust partner to give the relationship another chance, and help the second partner to increase its attractiveness. But this entails that the second partner further increases its relation-specific investments, while a threat has appeared that the relation may break. Here also, trusted counselling may be needed.

BOUNDARY SPANNING

An argument against embedded, voice based network systems is that the relations between fums, oriented towards the longer term, and based on relation-specific investments and trust, create exit barriers and entry barriers and thereby limit the efficiency, flexibility and adaptiveness of the system. Uzzi (1997a) called this the 'paradox of embedded ness.' This leads to a sixth role for the go-between: to act as a boundary spanner between an existing network and potential outside sources of innovation. It may be threatening for partners who are active in ongoing exchange relations to maintain outside exploratory relations for scanning novel opportunities. In fact, one of the most threatening and potentially destabilizing events in a trust relationship is the appearance on the stage of a new player who might present a more attractive substitute for one of the partners involved. The fear of this may create a taboo of outside scanning in order to maintain the integrity of the network, and this can be disastrous for innovation on the basis of outside sources. The problem of losing positional advantage to an outsider, and being left with idle dedicated assets, is aggravated by spill-over risk: one may also lose part of one's core competence. A go-between who does not himself participate in exchange can perform outside scanning under less suspicion, since it is not in his own direct interest to defect to more promising outsiders, while it is in his, interest to maintain his reputation for confidentiality and


Table 2. Fonns of governance

exit system

Elements of governance specific investments low switching costs low tailored value partner low contractual restrictions high opportunism high

Results product costs low transaction costs high productdUferentiation low incremental innovation low creative destruction high

voice system

high high high low low

high low high high constrained

fair dealing. Summing up: outside scanning is less threatening when done by a gobetween without a direct stake in exchange, and this opens up the network to potential novelty from outside.

In fact, this role is related to some of the previous ones: the role of revelation, spill-over control and the management of trust. One can bring in judgements of the potential of outside novelty while maintaining confidentiality of its content and of its source, without which one would not have obtained that information. And conversely: one can give an assessment of the value of what the network could offer to outsiders without giving it away and without specifying the precise source in the network (which may tempt the outsider to pry that loose from the network). When novel opportunities for redesigning the network have thus been identified and assessed, the go-between can next help in the phasing out of the 'losers' in the old network, by providing ways out, in exit to another network, or helping them to break up and redistribute their resources. The go-between might even administer some insurance developed by the network to deal with such contingencies.

INCREASING FLEXmILITY

When translated to specific aspects and outcomes of governance of inter-firm relations (Nooteboom 1996), the differences between the exit based and voice based systems of Table 1 are as summarized in Table 2.

As already indicated, the main argument against the voice based network systems of capitalism is that the relations between firms, oriented towards the longer term, and based on relation-specific investments and trust, create exit barriers and limit the flexibility of the system, and thereby obstructs especially radical innovation. However, it should be noted that in the Anglo-American system due to lack of network structures between firms activities are more integrated within larger firms. That system must derive its flexibility from easy break-up of firms, which also has disadvantages: shorter contracts and assignments of labour, with less job security, yielding less willingness to engage in firm-specific learning, less scope for cooperation in teams,


and a short term orientation which hinders innovation. However that may be, it cannot be denied that the flexibility of voice based systems could be constrained. It then is important to make relations as flexible as possible, while still being consistent with the orientation towards mutual advantage and cooperation.

To make the voice-based system more flexible, relations between frrms should be long enough to ensure pay-back of specific investments, but short enough to allow for sufficient flexibility of re-aligning relations. When trust is not already present, and has to be built up within relations, this also takes time and creates a danger of rigidity. Setup and maintenance costs of governance should be low enough to facilitate mUltiple and flexible relations. Spill-over control should not constrain the formation of multiple relations too much. The third party can help here, to offer trilateral governance, keeping of hostages, revelation, spill-over control, the building of trust and boundary spanning. In other words: The third party can help to make the system as flexible as possible. Thus the role of the go-between is part of the social capital of voice based systems, and helps to make it flexible.

EVIDENCE

I propose that the roles of the go-between explain the important role of banks and trade and craft associations in European systems, and of banks and trading companies in Japanese enterprise systems (Kigyo Shudan, cf. Scher 1996).

There is evidence that in the Netherlands the recently instituted 'Innovation Centres' play such roles. Not entirely consciously, and largely by trial and error. This evidence comes from my own experience in giving lectures along the lines of this article to the centres, and informally testing to what extent they play the roles indicated, by asking them to give specific instances. The role of trilateral governance was recognized clearly: the coaching of partners in lieu of contracts; particularly the setting of realistic goals, monitoring their achievement, eliminating misunderstandings and making adjustments along the way. This role was clearly integrated with the role of building and managing trust, as one might expect. The role of revelation was also clearly recognized: here lies the central mission of the centres. Hostage keeping and spill-over control were not recognized directly, but part of the experience was that the care and confidentiality with which information was treated was enhanced by the fact that the centre acted as a go-between. This could be interpreted as a hostage mechanism: players were careful with information received from their partner because sensitive information from themselves was held by the go-between. It could also be interpreted as a reputation mechanism: they want to maintain a good reputation with the centre, not to jeopardize future dealings and potential future partnerships. The role of the boundary spanner was not relevant here because the networks in which the centres operated had not yet reached the degree of consolidation in which such a role becomes relevant.

Evidence is also found in a longitudinal case study by Klein Woolthuis (1996) of cooperation between eleven frrms, in various configurations, in the development of medical products, based on an initiative by an Innovation Centre and a Regional Development Centre (ROC) in the Dutch province of Overijsel. Development projects were submitted for subsidy in the EU EFRO programme (when awarded, it yields a 50% subsidy of product development costs). The Centre and the RDC were


particularly important in the start-up stage. An important factor for the firms was that they trusted the initiators, and hence also the partner firms that they brought together. Yet, 'parties were reluctant to reveal information and were hesitant to engage in any relationship before it was completely clear what the other's motives were and what role they would fulfil... This made the start difficult and time consuming .... Therefore it was important that the Centre and the ROC took on a guiding role ..... they could ease the negotiations .... this included chairmanship at meetings, encouraging and guiding the first contacts.' This gives evidence of the roles of the go-between in trilateral governance and the building of trust.

An external committee was instituted, with one member from the Ministry of Economic Affairs, and two managers from local hospitals, to help in the assessment of project proposals and to watch over fair distribution of work and subsidies from EFRO. It is noted that 'It is very important that the committee should make fair evaluations. Parties should be able to trust the board ... The perception of equal treatment is therefore of crucial importance.' The task of the 'umbrella' of the gobetweens, consisting of the Centre, the ROC and the advisory council, was summarized as follows: 'In the first stage the umbrella performed an important role because it provided firms with a platform to get to know the others. The umbrella became of decreasing importance when direct links between firms evolved .... frrms started to know each other and envisage future potential for more dense relationships.' This again indicates the role of trust building, and also suggests the role of revelation. There is no mention of the roles of hostage keeping and spill-over control. They may have been present but unnoticed by the researchers reporting the case.

It is interesting that the go-betweens who were the initiators (the Centre and the RDC) instituted a separate advisory council to guide the development of projects. This makes sense, for two reasons. First, it was noted that the go-between requires expertise to generate competence trust, and it may be too much to expect relevant and sufficiently deep expertise of a single agency in a wide variety of relations. With a separate council the go-between can seek specialized support that is tailored to the situation. That is why in this case hospitals were included. Second, by creating a special council, the go-between insulates itself to some extent from the risk of losing reputation when something goes wrong in the judgement and guidance of cooperation: then the advisory council takes the direct blame, and since it is unique to the situation, its loss of reputation is less serious.

DISCUSSION

Six roles have been identified for the go-between: trilateral governance, keeping of hostages, revelation, spill-over control, trust building and maintenance, and boundary spanning. All roles except the first are especially important in the building of new relations in innovation. It is especially there that the use of complementary competencies is important, while they are not always visible, and exchange of sensitive information is important. By playing these roles, the go-between can contribute to the flexibility of network systems of frrms. Thus the go-between forms an important ingredient of the social capital on which the networks of the voice-based system of capitalism are based.


However, the roles of the go-between are not easy. They require thorough knowledge of the technologies involved in the firms involved, as well as the commercial interests and strategic interests and risks at stake. A sharp insight is required of the factors which play a role in relationships, instruments of governance and skill to handle them (Nooteboom 1996). Reputation, in the both competence and fairness, is crucial, is difficult to build up and easy to destroy. The demands and risks involved may be too much for any single agency. The go-between can solve this problem by using special committees or councils to guide specific relations, which are tailor-made in their expertise, to generate competence trust, while the go-between itself monitors intentional trust. In this way, the go-between also insulates itself to some extent from the risk of failing to perform all the balancing acts without a flaw: part of the blame is then taken by the committee, that can be modified or replaced while maintaining the position of the go-between himself.

The roles need not all be performed by a single agent, as the above analysis and the preceding case study show. They can be distributed over several agents. One reason for this can be to distribute risks of loss of reputation due to error. As I indicated from the start, the roles are not the traditional roles of middlemen or entrepreneurs, but that does not preclude that some of the roles are in fact played by parties who are directly involved in production and distribution. In any network or team, any participant may help to arbitrate, keep a hostage, reveal competence, control spill-over, build trust and act as a boundary spanner for other participants. Some of this is found in the notion of the 'closure' of social structure (Coleman 1990). But some of the roles require some degree of independence; no direct involvement in exchange between the others.

The analysis indicates that there is an emerging market for go-betweens. The roles may be played by government or semi-government organizations: municipalities, provincial development associations, innovation or technology transfer centres, but also by banks, commercial consultants. Government agencies tend to be weak in generating competence trust but strong in generating intentional trust. Thus combinations of local semi-government agencies coordinating and monitoring professionals from business may yield the optimal form.

•

The Management of Social Capital in R&D Collaboration

ABSTRACT

19 OnnoOmta

Wouter van Rossum

Rapidly developing technologies, increased global competition, and more stringent customer demands are compelling companies to improve the pace and quality of their product and process innovation. As a consequence, companies are increasingly pushed to form technological partnerships to share development costs and to reduce time-to-market. This chapter focuses on the 'dark side of cooperation' by concentrating on the complications and pitfalls encountered by thirty general and R&D managers from ten leading companies in R&D collaborations in the automotive, electronic, energy, heavy, and pharmaceutical industries. The partnering relationships described include strategic R&D alliances, supplier-customer partnerships, and cooperative relationships with knowledge institutions. The empirical research shows that these relationships are extremely vulnerable. For example, the partners may worry about releasing too much confidential information and technology. The occurrence ofthese problems can be explained by social capital theory. Network relations may enhance the social capital of a company by making it feasible for it to get easier access to information, technical know-how, and financial support. But, at the same time, these relationships may lead to social liability, e.g., by reducing the possibilities for relating to companies outside the network, risk of spillover, and high coordination costs of the networkrelations. R&D relationships, especially, are, for the most part, not very tightly knit, and hence their problems relate to lack of information and to opportunism. This chapter then focuses on the methods managers can use to minimize these problems by reducing social liability and hence enhancing social capital. Partnership management is considered to be, to a great extent, management of trust and goes far beyond signing confidentiality agreements and agreeing to follow guidelines. This chapter ends with possible

The Management of Social Capital in R&D Collaboration - 357

solutions for every phase of the collaborative lifecycle. It includes an outsourcing and partnering matrix as a management tool for the strategic foundation of makebuy-or-collaborate decisions.

INTRODUCTION

Rapidly developing technologies, increased global competition, and more stringent customer demands are compelling companies to improve the pace and quality of their product and process innovation. Huge R&D investments may be lost when a final product does not measure up to the required standards or when a similar product of a competing company reaches the market earlier. The pressure to do more with less inexorably pushes companies to focus on few, unique, hard-to-imitate, and distinctive core competencies by continually nurturing and enhancing them, while abandoning activities in which they do not possess distinctive competencies. These R&D-intensive companies, while emphasizing their core competencies, are increasingly seeking new upstream and downstream partners to share development costs, accelerate product and process development and maximize commercialization opportunities. The ability to build and maintain interorganizational network relationships-such as joint ventures, license agreements, supplier-customer partnerships, and strategic alliances-is increasingly viewed as the key factor to sustained competitive advantage. Several authors refer to such R&D networks as virtual corporations in which a number of firms create flexible linkages to attain common or complementary objectives (e.g., Davidow and Malone 1992; Campbell 1996; Upton and McAfee 1996; Yoshino and Srinivasa Rangan 1995).

Innovation within technology networks is extensive because of the sustained interaction between institutions and commercial organizations of different size, capabilities, and expertise. In Silicon Valley for instance, various forums exist for high-tech companies where relationships can be formed and maintained, information exchanged, contacts made, and new ventures formed (Saxenian 1990). Forrest and Martin (1992) note that in biotechnology alliances are frequently successful, since the peculiar strengths of small firms (entrepreneurship and innovational climate) are complementary to those of the large pharmaceutical firms (critical mass for development activities and scaling-up expertise). As Gambardella (1992) concludes, based on an extensive study of the relations between in-house scientific research and external scientific knowledge in the U.S. pharmaceutical industry: To be part of a network, and to be able to effectively exploit the information that circulates in the network, has become even more valuable than being able to generate new knowledge autonomously.'

As a result, in some high-technology areas, such as biotechnology, electronics, and software, a large portion of the knowledge base for new products is shared, rather than proprietary, knowledge (e.g., Cabo et al. 1996; Chatterji 1996). R&D collaboration between firms and knowledge institutions is rapidly becoming the rule rather than the exception (e.g., Biemans 1992; Millson et al. 1996). Currently, more than 50 percent of Du Pont's new product leads in agriculture stems from university laboratories (MacLachlan 1995). In the pharmaceutical industry, too, knowledge institutions are critical partners in all aspects of new drug development. Taylor (1994) estimates the innovative pharmaceutical industry to spend ten to twenty


percent of the total R&D budget on collaboration and sponsoring of universities and government laboratories. This current trend of forming formal and informal R&D partnerships is likely to continue. CEOs of fifty leading companies believe that by the year 2000 close to 50 percent or more of their technological competitiveness will derive from external sources and partnering (Jonash 1996).

These external relationships might be a reasonable response to the business pressures, but, at the same time, they may create new long-term dependencies and vulnerabilities, as companies are becoming increasingly dependent on outside sources for their technological advances (e.g., Millson et al. 1996). For instance, if industry is going to entrust critical parts of its research to outsiders, there must be confidence that timing to produce results will be respected. Business heads worry about security, cost-effectiveness, and relevance of results. They have the view that only R&D under their direct control can be held to a focus on business needs (Jonash 1996).

The occurrence of these problems can be explained by social capital theory. Network relations may enhance the social capital of a company, by making it easier to get access to information, technical know-how, and financial support. But, at the same time, these relationships may lead to social liability (by reducing the possibilities to relate to companies outside the network, risk of spillover, and high coordination costs of the networkrelations). R&D relationships, especially, are, for the most part, not very tightly knit, and hence the problems relate to lack of information and to opportunism. This chapter focuses on social liability by concentrating on the complications and pitfalls encountered in ten R&D collaborations and then on the methods managers can use to reduce social liability and hence enhance social capital.

TECHNOLOGICAL COLLABORATION

In management literature a variety of terms describe the phenomenon that organizations are no longer defined in terms of their assets alone but increasingly by their relationships with other organizations, as well. In reviewing the literature, Campbell (1996) mentions the virtual corporation, the virtual organization, the network organisation, the modular cooperation, and the web as examples of this notion. The terminology used to describe technological collaboration is variable, as well. The present study starts from the general definition given by Dodgson (1993), who describes technological collaboration: 'as any activity where two or more partners contribute differential resources and know-how to agreed complementary aims.' Chatterji (1996) defines this broad terrain of business relationships to include acquisitions of small companies, exclusive licensing of specific technologies, joint ventures, minority equities, options for future licenses and joint development, R&D contracts, and seed funding of exploratory research at universities, independent research organizations, and start-up companies. This study concentrates on technological collaborations (joint ventures, joint developments, and R&D contracts) that continue for a definite period to achieve a common objective. These might or might not shift into long-term alliances. The focus is on the complications and pitfalls encountered by the management of the leading company. The leading company brings in the original concept for the cooperation, where it is then further


developed in cooperation with the partners, which add varying degrees of value (Lorenzoni and Baden-Fuller 1995).

Motivation to start and factors undermining the success of a collaboration A number of reasons for starting a collaboration are identified in literature (e.g., Dodgson 1993; Hagedoorn 1993; Bruce et al. 1995; Bailey et al. 1996; Chiesa and Manzini 1996). These include improving the efficiency of the developmental process through spreading costs and reducing risks of product development, allowing speedy access to new technology, widening the firm's breadth of technological competence, and providing a window for monitoring technological advances. In addition, technological collaboration can play a key role in implementing corporate strategy-for example, in diversifying, globalizing, and overcoming trade barriers.

Bruce et al. (1995) conducted a survey into the advantages and risks of technological collaboration in 106 UK-based companies in the information and communication industry. More than half of the respondents expressed as their view that collaborations do not accelerate product development but make product development more costly, in general. Bruce et al. expect that some of these problems may be overcome by organizational learning because the companies in their sample with much experience in collaboration (more than 25 percent of R&D-effort stemmed from collaborative effort in the preceding two years in 37 percent of the sample) noted these problems less often. Bruce et al. extract a number of risks in technological collaboration from the management literature: • Alliances are unstable and difficult to manage because of reduced direct

management control (MacLachlan 1995); • There is a risk of creating a rival (Lorange et al. 1992); • There is a risk of creating a dependency on a key supplier or partner; • There is a risk of leakage of skills, experiences, and competencies that may

form the basis of the competitiveness of the firm; • There is a risk of leakage of information and insights into possible markets and

future possibilities; • There may be hidden administering costs of setting up and monitoring of a

collaboration; • There may be costs of time in harmonizing different management styles and

budgeting processes (Farr and Fischer 1992). Although most authors don't refer to social capital theory, the reasons for starting a collaboration and the pitfalls and complications encountered refer to corporate social capital and liability.

Technological collaboration as a process Technological collaboration should not be considered a discrete event but a process (e.g., Hamel et al. 1989; Chatterji 1996; Millson et al. 1996) characterized by a sequence of phases that require the allocation of tangible and intangible resources, and need a structure of organizational procedures and routines for managing them. Chiesa and Manzini (1996) describe the process of technological collaboration through a phase model, as shown in Figure 1.


Define the Select the Establish the Implement the firm's goals ~ partners f+ partnership f+ partnership ..... Outputs

t f f f

I Implement corrective actions '- Measure network's I results

Figure 1. Phase model of technological collaboration (Chiesa and Manzini 1996)

An R&D co-operation is composed of the following phases:

1. Goal definition In the goal-definition phase the firm defines the goals for the collaboration in terms of its short-term objectives (such as exploitation of a definite innovation or development of a new product) and long-term objectives (such as building skills in a technology area). This phase is extremely important, because by doing its homework carefully, a company may avoid a lot of trouble later in the collaboration.

2. Partner selection In the-phase one or more potential partners are identified. This includes collecting information about and exploring the strengths and weaknesses of potential partners and carefully assessing the most appropriate.

3. Establishment of collaborative agreement In this phase the partners get to know each other and define the guidelines for the collaboration. Agreements are made concerning aim, scope, time, type (such as strategic alliance or joint venture) and organization of the collaboration. In this essential phase the basis of trust on which a successful technological collaboration stands is built. Despite its obvious importance, it is seldom distinguished as a separate phase in management literature.

4. Implementation In this phase the partners start to collaborate in line with the guidelines defined in the previous phase. Network activities should be stimulated to maximize integration, communication, and learning.

5. Performance assessment and implementation of corrective actions During and at the end of the collaboration, the partners evaluate whether the objectives are achieved, measure the results, and verify whether the guidelines are actually respected. Decisions on corrective actions should be taken collectively by the partners if they relate to modifying the rules for the collaboration.

Research Questions The present study was designed as an explorative study regarding factors influencing R&D collaborations. An explorative approach was chosen because there is still a mismatch between the literature, which emphasizes the importance of R&D collaboration and indicates ideal typical solutions for starting and developing partnerships, and empirical reality in which a range of many different forms of


f Table 1. Partners and objectives 0 the collaborations in different industry sectors

Partners Objectives

Energy Dutch power plant and knowledge institution Simulation-based power plant

and U.S. software company controlling system

Dutch power plant and knowledge institution Controlling systems for small-scale

and manufacturer of energy controlling power generation (two collaborations)

systems

Dutch power plant and knowledge institution Environmental friendly wood- and

and U.S. software fmn and Swedish wood coal-based energy generation

supplier

Dutch gas distributor and three U.S. power Advanced motor management

plant manufacturers Systems

Automotive Two automotive companies (Gennany, France) New coating for motor bodies

and Dutch steel company and Chemical

conglomerate

Six auto motor body companies Conversion of aluminium motor

bodies

Office Dutch copier manufacturer and suppliers New generation of high-end copiers

systems

Software Irish software company and major clients Tailor-made software solutions

Electronics Japanese e.Iectronics company and four High-perfonnance parallel and optical

universities (UK, France, Gennany, Italy) computing

R&D collaborative efforts can be found, each with its own problems and pitfalls. In this situation it seemed sensible to first collect information on different forms of R&D collaborations, then try to explain the findings, and finally formulate the practical consequences for managing R&D collaborations.

Study Sample The present study was conducted in 1996. Thirty half-structured interviews were conducted with general and R&D managers of ten leading companies in technological collaboration about the complications and pitfalls encountered. I All managers had been actively involved in the preparation and/or the execution of the technological collaborations at issue.

Five collaborations took place in the energy sector (see Table 1). Three of them were initiated by different Dutch power plants and a central research institution concentrating on energy and environmental issues. The other two were initiated by the largest gas distributor in the Netherlands. They were all co-developments. The first collaboration was directed toward the development of a new simulation-based controlling system for power plants. Two others were directed towards the development of advanced controlling systems for small-scale power generation. The fourth concentrated on the development of environmental friendly wood- and coal-based energy generation. The fifth was concerned with advanced motor management.


The partner in the first collaboration was a U.S. software company, in the second a Dutch manufacturer of energy controlling systems, in the third (another) U.S. software firm and a Swedish wood supplier, in the fourth, three U.S. power plant manufacturers; and in the fifth a Dutch manufacturer of motor systems.

Two collaborations took place in the automotive industry. The first was a codevelopment of a new coating for motor bodies between two automotive companies in Germany and France, a large Dutch steel company, and a Dutch-based multinational chemical conglomerate. The second was a long-term R&D collaboration of six motor body companies, currently concentrating on the possibilities of converting aluminium bodies. The seventh collaboration was a codevelopment between the largest copier manufacturer in the Netherlands and its main suppliers to develop a new product family of high-end copiers. The last two were networks-the first in the field of software development and the other in the electronics industry. A worldwide operating software company, based in Ireland, daughter of a large multinational company, works together in supplier-customer partnerships with their clients to provide them with tailor-made software solutions. The last collaboration was a university-industry collaborative network in the area of high-performance parallel and optical computing in four university departments in the United Kingdom, France, Germany, and Italy with a Japanese electronics company.

RESULTS OF THE EXPLORATIVE STUDY

Motivation to Start a Technological Collaboration From the interviews it became apparent that the general motives to start a partnership were the following: • To accelerate time-to-market, • To improve the cost-effectiveness of R&D, • To develop stronger technology competencies, and • To broaden the scope of technology reach and/or geographic coverage.

Unfortunately, in none of the cases were all these objectives met because of communication and coordination problems, misunderstandings and hidden agendas, which led to overrunning of budget or loss of interest by one of the partners. From the interviews it became apparent that technological collaborations are extremely vulnerable, even more than other (supplier-customer) partnerships.

Complications and Pitfalls The following complications and pitfalls were encountered in the different phases of the technological collaboration.

Goal Definition R&D partnerships are mostly directed toward new technologies and markets. Therefore, careful upfront study is essential for a well-balanced assessment of the technological and business opportunities. In two cases the partners were so enthusiastic about the collaboration, that without further study the business opportunities were considered to be good. The respondents stated that, although no


market turned out to exist for the developed product, there was gain of beneficial experience in terms of organizational learning and technical know how. MacLaghlan (1995) comments that where termination of an R&D project can create trauma inside the R&D organization of an individual company, this is even more so in a collaborative effort. Bruce et at. (1995) point in the same direction, by commenting that the collaboration as such might establish its own agenda. The overriding desire of the partners to ensure that the collaboration will be successful may cause the partners to become blind to technological and market reality.

Partner Selection Insufficient monitoring of the R&D environment may lead to obvious partners being overlooked. This is often caused by insufficient management commitment to finding a partner. In one case it took five years to find a partner with the required competencies. The resulting partnership missed the business opportunity because the market had changed in the five-year period. In another case, a leading university department in the country of the lead company was ignored for more than half a year because the company was searching worldwide, ignoring the possibility of finding excellence around the corner. In two cases the technological capability and the financial resources of the partner turned out to be insufficient to conduct its part of the collaboration successfully.

Establishment of Collaborative Agreement One of the problems that emerged from the interviews was that the technological collaboration had evolved in a merely ad hoc way, while the interests of the company were insufficiently secured. In five collaborations inconsistency with corporate interests was mentioned to have caused (severe) problems later in the collaboration. This included lack of clear agreements about the division of the financial and R&D efforts among the partners and insufficient clarity about the way the collaboration should be organized and managed. The omission with the most severe consequences, however, was the lack of a clause into the contract about the distribution of the potential gains. The partner stepped out of the collaboration to use the knowledge competitively.

A number of respondents stated, however, that R&D partnership gains are often too uncertain, and gains too unexpected to capture them all in contracts. Too detailed contracts are also contraproductive. In two collaborations the respondents indicated that their (American) counterparts showed up at the constitutional meeting with their corporate lawyers and fist-sized contracts. This deterred the Europeans and extended the negotiations considerably.

Implementation The most important problem encountered was fear and distrust. Six out of the ten technological collaborations suffered from this at any time during the collaboration. The partners worried about releasing too much confidential information and technology because they feared a hidden agenda and opportunistic behavior of the partner. Two collaborations failed, because one of the partners was more interested in the short-term exploitation of strategic information than in the success of the joint


program. In the pharmaceutical industry, where concentration tendencies are widespread (Omta 1995), fear was expressed that the current R&D partner might later merge with a competitor, which could lead to a drainage of sensitive information. One of the respondents of a university-industry collaboration expressed a fear that strategic information would leak out because a number of the graduate students of the university department got jobs in competitors' R&D departments. Furthermore, respondents warned researchers about asymmetric technological collaborations between large and small companies. The small partner may fear (or hope) that if the technological collaboration turns out successfully, it will lose independence and be taken over by the financially stronger partner.

In three cases intercultural problems were encountered. In one case, language problems and cultural differences hindered open communication between Japanese and European companies. But problems also arose with American companies. The problems caused by differences in negotiating culture were already mentioned. In two cases there were also complaints about lack of openness and the provision of scanty, not up-to-date information by the American counterparts.

Peiformance Assessment and Implementation of Corrective Action The partners of large R&D networks tend to underestimate the communication and coordination problems and costs that are encountered in large collaborations. Four large R&D partnerships overran their budgets. Obviously, management still encounters difficulties in coping with the coordination of networks. The central problem in this respect is the absence of a steering agency. The network, or rather the structure of the network, to a large extent determines how actors work together. Consequently, executing R&D within the confines of a network requires different steering solutions than the execution of in-house R&D. Moreover, the fit between the nature of the research problems to be addressed in the network and the type of the structure of the network is not always self-evident. Quite often the structure of the network is not explicitly chosen but merely accidental-for example, based upon previous experiences, as was found in a recent study of upstream R&D cooperation between firms and knowledge institutions within the BIOTECH program (Cabo et al. 1996).

A Social Capital Explanation of Problems in R&D Collaboration The explorative study reveals that R&D collaboration is widespread but that many relationships are characterized by social liability. More in particular, the study shows that fear and distrust-that is, the risk of opportunism by the partner and the lack of adequate information on (possible) partners-seem to be attributes of many relationships. Apparently many of the firms investigated have difficulties adequately managing their R&D partnerships and hence enhancing their social capital. How do we explain this high frequency of problematic R&D relationships?

A first explanation refers to the nature of R&D and its role with respect to corporate goals. The results of R&D are difficult to appropriate, patenting only covers part of these results and hence firms will be secretive about their R&D activities. According to this view it is considered to be rational to perform R&D in-house and refrain from collaborative R&D activities. The empirical evidence


indicating complications and pitfalls is then an indication of the nonrationality of R&D collaboration. However, there are several reasons why this explanation does not suffice.

First, R&D ranges from fundamental scientific research to concrete development of products and processes. No firm can encompass all these R&D activities and thus it has to at least cooperate with the firms and research institutes having the competencies that are not available (and are also not economic to have) in the firm itself. The recent tendency to outsource those activities that are not central to the central mission of the firm results in an increase of collaborative activity in this realm, as well.

Second, some R&D activities are simply too expensive for the single firm; in these cases firms are forced to cooperate, sometimes even with competitors.

Third, in some fields oftechnology, such as biotechnology, a large generic body of knowledge is shared by firms and research institutes. Here it is economic to engage in collaborations instead of developing a proprietary knowledge base. In other words, if R&D collaboration is more the rule rather than the exception, one needs to look for other explanations, instead of pointing to the nonrationality of R&D collaboration.

The social capital argument assumes that the behavior and expectations of actors are constrained by the degree to which the relationship between the actors is embedded in the network structure. Consequently, one may distinguish between a situation in which network structure is closely knit and relationships are redundant (Le., actor A has relationships with actors Band C, and Band C also have a mutual relationship) and a situation in which more non-redundant relationships prevail. Coleman (1988) describes this phenomenon as the degree of closure of a network. In a similar vein, Granovetter speaks about strong versus weak ties in a network (Granovetter 1985). Burt (1992) formulates it slightly different as the occurrence of structural holes. Conversely, it can also be argued that, considering the role of social relationships in technological fields such as biotechnology, the absence of adequate social relationships can be seen as a liability of research laboratories in such fields, as well.

Networks with relatively many nonredundant relationships provide less social constraints. Granovetter (1985) argues that this network effect occurs via the mechanism of reputation (see also Raub and Weesie 1990). If a network comprises many redundant relationships, relatively many actors will have information on the actions of an actor. The actor's reputation (as assigned by the others in the network) depends on this information. Accordingly, actors will refrain from (otherwise individual profitable) opportunistic behavior because they anticipate that it will affect their own reputation in the network. As Raub and Weesie (1990) show in a series of game-theoretic experiments. this effect is diminished when the degree of embedded ness decreases.

In our view. the network of R&D relationships is a prime example of a network with relatively few redundant relationships. Actors do not look for similarity in the attributes of their partners in the case of R&D relationships but, instead, attempt to acquire partnerships with actors having complementary rather than similar attributes. This will result in less densely knit networks. In fact. one may expect that firms


looking for R&D relationships will attempt to acquire partners not in their direct vicinity but in remote locations of the network. Firms will, especially in the case of R&D relationships, strategically invest in their (R&D) social capital by emphasizing nonredundant relationships or, in Burt's words, relationships that span structural holes.

If this argument is correct and the number of nonredundant relationships is relatively high in networks of R&D relationships, one may expect a lack of social constraints and as a result the occurrence of problems as indicated by the exploratory study. These problems will be characteristic of all R&D relationships: R&D networks are less dense and therefore do not generate the social norms facilitating consensus on mutual expectations.

Although the exploratory study reveals a list of various different types of pitfalls and complications, they nevertheless all point to inadequacies of network structure. The problems indicated by the respondents are either problems related to their own behavior toward partners (such as the uncritical acceptance of partners without an adequate check of their relevancy for a project), the behavior of the partner (such as a short-term exploitation of strategic information), or, and predominantly, problems of inadequate mutual expectations resulting in fear and distrust (suc as the fear of leaking of strategic information). Nooteboom's (this volume) argument about the importance of including a trusted third party (the go-between) in an R&D collaboration, also refers to the importance of increasing the information level about the partner(s), enhancing trust by a party from outside the network.

In the exploratory study we did not systematically collect information on the nature of the respective R&D network structures. This will be the topic of a followup project. In the remainder of this chapter we want to focus on the practical consequences for the management of such R&D partnerships.

In the concluding chapter in Nohria and Eccles' collection of articles on networks and organizations (Nohria and Eccles 1992), Moss Kanter and Eccles correctly observe that there is a large discrepancy between the academic discussion on attributes of networks of organizations and the completely different perspective from which managers deal with their networks (Kanter and Eccles 1992: 521-527). Accordingly, these authors request more practically inclined network analyses.

Our intentions in this chapter are, practical, although less far-reaching. If pitfalls and complications of R&D partnerships are related to attributes of the R&D network structure, then management tools should be developed that enable managers to adequately reduce social liability and hence enhance the social capital of such loosely knit networks where they cannot assume adequately developed expectations and mutually attuned behavior.

GOAL DEFINITION

A number of problems could have been avoided if the technology forecasting was more carefully assessed. Companies need to establish robust technology forecasting systems, that express technology needs for the next five to ten years, and monitor and interpret developments in different technologies, the emerging trends in customer needs, and competitor actions. Based on these forecasting activities


Table 2. Competitive impact of technologies

Technology Description

Emerging

Pacing

Key

Base

Has not yet demonstrated potential for changing the basis of

competition.

Has demonstrated potential to change the basis of competition, but has

not yet been embodied in products or processes.

Is embodied in products and processes. It has a major impact on

product or process value-adding (in terms of cost, quality or

performance); and allows proprietyl patented positions.

Is required to be in business. It offers little potential for value-adding,

because it is widespread and shared among competitors

technology road maps can be set up that link future product plans to the technologies required to achieve them. Many authors (e.g., Hamel et al. 1989; Roussel et al. 1991) have emphasized that firms should develop only a few strategic technological capabilities and should outsource the other ones. The emerging role of R&D management is to balance internal and external technological capabilities by identifying the projects that are feasible to take outside, and match these with external sources. Via management tools such as the outsourcing and partnering matrix (discussed below), firms can decide which technologies should be developed in-house, which in collaboration with one or more partners, and which technologies are better outsourced.

Outsourcing and Partnering Matrix The first step in framing the outsourcing and partnering matrix is to establish the competitive impact of the company's technologies by dividing them in emerging (embryonic), pacing, key, and base (existing) technologies (Roussel et al. 1991), as shown in Table 2. To this end, for each technology the following questions must be answered: • Does this technology have the potential for competitive differentiation? • Could it become critical to the company? • What is its market value?

The second step is to assess the internal technological capability for each of the technologies using the categories weak, moderate, and strong. By combining the competitive technology impact with the internal technological capabilities, we get the outsourcing and partnering matrix, that can serve as a comprehensive management tool for the strategic foundation of make-buy-or-collaborate decisions (see Table 3).

Emerging Technologies An emerging technology may have a competitIve impact in the future. If the technological capability is strong, optimizing the technological capability to reinforce the potential competitive advantage is called for. If the internal technological capability is moderate or weak, catching up may be necessary. However, uncertainty demands that the R&D environment be scanned-for many


Table 3. Outsourcing and partnering matrix (Harris et a1. 1996)

Competitive impact of Internal technological capability Technology Weak Moderate Strong

Emerging Scan Scan/collaborate Collaborate Pacing Collaborate Share risks In-house

Key Optimise Optimise In-house

Base Outsource Outsourcelexchange SelVexchange

partners and flexible relationships, preferably in strategic partnerships and alliances or via contract research and sponsoring of knowledge institutions. In all cases, adequate patent protection strategies need to be considered.

Pacing Technologies Pacing technologies may have a strong competitive impact on the short or medium-term. If one's technological capability is strong relative to the rest of the world, the bias should be towards doing the work in-house. Extra investments might be required for research into the application of the technology in new products and markets. If one's technological capability is moderate, sharing the risk by strategic alliances with partner firms makes the most sense. If one's technological capability is weak, acquiring licences or joint development may be viable alternatives. Pacing technologies need utmost management care, especially if the technology is maturing rapidly, because these might become essential for tomorrow's business. It is therefore necessary to scan research efforts of competitors and potential technology sources intensively. Furthermore, the technologies need to be protected carefully.

Key Technologies Generally speaking, key technologies, being critical to current competitiveness, should be owned by the company. If one's technological capability is weak or moderate in the technology area at issue, one should acquire extra technological capability for building in-house R&D strength by acquisition or by introduction of a substitute technology.

Base Technologies For noncritical base technologies outsourcing might be the appropriate choice if one's technological capability in the field is weak. If it is moderate, it may serve as a means of exchange in a partnership. If it is strong, it either may serve as a means of exchange or may be sold to focus the internal technological capabilities on key technologies.

It is important to remember that the outsourcing and partnering matrix does not render a static model. For instance, by acquiring extra technological capacity in a field of key technology where one's technological capability is weak, one should gradually shift from the left-hand to the right-hand side of the third row in the matrix.


Partner Selection Erens et al. (1996) conclude, based on a study of fifty companies in Europe, U.S. and the Far East (including Airbus, Boeing, Canon, Hitachi, IBM, Matsushita Philips, and Toyota), that many leading companies are too selfish in the search for an appropriate partner. They emphasize that companies should not only look at what they need and want from a potential partner but also what they can deliver to a partner in terms of skills, market access, and economies of scale. In addition, they emphasize that the companies are too much oriented toward the hard aspects of the collaboration, whereas a good match of the soft aspects, including business culture and chemistry of (top) managers are far more important for successful cooperation. Bailey et al. (1996), based on a recent study of seventy UK-based companies in different industry sectors, also conclude that selecting partners for collaboration on technical merits alone is clearly a suboptimal solution. Relying on the partner's track record in previous collaborations has turned out to be a poor basis for collaborator selection, as well. The authors even call this a recipe for disappointment. In short, a company has to be very careful in the selection of the potential partner(s}. An ideal partner should: • Have an interest in and expect equal advantages of the collaboration, • Have complementary technological capabilities and knowledge, • Be capable and willing to share financial risks, • Have no record of opportunistic behavior in former collaborations, and • Have a business culture that favors collaboration with open communication and

a quality vision. Its management should not be afraid of losing some of its authority.

Less suitable partners are: • Partners with comparable core business or geographical markets (security

conflicts may easily arise where partners operate as direct competitors in other markets),

• Partners with business cultures that differ too radically (Lorange et al. (1992) also identify compatible organizational cultures as important, encompassing similar perception of the environment, organizational values, and operational routines), and

• Dependent companies (this finding is in line with that of Saxenian (1990); high-tech Silicon Valley companies receive no more than 20 percent of any supplier's output to ensure that they do not become too dependent on external partners).

Supporting the Outsourcing and Partnering Decisions Choosing a technology field in order to sustain long-term competitive advantage, choosing outsourcing or not, and choosing of partners are-in the concrete practice of managing a firm-decisions that are closely related and difficult to disentangle. Consequently, if managers do not sufficiently separate aspects of these three decisions, decision making will be less than optimal.

For example, suppose that a firm wants to develop a high-tech product that includes different technologies. Further presume that some of the technologies still


have to be developed and are unrelated to key technologies within the firm. In this case, it would be advantageous for the firm to find a partner with the relevant competencies. Unfortunately, the most adequate, or perhaps the only feasible, partner for this activity is one of the firm's main competitors in other fields. This being the case, the firm will probably refrain from considering this potential partner, even when the collaboration would be necessary from the point of view of developing the new product.

But if the partner search would initially be confined to identifying the relevant attributes of a potential partner, and then assessing the relative importance of these attributes, then the first selection would be configurations of attributes rather than concrete firms. By later adding names to the relevant configurations of attributes, the search would probably include the competitor. In this way, the firm can systematically scan the social network for potential partners in the periphery of the network, which, as we indicated earlier are in the case of R&D relationships, can sometimes be very significant in terms of reinforcement of R&D competencies.

In our view, this implies that managers should consider the set of decision problems as one complex decision problem that first has to be analytically unpacked in various aspects, after which the aspects should be synthesized in a meaningful way. A powerful way of supporting this activity is the application of multicriteria decision analysis. In a related project (see Hummel et al. 1998) we are using Saaty's Analytic Hierarchy Process (AHP), and the related Expert Choice software, to develop a support system for such decisions.

The Analytic Hierarchy Process method comprises four steps: 1) decomposition of the complex problem into dimensions, criteria, aspects, and so on; 2) hierarchization in terms of different levels at which the components can be seen; 3) pair-wise comparison of the dimensions, criteria, and so on, at each level, using a nine-point quantitative scale, resulting in ranks of importance of the dimensions, criteria, and so on for each of the levels; and 4) synthesizing the respective rankings by combining the scores of each of the levels, resulting in eigenvectors for each of the possibilities. We refer to Saaty (1990) for a further elaboration of the method.

Although different types of AHPs can be developed regarding decisions about R&D management (for example, a cost benefit analysis of the R&D portfolio, see Liberatore et al. 1992), it is our contention that R&D decisions involving other actors should be supported by a more elaborate model. This model combines two different approaches: a forward and a backward planning approach. The forward planning approach should start with the assessment of the actual R&D capabilities and existing collaborations and consider the potential objectives that can be reached with these capabilities and collaborations (also including the forces in the environment and the behavior of other actors). In contrast, the backward planning approach should take as its starting point the desired future (that is, the improvement of the competitive position of the firm), considering various potential strategies, and indicate the necessary R&D capabilities and collaborations in order to reach this objective (taking into account forces in the environment and the behavior of other actors). Discrepancies between the two approaches are diminished by subsequent iterations of both approaches. In Figure 2 the two models are exemplified.


Establishment of a Collaborative Agreement Too extensive contractual arrangements should be avoided because they are likely to be interpreted as distrust of the good intentions of the partner. However, we don't go as far as Wolff (1994), who states that contractual arrangements are important only insofar as they define the terms under which a partner may exit the collaboration. According to Chiesa and Manzini (1996) the partners should agree on the objectives of the collaboration (expected results), the time required for the project, the expected contribution of each partner in terms of time and resources, the organizational structure, the role of each partner (allocation of tasks), the coordination mechanisms, and criteria for assessing the collaboration's performance and evaluating results. In accordance with Chiesa and Manzini's conclusions, our respondents indicated that a contractual arrangement should minimally codify the following: • The financial and personal responsibilities of the partners, • The division of the possible gains among the partners, • The way that knowledge will be protected, including patent and trade secret

rights and confidentiality agreements, • Criteria for measuring and monitoring progress, so that deviations can be

identified and potential problems can be overcome (this includes milestones and deadlines of the project, responsibilities and accountability of the project team and the founding of a steering committee; for example, a contract between the biotechnology company Immunex and the pharmaceutical company Smith Kline Beecham included a list of the principal scientists who would be responsible, a detailed schedule of at least weekly telephone conferences, and a provision for at least quarterly joint meetings) (Leonard-Barton 1995), and

• Penalty clauses to discourage opportunistic behavior.

Implementation We are convinced that partnership management is to a great extent management of trust and goes far beyond signing confidentiality agreements and agreeing to guidelines. To be effective, a collaboration requires the bridging across different business cultures and lines of responsibility in the participating companies. Lewis (1990) states that lack of trust is the major reason why many R&D managers don't think their alliances are working as well as they should. Based on a study of 84 alliances, Lorange et al. (1992) conclude that trust and commitment are necessary conditions for long-term collaboration. Alliances have to be designed to create win-win situations rather than some form of a zero-sum game, otherwise they will certainly fail (Rai et al. 1996). Bruce et al. (1995) comment that the creation of a climate of trust might appear to be in direct conflict with the notion of establishing limits to the knowledge exchanged. It is the challenge for the partners to find the critical balance of openness and confidentiality. It seems feasible that over time, as trust is built, the need to limit the scope of the collaboration might decrease. Trust is built up by ensuring that partners receive suitable rewards for their efforts. To show their interest in the venture, each partner should contribute high-quality R&D staff. During the whole cooperation it is critical that the partners keep each other informed about what they are doing. Frequent communication in building up mutual understanding and in checking on the progress of the collaboration saves time and


Focus

Objectives

Strategy

Environment

Technological Fields - saturated - growth • emergent

Policy

Technologies

Type of technologies

Location

Partners

• beginning · mature • declining

. stable

. changing potential - fast changing

· emerging · pacing • key · base

• horizontal • supplier • buyer

Figure 28. A forward planning approach on R&D collaboration decisions: asssessment starts from the current technology portfolio and partnerships


Focus Competitive position of firm

Scenario

Strategy

Technology focus on - Emerging - Pacing - Key - Base

Location

Partners

Figure 2b. A backward planning approach on R&D collaboration decisions: asssessment starts from the desired technology portfolio and partnerships

costs by preventing far more costly adjustments later in the collaboration. To communicate frequently, exchanging all relevant memos and team reports helps in creating a climate of trust.

Alliances can't be run over the telephone (Wolff 1994). The R&D staff of each partner should visit and be allowed to work temporarily in the partner's laboratory to gain a hands-on understanding of the situation. If geographical colocation is not feasible or too expensive, the use of advanced telecommunication systems (videoconferencing) might be the best alternative that allows direct face-to-face contacts. In most cases, successful cooperative relations were established gradually and were started with small and uncomplicated projects. It is important that the partners learn as much as possible about each other before the partnership starts. Leonard-Barton (1995) notes that the more managers understand the classes of problems that may occur and develop the ability to anticipate issues through preagreement diagnosis, the greater the likelihood of success. Managers have to map possible differences in business culture among countries and differences in management style among companies and try to foresee and anticipate possible key issues of disagreement. Ideally, an in-depth organizational report should be made of the business culture of the partner in advance of the collaborative agreement. This is often problematic because differences are more likely to emerge as the collaboration proceeds (Bruce et al. 1995). Nooteboom's (this volume) argument about the


importance of including a trusted third party (the go-between) in an R&D collaboration, also refers to the importance of increasing the information level about the partner(s) and thereby enhancing trust by a party from outside the network.

Performance Assessment and Implementation of Corrective Action The partners should strive to clear accountability through performance measures equivalent to those used for measuring internal R&D performance. A joint steering committee with enough authority should meet periodically to review goals and progress against schedules. This is very important because it creates a feeling of urgency. In the home companies of the partners there are always R&D activities that seem more urgent than a collaborative project with a far-away partner. Stringent deadlines help to avoid arrears and budget overruns. Bruce et al. (1995) also underline the importance of a steering committee as the management's visible commitment to the collaboration.

CONCLUSIONS

As companies strive for more sustainable technological competitiveness, they must rethink a number of critical technology management processes. New strategies, processes, and organizational roles are required to develop and utilize the wider range of external resources and competencies needed to optimize the social capital of the company. According to this study, partnerships to be effective, should meet the following requirements: • The partnership is based on a good business opportunity, • The goals are equally beneficial to and considered important by both partners. • The scope of the partnership is well defined, • Each partner makes an equal contribution, based on complementary strengths, • There are no strategic conflicts between the partners, and no side dominates the

partnership, • The partners show their interest in the venture, and each partner contributes a

high-quality R&D staff, • A joint steering committee meets periodically to review goals and progress

against schedules, and • The partners anticipate and respect the inevitable differences in business

culture.

These requirements align with the key attributes for strategic alliances as distinguished by Hampson and Kwok (l996)-trust (reliability), commitment (a win-win attitude), (acknowledged) interdependence, cooperation (self-interest to achieve mutual goals), communication (accurate, timely and relevant), and (open) joint problem solving.

It should be kept in mind that the survey reported here focused on social liability occurring in single cooperative efforts. Bruce et al. (1995) rightly point to the potential danger of a set of rules for success on collaboration that is based on information about single collaborations only. Cooperation is basically an evolutionary process. It may start as a highly structured project with clearly defined


responsibilities and objectives in terms of developing a single product, and it may gradually develop into a profitable but less structured long term alliance as trust and mutual confidence grow. There are also many intangible and unpredictable factors that might affect the manner in which collaborations develop. The partners may have more reasons to enter the collaboration (for example, turning potential rivals into allies or getting access to foreign markets), and circumstances may change. The collaboration may have unintended side-effects and present unanticipated opportunities, as well. As was shown before, even unsuccessful cooperations may lead to extensive organizational learning, and hence may enhance the social capital of the company.

As a consequence, the relationships between the various aspects mentioned above are not self-evident. For instance, the outsourcinglpartnering matrix might indicate that a particular competitor is potentially an adequate collaborator for a specific R&D project, whereas previous experiences with the competitor may hamper the firm's ability to see this possibility. Consequently, it is paramount that the various aspects are addressed in a systematic way, by weighing the various aspects against each other. In this chapter we showed the potential usefulness of Saaty's Analytic Hierarchy Process to support this process.

NOTES

The interviews were conducted by 26 students of Technical Business Administration. School of Management and Organization, University of Groningen. The Netherlands.

•

Technological Prestige and the Accumulation of Alliance Capital

ABSTRACT

20 Toby E. Stuart

In this chapter, I argue that the prestige of an actor is a primary determinant of its ability to access resources held by others. The reason for this is that relationships with prestigious actors are inherently valuable because they convey status to affiliates, and so prestigious actors have many opportunities to form new relationships in which they exchange status for other kinds of resources. The chapter contains an empirical analysis that shows that high prestige semiconductor firms establish many license alliances in which they gain the rights to produce and sell the proprietary technologies of competing organizations. If we conceive of a portfolio of interorganizational access relationships as a component of corporate social capital, the findings show that social capital accrues at a high rate to high status organizations.

INTRODUCTION

Social capital has been defined in many different ways, but it is always presumed to inhere in the relationships that bring actors together (Coleman 1988) or the capacity of an actor to bring others together where no relationships exist (Burt 1992). When social relationships are completely absent, social capital is nonexistent. For this reason, one must understand why and where relationships are formed in a social structure to understand how and how much social capital is accumulated.

Sociologists have posited numerous behavioral tendencies that limit the incidence of contact between actors, and so affect the accumulation of social capital in a system. These tendencies are the mechanisms that create highly circumscribed

Technological Prestige and the Accumulation of Alliance Capital - 377

patterns of relationship formation; their effect is to cluster relations within and around particular social structural locations. The behavioral proclivity that has garnered the greatest attention in sociology is homophily: if strong homophilous tendencies exist within a population, then the demographic composition (broadly defined) of that community of actors will dictate the frequency and pattern of relationship formation within it (Homans 1950; McPherson, Popielarz and Drobnic 1992). Marsden (1983) refers to factors that confine relationship formation under the rubric of 'restricted access,' which suggests the operation of social structural and/or social psychological mechanisms that sharply delimit patterns of interaction. While access restrictions create the structural conditions necessary for the brokerage role (Burt 1992; White, Breiger, and Boorman 1976) and therefore give substance to the weak tie metaphor (Granovetter 1973), they can also limit the aggregate amount of social capital in a community. Broadly speaking, access restrictions imply that social capital will accrue unevenly and at the greatest rate to the occupants of positions that, for one reason or another, convey access to a diverse array of others.

The general phenomenon that I investigate in this chapter is the association between actor prestige and the ability to accrue social capital by building exchange relations with one's alters. The specific instance of this more general phenomenon that I explore here concerns whether the level of technological prestige of a firm influences its propensity to engage in strategic alliances that convey access to the technological resources of alliance partners. My thesis is that high prestige creates opportunities for access: organizations with high-status generally possess many opportunities to build new intercorporate coalitions. Not only are they able to pick and choose among potential associates, high prestige firms are able to negotiate access to the key resources of their competitors on appealing terms. In its essence, the argument that I propose is that high-prestige firms exchange a social resourcetheir imprimatur-for a technological resource-access to their competitors' innovation. Hence, this chapter investigates the effect of an organization's level of prestige on its accumulation of material and alliance capital through a portfolio of strategic alliances.

NETWORK-BASED PERSPECTIVES ON ACCESS RESTRICTIONS IN STRA TEGIC ALLIANCE NETWORKS

In organization theory and sociology, a number of studies have examined the structural foundations of alliance formations. Perhaps the most widely influential work in this area has been the group of 'embeddedness' studies (Granovetter 1985). Scholars working from this view have argued that the pattern of new alliance formations depends upon from the configuration of previously-established coalitions. The reason that existing (or past) alliance ties are important for the pattern and volume of formation of new coalitions is that cheap information about the quality and the trustworthiness of potential alliance partners diffuses across existing inter-organizational relations. Information on the attributes of possible alliance partners is available from previous, direct interactions with them (Gulati 1995a; Podolny 1994; Powell, Koput, and Smith-Doerr 1996; Granovetter 1985; Walker, Kogut, and Shan 1997) and from the experiences conveyed by other parties who have had relations with them (Burt and Knez 1995; Nooteboom, this volume).


However, because firms differ in terms of how they are positioned in the interorganizational alliance network, they also differ in the extent to which they are able to gather information about potential alliance partners.

Possessing knowledge of the characteristics of potential exchange partners is important because it influences the transactions costs associated with forming new relationships. There are two reasons for this. First, intimate knowledge of a potential partner eliminates the cost of investigating the quality of that organization, and thereby removes one of the more significant expenses of alliance formation. Second, when a potential partner is known to be reliable and trustworthy, a focal firm may be willing to enter into a business relationship with that partner without contractually specifying a priori all of the terms of the association (see the discussion in Knoke, this volume). Conforming to Marsden's (1983) notion of restricted access, embeddedness perspectives on inter-firm strategic alliances have argued that existing alliance ties structure the search for new associates, directing attention toward previous associates and thereby influencing the rate of accrual of social capital.

From this very brief review, it is apparent that much of the sociologicallyinformed literature on interorganizational relationships explains alliance formations according to the influence of previously-established intercorporate relations on the search for and selection of new partners. In contrast to the existing sociological literature, I will use this chapter to investigate how the technological structure of the industry in which I locate the empirical analysis influences the formation of new (technology trading) alliance ties. In particular, I will investigate the prediction that prestige differences between firms have a systematic effect on their proclivities to enter into new intercorporate coalitions. The argument that I make is that prestige creates invitations for association because of the benefits that others derive from relationships with well-known actors. Therefore, high-prestige raises the ability of organizations to develop 'alliance capital'-a collection of relationships with other firms that can be a valuable source of information, know-how, intellectual property, and new customers (see Smith-Doerr et aI., this volume).

Because the empirical analysis will focus on strategic alliances between hightechnology firms, I will use a measure of corporate technological prestige. In general, organizations acquire prestige when they make significant contributions to the communities to which they belong. In the context of a high-technology market, firms attain prestige by forging the paths for new technologies and opening up possibilities for follow-on inventions (Podolny and Stuart 1995). Hence, developing inventions that other innovators recognize as important technological achievements is the underpinning of high technological prestige. Just as the members of any community of actors can be ranked by relative prestige, every industrial community can be described as an ordering of status positions (Podolny 1993). Consider any high-technology arena at a particular instant in time, and the names of its most prestigious members will come quickly to mind.

In this chapter, I specifically focus on the relationship between corporate technological prestige and the incidence at which firms establish strategic license alliances with competing organizations. License alliances are inter-firm agreements in which one organization gains the rights to produce or sell the proprietary


technology of a second firm, usually in exchange for a royalty on units sold or a lump sum payment. Therefore, firms that serve as the licensee in these asymmetric alliances do so to gain access to the proprietary innovations of a competitor. I My thesis is that organizations with high prestige will be the most frequent users of license alliances to acquire the rights to utilize technologies developed by their competitors. A number of factors produce the association between actor prestige and license alliance opportunities, but for the objectives of the present analysis, one determinant merits particular emphasis.

High prestige firms encounter many opportunities to enter alliances because status is implicitly transferred across inter-corporate ties (Podolny 1994), which creates benefits for the affiliates of high-status organizations (Stuart 1998a). Moreover, the advantages of affiliations with high-status firms are significant enough that other organizations are likely to concede generous contractual terms in order to secure the organization. This means that high-status firms are able to negotiate favorable contract terms when they establish alliances.

Relationships with high-prestige firms create value because external resource holders such as potential customers pay close attention to the strategic initiatives of prestigious firms. Because of the presumption that prestigious firms will eschew collaborations with those of mundane quality, prestige is transferred to affiliates when associations with prestigious actors are publicized. Therefore, associations with prestigious actors are status-enhancing: they can substantially improve the reputation of the 'connected' actor. An enhanced ability to attract resources is likely to follow the establishment of an exchange relationship with a prestigious affiliate, and this implies that prestigious associations are a form of alliance capital that can create a meaningful competitive advantage (Podolny 1994; Baum and Oliver 1991; Stuart, Hoang, and Hybels 1999).

In high-technology industries, the transfer of status from high-prestige actors to other organizations or to new initiatives is an every-day phenomenon. One very notable instance of this dynamic occurred when IBM, a high-status manufacturer of mainframe computers, chose to enter the fledgling personal computer industry in the early 1980s (Anderson 1995). IBM's entry into the computer business is widely thought to have brought legitimacy to the industry as a whole. A similar dynamic occurs when a prestigious enterprise backs the initiatives of an unknown firm by establishing an alliance with it, such as when three well-regarded pharmaceutical companies (Pfizer, Glaxo Wellcome, and Smith-Kline Beecham) recently formed technology alliances with Cantab, a young biotech company on the brink of financial distress. Because there is a presumption that prestigious organizations perform a thorough evaluation (due diligence) before entering into a cooperative venture, a technology alliance with a prestigious partner is precisely the type of tie that raises public regard for the initiatives of the lesser-known firm. Particularly when a prestigious firm licenses technology from a different organization, attention is directed to the endeavors of the affiliate and greater interest may be expressed in the exchanged technologies. A license alliance conveys an implicit endorsement, and when the licensee is a highly-regarded enterprise, it is a particularly valuable certification.


High prestige firms are willing to purvey endorsements for technology access for two reasons. First, as previously noted, prestigious firms possess the ability to influence the terms of trade so that they can strike favorable alliance deals. Because they are distinguished players, other actors will realize an advantage from associating their initiatives with prestigious firms. Given that they can significantly raise the value of their associate's endeavors, prestigious actors will enjoy an advantage in negotiating the terms of an alliance contract.

Second, it is inexpensive to govern alliances when prestigious organizations are participants in them, particularly when their affiliates possess lower status. The reason for this is that high-prestige firms-central and influential actors in the communities to which they belong-have a unique ability to enforce contract terms and to block attempts on the part of their strategic partners to act against their interests. From the vantage point of high-prestige firms, the transaction costs associated with alliance formation are low because a positive reference from them will be of great future benefit to their alliance partners, while negative referrals from them will be particularly damaging to the reputations of young, small, and low status organizations. The importance of staying on the good side of a prestigious firm creates a strong disincentive on the part of their partners to defect against the terms of an alliance contract or to otherwise behave in a manner that is disadvantageous to the interests of a high prestige partner.

For the reasons just outlined, my central contention is that prestigious organizations, by virtue of their standing in an industry's status order, are able to choose among potential alliance partners and probably at favorable terms. Therefore, I expect that technologically prestigious firms will have a relatively high number of options to establish strategic coalitions on terms that appeal to them, and so are likely to form many alliances. In particular, I investigate the extent to which its level of prestige affects the likelihood that a firm will establish strategic license alliances in which it receives (in-licenses) technology from a collaborator. I test this prediction in an analysis of the effects of technological prestige on the rate of formation of licensing alliances in the worldwide semiconductor industry.

THE SEMICONDUCTOR INDUSTRY

The sample that I analyze consists of all semiconductors firms for which I was able to collect annual semiconductor sales during the analysis window (1986-1992)? Because many semiconductor producers are broadly diversified (e.g., IBM, Siemens, Fujitsu), it is typically not possible to collect semiconductor revenues for many of the firms in the industry from securities filings. Therefore, I acquired sales data from Dataquest, a high-technology consulting firm. Dataquest tracks all merchant semiconductor firms with annual sales in excess of $10 million; given their data collection procedure, the sample consists of the firms in the industry with the largest sales volume. The Dataquest database consists of 150 companies, although some are not present in the data during all of the years. The firms in the sample are headquartered in the US, Europe, Japan, and other Southeast Asian nations. As a group, the firms in the sample accounted for over 90 percent of the total, worldwide semiconductor production volume in 1991.

Technological Prestige and the Accumulation of AlIiance Capital - 381

ESTIMATION

I treat the organization as the unit of analysis in the alliance formation process and estimate the rate at which a focal organization forms new relations.3 Assuming a continuous-time event history, the hazard of alIiance formation, denoted rit), is expressed:

(1)

where Pk(t,t+At) represents the probability that a firm will experience an alliance of

type k during the interval from t to t+L1t (Tuma and Hannan 1984). In this chapter, I will only examine the determinants of one-way (i.e., directional) technology license agreements. These are agreements in which technology is transferred from one firm to another. Therefore, the data are asymmetric ties and it is possible to designate which partner sends and which one receives technology. 'Receive' alliances are considered to be those in which a focal firm is the licensee or recipient of technology from a competitor. 'Supply' alliances are those in which a focal firm is the licenser or source of technology. which is transferred to a competitor. Hence, I specify the state space to allow for the occurrence of two event types.

Estimating equation 1 requires an assumption regarding the form of duration dependence in the alIiance formation process. I have selected a piecewise exponential model to estimate the hazard because it is extremely flexible with respect to the form of duration dependence. The model assumes that the baseline transition rate, given by the coefficient estimates on a vector of m time periods, is constant within each time period. However, the baseline rate may change across time periods. The piecewise exponential can be written as:

rk (t) = exp(am + f3Xt) (2)

where <Xm represents a vector of m constant coefficients designating the baseline hazard (one coefficient is estimated for each time period), XI is an nxq matrix of q covariates, some of which are time-varying and updated annualIy, and ~ is a vector of q time-constant coefficients to be estimated. Note also that the formation of a technology alliance is a repeatable event. Consequently, firms remain in the risk set for the formation of an alliance even after they have established previous alliances. Repeatable events are unproblematic in event history analysis: in practice, the episodes between alliance events are separate observations, all spells are split at the end of each calendar year to update the covariates, and all observations are then pooled across units and time (Tuma and Hannan 1984).

Because I distinguish between technology supply and receive alliances in the analysis, I can contrast the effect of prestige on the rate at which a focal firm is a licensee (recipient) versus licenser (source) of technology. Hence, k indexes two types of events in (1) and (2) above.


VARIABLES

Dependent Variable The database for the analysis includes all publicly-reported technology alliances involving the firms in the semiconductor industry between 1981 to 1992. An event occurs on any month during which a firm in the sample announces the formation of a license alliance with a different semiconductor firm (the partner firm can be inside or outside of the sample of 150 that I analyze). The sources for the alliance data were the electronics trade press, the business press, company press releases, annual reports and 10Ks, and periodical indices.

Technological Prestige Following Podolny and Stuart (1995), I define a firm's technological prestige as its indegree score in a network of patent citations consisting of all semiconductor inventions patented in the US.

Patents grant inventors exclusive property rights for the commercial use of their inventions for a fixed period of time (see Smith-Doerr et aI., this volume, for an extensive disucssion on patents). Patents are issued by the government in exchange for the inventor's consent to publish a detailed description of an invention. To receive a patent, an inventor must file an application that fully discloses a nonobvious and industrially useful invention. The Patent Office publishes descriptions of all patented inventions in successful (granted) patent applications.

Patent applications must contain a list of citations to the existing patents that had made technological claims similar to those claimed in the application. Hence, citations are issued to the patented inventions that are nearest in technical content to the proposed invention. When the Patent Office receives a patent application, it is assigned to a patent examiner who is knowledgeable in the pertinent area of technology. The examiner then searches through a database of existing patents to verify that the list of references in the patent application, known as the 'prior art,' is complete. If the patent application is ultimately approved, it is published with the list of prior art.

Citing a preexisting patent does not protect a patent holder from a legal action initiated by other inventors: any party may contest the validity of a patent, even if the action is filed by the holder of a patent that is cited in the disputed claim. For this reason, the patentee accomplishes nothing by including superfluous citations in its application. On the other hand, failure to list known citations can delay the issue date (the date when the application is approved). Therefore, it is generally assumed that the applicant possesses the incentive to cite relevant prior art, and the patent examiner's prior art search safeguards the integrity of the process (Office of Technology Assessment 1976).

Recently, social scientists have taken advantage of patent citation data because they manifest technological similarities between inventions. Researchers have used citations to identify pairs or groups of firms have been investing in similar technologies. For example, Podolny, Stuart, and Hannan (1996) used patent cocitations (when two patents cite a common, third patent) to measure the technological overlap of pairs of organizations. In a similar vain, economists have


traced patent citation patterns to document technological knowledge spillovers between firms (Jaffe, Trajtenberg, and Henderson 1993).

The US semiconductor patents assigned to the 150 firms in the sample are used to compute technological prestige scores. Semiconductor firms routinely fIle for patent protection for their inventions in the US. The importance of patenting in the industry is underscored by the fact that the six firms awarded the greatest number of US patents during calendar 1996-IBM, Motorola, NEC, Hitachi, Canon, and Mitsubishi-all have sizable operations in semiconductors.

The US is the world's largest technology marketplace. For this reason, it is routine for non-US-based organizations to patent in this country (see Albert et al. 1991; also, note that four Japanese companies were among the six largest holders of US patent with a 1996 issue date). Thus, I collected all US semiconductor patents in 2400 distinct patent classes that contained semiconductor product, device, and design inventions. After assembling all US semiconductor patents, I then constructed detailed corporate family ownership trees for the 150 firms in the sample. Using these ownership trees, I was able to assign the patents of subsidiaries and divisions to the level of the corporate parent. Finally, after identifying the corporations that owned the 48000 patents in the database, I configured the data into a network to compute a measure of prestige for the statistical model.

To operationalize technological prestige, I follow the definition of Knoke and Burt (1983), who stated that an actor is prominent to the degree that its network position makes it visible to other actors, and it is prestigious when it is the object of relationships from other actors in a network of directed ties. Among the many measures of prestige developed in the network literature, the simplest is an actor's indegree score, which is a simple count of the number of relations directed to a focal actor.

Adopting this measure to a directed network of patent citations, a prestigious innovator would be a firm with a patent portfolio that is highly cited by other innovators (Podolny and Stuart 1995). Because patent citations are akin to building relationships between inventions, highly-cited patents are those which have been important building blocks for the efforts of other innovators (see above). In this sense, patent citations are similar in meaning to citations between journal articles: they are deference relations because they implicitly acknowledge the importance of a contribution of a competitor (Podol ny, Stuart, and Hannan 1996). Therefore, just as highly-cited papers are the origin for the prestige of a scientist, I argue that highly-cited patents engender prestige for their corporate developers. Accordingly, the technological prestige of semiconductor firm i is defined to be the proportion of all patent citations directed to its patents. Specifically:

LCjit' D. =_1_' __

,t 4, i:1! j (3)

where Dil denotes the prestige of firm i at time t, Cjil, is coded as '1' if a patent of firm j cites a patent of firm i during the interval t' (and '0' when there is no


citation), and 4' is a count of all patent citations accruing to the firms in the sample during the interval t'. Thus, Djt is the proportion of patent citations made during the interval t' which are directed to the patents in the portfolio of each firm i. The restriction i :F j is included so that self-citations do not contribute to a fIrm's prestige. I have chosen to normalize the indegree score with the total number of patent citations made in the industry to correct for changes in the total volume of citations accruing to the sampled firms over time. Because of the normalization, the prestige measure maintains a consistent meaning across time periods (it is the proportion of all patent citations that accrue to the portfolio of each firm in the sample within a fixed period of time).

Prestige levels for the firms in the sample can be expected to change over time as firms develop new inventions and as the characteristics of old ones lose their relevance for a firm's current-day initiatives. Therefore, I have used a five-year, moving window to compute technological prestige (t' in equation 3 designates the five year interval that precedes year t). Five years was chosen because this is the approximate length of the product life cycle in the semiconductor industry. Following equation 3, Texas Instrument's prestige in the year 1991 is the number of patent citations that were received by all of its inventions during the 1986-1990 interval, divided by the total number of patent citations received by all firms in the sample during the same time interval.

Control Variables A number of control variables are included in the analysis. First, firm size, measured as annual semiconductor sales, is included in the models. This variable is added because there is likely to be a positive correlation between size and prestige. Hence, without size in the regressions, any effect of prestige may be spurious because of the correlation between prestige and this omitted regressor. Second, the models include the age of firms' semiconductor operations, defined as the number of years since initial entry into semiconductors for diversified semiconductor producers and to be the time since founding for dedicated producers. Third, the models include a dummy variable coded as '1' if a firm is publicly traded.

I have also constructed a 'no patent' dummy variable, coded as '1' if a firm had no semiconductor patents before the start of a year. Based upon equation 3, firms without patents have '0' technological prestige because they are not at risk of garnering patent citations. I have included these firms in the analysis with zero prestige, but allow for an intercept adjustment for them (reflected in the coefficient on the 'no patent' dummy). It should be noted that results are comparable when firms without patents are omitted from the sample.

The final control variable in the hazard rate models is an endogenous occurrence dependence term defined as the total number of alliances formed by each firm in the sample during the previous five years. It is understood that unobserved heterogeneity across observations produces occurrence dependence in event data (Barron 1992). One strategy to control for unobserved heterogeneity is therefore to include a variable that indicates the number of times that each actor has previously experienced the event being modeled. Including the previous alliance count as a variable in the event history analysis should therefore help to control for the effects


of unobserved factors that produce variance across firms in their proclivity to form alliances.

The alliance data I possess are from 1981 to 1992, but because of the lagged alliance count variable, I have modeled alliance formations during the period from 1986 to 1991, while using the first five years of the data to construct the occurrence dependence variable. All firms in existence in 1986 are considered to be at risk of a new alliance formation on January 1 of that year. For all firms, spells have been split and covariates are updated on the first day of calendar 1987, as they are in each subsequent year. All time-varying covariates have been entered into the models as one year lags. The waiting time clock for each firm's initial spell has been set to be the amount of time which has passed since the firm formed its last alliance (prior to 1986; this is defined as the time since founding for firms that had no pre-sample alliances). Guo (1993) has demonstrated that this approach yields unbiased estimates when using the exponential model. Therefore, because of the distribution, the results are untainted by left censoring, even though some firms had formed alliances prior to the time at which I begin to observe them.

RESULTS

Table 1 reports the results from the hazard rate alliance formation models. As discussed in the 'Estimation' section, technology supply and receive license alliances are treated as 'competing risks' in the analysis. Therefore, Table 1 reports a separate set of coefficients for the effects of the variable on each of the two outcomes. Beginning with the control variables, it is no surprise to find that firms without patents ('no patent' indicator ='1 ') are far less likely to enter alliances, particularly technology supply agreements. As one would expect, the count of previous alliance formations (the occurrence dependence variable) has a positive and statistically significant4 effect on both events. Among the remaining control variables, firm size has a statistically significant effect only on the rate of entry into technology supply alliances, while the age of the firm in semiconductors has a statistically significant effect only on the rate of entry into technology receive alliances. It is interesting to note that older firms, perhaps more inert and lagging behind their younger counterparts in the adoption of the latest technological developments, have a greater proclivity to utilize the social capital of license alliances as a strategy for sourcing technology.

Turning now to the effect of technological prestige on the alliance formation rate, the results demonstrate that the variable has a large and statistically significant, positive effect on the rate at which a firm receives technology from its competitors (the magnitude of the effect through the 95th percentile of the distribution of the variable is plotted in Figure 1). In contrast, while the effect of prestige on the rate of supplying technology is also positive, the magnitude of the coefficient is small and it is not statistically different from zero. This pattern of results is consistent with the hypothesized status transference process. By the signing of a license contract, a prestigious innovator publicly certifies another organization's technology. If (as I have hypothesized) firms attempt to affiliate their endeavors with prestigious firms


~ ~

.s 'C)

~ C\,

':::1 "3 ::E

Table 1. Maximum likelihood estimates of the technology supply and receive alliance formation ratea

Receive Supply

Firm is public .4184* (.2291) .2528 (.2125)

Firm has no patents -.5828* (.2220) -.7828* (.2268)

Lagged alliance count .0237* (.0041) .0214* (.0038)

Firm age -.0122* (.0062) .0160 (.0057)

Firm sales ($8) .0039 (.0065) .0117* (.0056)

Technological prestige (DiI) .0237* (.0041) .0038 (.0033)

aAIl models include II unreported period effects and unreported calendar year effects. Receive alliances are those in which a fmn is a technology licensee and supply alliances are those in which a firm is the source of licensed technology. *p<.05.

2,5

2

1,5

0,5

0 0 0

~

Log-Likelihood Episodes Number supply alliances Number receive alliances

0 ~ N N

~ .., ..,

~ 0 ~ ~ 0 0 0 0

905.84 3127 338 409

.., ~ 0

0 0

Technological Prestige ~ ~

.,., ~

.,., ~ ~ ~ 0 0

Figure 1. Multiplier of the receive alliance rate Legend: The Figure portrays the technology receive alliance rate multiplier plotted against the level of focal fmn technological prestige. The curve is based on the parameter estimate for the prestige in the first column of Table I.


to attract attention to their undertakings, we would expect to find that prestigious firms are frequently on the receiving end of directed alliances.5 As the 'Implications' section elaborates, this finding may reflect the changing roles of up-start and prestigious innovators in high-technology marketplaces.

The statistically significant effect of prestige on receive alliances coupled with the weak effect of the variable on supply alliances demonstrates the empirical leverage that can be gained by distinguishing between alliance types in a competing risks formulation of the alliance formation process. High prestige firms are the industry'S leading innovators. As I have operationalized the variable, firms gamer prestige by making important additions to the corpus of semiconductor technology. Because high prestige firms control many of the industry' s important innovations, it might have been expected that they would frequently serve as the source of technology in asymmetric license alliances. Indeed, competence-based explanations of the alliance formation process have emphasized that the most accomplished innovators are likely to enter into alliances because they possess capabilities that other firms will attempt to access through intercorporate alliances (Eisenhardt and Schoonhoven 1996). As the results in Table 1 demonstrate, however, high prestige only augments the baseline rate of licensing technology from other organizations. This finding is in accord with the predictions of a status-based model of alliance formations, but it is inconsistent with a capabilities-based explanation of the phenomenon.

Before concluding, it is useful to present some descriptive data on the degree to which prestige asymmetries between alliance partners appear in the patterns of affiliation among semiconductor producers during the decade from the early 1980s through the early 1990s. These data bear upon the arguments in this chapter that governance considerations and the ability to exchange technology access for endorsements often create the incentive for firms at different prestige levels to form strategic coalitions. Because this suggestion stands in contrast to the prevailing thinking on the effects of status differences on the probability of inter-actor associations, I present some descriptive data to document the extent to which there have been alliances in the industry involving firms at different prestige levels.6

To explore the extent of status-based homophily in alliances between semiconductor firms, I have sorted the 150 firms in the sample into three prestige categories: low, medium, and high. To construct these categories, I simply rankordered the firms in the sample by their technological prestige scores and then trifurcated the sample into three groups of equal size (using cut-points at the 33rd and 67 th percentile of the prestige distribution to demarcate group boundaries). Table 2 presents the distribution of alliances within- and between these three status levels. The data for the Table add joint product development, technology exchange alliances, and joint ventures to the license alliances that were analyzed in Table 1. Each cell in Table 2 reports the percentage of all innovation-oriented alliances established between firms at the prestige levels denoted on the rows and on the columns of the Table. Relationships among firms of comparable status (withinprestige level ties) appear on the main diagonal of the Table because the cells on the diagonal denote the proportion of all affiliations between firms in particular status


Table 2. Proportion of alliances between finns in different prestige categoriesa

Prestige Level Low Medium High

Low 0.02 0.07 0.27

Medium 0.07 0.02 0.21

High 0.27 0.21 0.41

a The Table reports affiliation patterns based upon 1000 strategic alliances between semiconductor firms during the period from 1987 to 1992.

categories. Alliances between organizations at different levels of prestige fall in the off-diagonal cells.

A number of patterns appear in Table 2. First, at least one high prestige firm was involved in the vast majority of the strategic coalitions in the industry: only 11 percent of the alliances were between two low, one low and one medium, or two medium prestige firms. In other words: 89% of the alliances included at least one high-prestige firm. Second, a substantial proportion (27 percent) of the alliances were established between firms with significantly different levels of prestige (one low and one high prestige firm). Third, less than half of the alliances occur between firms in the same prestige category. Thus, while high prestige firms were the dominant players in the industry's alliance network-a fact that alone would seem to implicate technological prestige as a consideration in the selection of alliance partners-there were a significant number of coalitions between firms at different prestige levels. While the Table 2 proportions are only suggestive, it does appear that alliances were neither randomly spread across prestige levels nor did they occur exclusively between firms of comparable prestige.

IMPLICATIONS

The results have demonstrated that technologically prestigious firms accrue alliance capital at the greatest rate: they frequently utilize license alliances to access the technological assets of their competitors. While I have only looked at the effect of technological prestige on the accrual of alliance capital in a population of hightechnology organizations, it is likely that the findings are representative of a general association between status and access. I anticipate that the prestigious members of most populations will be able to gain access to the endeavors of other members of their community, due to the benefits that derive from associating with high status actors.

One of the traditional advantages of prestige in technology-intensive industries has been that those who have possessed it-firms such as AT&T and IBM-were able to attract top-quality engineers and scientists. Recent developments, however, may have threatened this benefit and promise to alter corporate roles in hightechnology industries. The strong market for initial public offerings and the opening up of new financing options for startup companies are among the developments that have brought sometimes extreme wealth to managerial and scientific talent who were early backers of successful new ventures. With sufficient private financing, young companies can offer salaries near to those paid by established firms. Of course, they can also offer equity, stock options and their concomitant: an outside


chance to make a small fortune. Because of these changes, prestigious companies now face greater competition for top-quality human resources.

I speculate that the profusion of startup firms in high-technology industries will mean that the social capital benefits of prestige, rather than dissipating, will appear in the unique capability of the prestigious enterprise to certify the initiatives of young or unknown organizations (see Stuart 1998a; Stuart, Hoang, and Hybels 1999). It is because of their ability to certify the initiatives of other organizations that high prestige firm will gain access to the endeavors of others. The correspondence between prestige and access implies that prestigious firms enjoy a powerful positional advantage.

NOTES

I. I refer to license alliances as asymmetric deals because technology travels in only one direction: from the licensee to the licenser. 2. Elsewhere (Stuart 1998a, 1998b) I discuss the reasons why the semiconductor industry is an appropriate context to study alliances and I report on the incidence of alliance activity in the business. 3. I have chosen to model the rate of alliance formation, rather than to model the probability that pairs of organizations form an alliance in some arbitrary time window (i.e., a panel dyad model; see Leenders 1995b, 1999). Treating the organization as the unit of analysis avoids issues of network autocorrelation (a problem of dyad models given currently available estimation procedures), but at the same time does not allow one to test hypotheses about which other organizations serve as a focal firm's alliance partners. Although I have estimated dyad models and find a prestige effect somewhat analogous to the results that I will report in this chapter (Stuart 1998a), the prediction is really about the effect of firm-level status on the rate of firm-level alliance formations, independent of the identity of alliance partners. 4. Statements on statistical significance refer to a .05 level. 5. Given the modeling strategy, it is possible for prestige to have a positive effect on both types of alliances, a negative effect on both types of alliances, or any combination thereof. 6. Podolny (1994), generalizing the argument that inter-individual as well as inter-corporate exchange relationships tend to be among demographically similar actors, asserts that high status organizations will exercise exclusivity in their selection of exchange partners and transact only with alters of similar status, particularly in uncertain situations.

•

Networks and Knowledge Production: Collaboration and Patenting in Biotechnology

ABSTRACT

21 Laurel Smith-Doerr Jason Owen-Smith Kenneth W. Koput

Walter W. Powell

We examine the link between social capital and intellectual output in the context of the biotechnology industry, a knowledge-intensive and expanding field. Due to a range of economic, social, and scientific factors, interorganizational collaboration is commonplace in the industry. Staying on top of fast-breaking research developments is critical, thus patents are an important indicator of intellectual output and property. But patents also provide a signal to organizational partners that a dedicated biotechnology firm (DBF) is a worthy collaborator, in effect providing a basis for building collaborative capital. We argue that collaborative capital will lead to patents and that this intellectual capital leads to subsequent collaboration, finding support for these claims. More centrally connected DBFs subsequently produce more patents; and DBFs with more patents subsequently become more centrally connected. Additionally, we look at the types of patents held, focusing on the effects of principle (the most intellectually broad) patents. DBFs with more principle patents become more centrally connected, and come to have more total patents. Firms rich in collaborative capital have greater opportunities for producing ideas with potential for timely payoff. We suggest that patents are a generative form of intellectual capital which provide scientific visibility that further enhances a firm's collaborative social capital.

INTRODUCTION

We examine the link between social capital and intellectual output in the context of the biotechnology industry, a knowledge-intensive and expanding field in which interorganizational collaborations are among a firm's most important activities. As

Collaboration and Patenting in Biotechnology - 391

an illustration, in public records such as annual reports and websites, biotechnology firms commonly list their most prominent collaborations with external organizations. For example, Biogen, a leading dedicated biotechnology firm, lists nine drugs at various stages in its research pipeline on its website (http://www.biogen.comI997). Four of these major projects are under joint development with other biotechnology firms. Research and development (R&D) collaborations are but one form of alliance, albeit a critical one, in an industry where firms manage interorganizational relationships for a wide range of organizational tasks, from financing to clinical trials to marketing and distribution (Powell, Koput, and Smith-Doerr 1996). Biogen, our exemplar above, has more than 50 formal ties, as of 1997 (Bioscan 1997). Of course, counts of formal collaborations exclude the many informal arrangements that are even more commonplace in this field. Thus, because the biotechnology industry has a strong basic science core and numerous collaborative arrangements, it provides an ideal setting in which to study the relationship between social and intellectual capital. We examine the effects of collaborative social capital on the intellectual capital in patenting, then the subsequent effects of patenting on social capital among biotech firms.

In the biotech industry, we have found evidence that network centrality is closely related to productivity (Powell, Koput, and Smith-Doerr 1996). The most centrally-connected firms have the highest market values and largest R&D budgets, have been the first to bring new drugs to market, and are the most visible in the scientific community. While network centrality brings many organizational benefits, here we are concerned with capturing the intellectual returns. In high-tech industries, patents are often used to indicate an organization's intellectual prowess. Our focus, then, is on the intellectual returns of social structure. That is, how does network centrality influence the assignment of patents? We begin with a discussion of research on collaboration and patenting, briefly reviewing the relationship between social capital and network position. Then, we characterize patenting as both an intellectual and organizational activity, leading us to formal hypotheses about the connection between centrality and patents. We then describe our data and methods and present results. In the final section, we discuss the implications for how organizations learn to develop and use social capital.

COLLABORATION AND SOCIAL CAPITAL

Networks and social capital are closely intertwined concepts. I Social capital has been characterized by Burt (1992, 1997) as the ability to exploit vacant positions in social networks, termed 'structural holes,' or potential relationships between nonredundant contacts. Those rich in social capital are commonly found at the center of social networks. Thus, network centrality provides an observable correlate of social capital.

In studies of inter-organizational relations, social capital has at least three variants? First, an individual may be a member of more than one organization. To the extent that these memberships link otherwise unconnected organizations, social capital would accrue to the individual (and the organizations) as a result. Interlocks among corporate boards of directors are a primary example of this form of interorganizational social capital. Second, individual members of organizations can


have social ties to members of other organizations, arising through friendship or common membership in trade associations or social clubs. These ties often facilitate goal attainment for these individuals. Third, organizations can have formal, contractual agreements to engage in collaborative activity.

The third variant of social capital, arising from formal organization-toorganization ties, is our concern here. In developing arguments about social capital associated with formal interorganizational relationships, we stress that organizations are not typically born with social capital on the basis of their parentage or bloodline (aside from that due to individual founders). Instead, firms must learn to manage relationships, build reputations, and ally with high-status partners by developing routines for collaborating with other organizations (see Omta and Van Rossum, this volume). For this reason, we refer to this variant of social capital as an organization's collaborative capital.

The collaborative capital reflected in a firm's central location in a network of organizations goes beyond the straight exchange advantage of having many partners (who in tum also have many partners). Centrality not only affords access to existing partners, but creates visibility and the opportunity to benefit from others' relationships. Firms steeped in collaborative capital are able to combine partners and resources in novel ways, fostering inclusivity and widening avenues of competition (Koput, Smith-Doerr, and Powell 1997). In industries where technological innovation is paramount, such as biotechnology, science-based start-up firms tum to collaborations to assemble a broad array of skills and resources needed to sustain their R&D efforts. In so doing, biotechnology firms gain experience at managing interorganizational relations that can be used to support a diverse portfolio of collaborative activities. Firms that utilize their initial R&D alliances as an admission ticket to the field's network, drawing on their experience to exploit diverse opportunities for collaboration, subsequently become more centrally-connected (Powell, Koput, and Smith-Doerr 1996), an effect magnified by the eliteness of a firm's partners (Koput, Powell, and Smith-Doerr 1998). Once centrally-placed in an interorganizational network, a firm is positioned to garner knowledge spillovers, generating further collaborative research opportunities. Hence, the more a firm learns how to position itself centrally in industry networks, increasing its collaborative capacity, the better positioned it is to generate patents.

PATENTS AS AN ADMISSION TICKET TO INTERORGANIZATIONAL COLLABORATION

In this section we describe what patents are, explain why centrally-connected firms in the biotech industry are more likely to patent, and argue that patents provide intellectual capital leading to further interorganizational collaboration.

What is a Patent? In the United States, patent protection for inventions is guaranteed by the Constitution. Article I, section 8, clause 8 states:

Congress shall have the power to promote the progress of science and the useful arts by securing, for a limited time, to authors and inventors, the exclusive right to their respective writings or discoveries.


The patent system in the u.s. is based on a policy that promotes the growth of technological knowledge and its public use by rewarding inventors for disclosing discoveries. The reward takes the form of exclusive rights to a discovery for 17 years. Patent rights were designed to insure that useful innovations would be transferred to the public domain despite considerable economic incentives to maintain secrecy.

To be patented an innovation must pass through a rigorous examination process, usually lasting at least 18 months. Patentable discoveries must pass three critical tests: they must be useful, novel, and non-obvious to 'a person of ordinary skill in the art' (Gregory et al. 1994). In the case of biotechnology, a person of ordinary skill would hold a Ph.D. in molecular biology or a related field and be practicing in that field at the time the innovation was developed. Such a criterion suggests that an innovation cannot be patented, even if it was not mentioned anywhere in prior art, if an ordinary practitioner could have arrived at the innovation by simply combining existing ideas at the time it was developed (Gregory et aI. 1994: 28). In order to meet the novelty criterion, an innovation must not have been 'anticipated' by public knowledge or use, description in print, patents outside the United States, or description in another U.S. patent application. The United States differs from other nations in following a 'first to invent' patent policy (Kinston 1992), which means that an innovation may not have been anticipated or invented by another individual prior to the patent applicant's discovery.

Drafting the patent application requires both intimate knowledge of the idea and of the prior art. The claims section of the specification sets forth the scope of protections claimed in the patent, defining the rights of the inventor. In essence, a claim is a one sentence definition of the invention (or specific aspects of the invention). 'An unauthorized product infringes a patent if it includes the equivalent of each and every element of any of the claims' (Schecter 1995: 71). The claims section and no other part of the specification is consulted to determine whether a patent has been infringed. Patent examiners may reject any or all of the claims in an application and the scope of protection afforded an innovation is defined solely by the claims that make it into the final patent. Writing claims is a strategic process by which an inventor or his representative attempts to draft claims that will afford the widest scope of protection possible while keeping each individual claim simple enough that infringement could be proven. Drafting a patent, particularly the claims section, has important implications for the success or failure of the application.

The Importance of Collaboration to Patents A wide range of competencies and knowledge, both technical and institutional, must be brought to bear in drafting a viable application. Undertaking research with the aim of patenting implies both a knowledge of the patent process and of the relevant literature and patent classes. Not only is it a challenge to determine what claims will avoid infringement challenges, but there is an anticipatory decision to be made about what to study in the first place. These decisions require a firm to be abreast of the latest research being conducted by others in the field. Centrally-connected firms, we argue, are best positioned to frame their findings in light of others' work, and to choose research projects that mark sufficient departures from existing work.


The patent-examining procedure also holds pitfalls for the inexperienced. The discovery of relevant prior art during an application requires a response by the applicant. Conflicts over priority of invention, which are really claims about infringement, take place during the examination process. The decisions of examiners can be appealed but knowledge of when and how to make the appeal is necessary. Patent applications provide no actual protection, but the examination process can be expedited at the request of an inventor in cases where infringement might occur before an innovation is patented. Each of these possibilities, and many others, can further the scope of protections guaranteed by a patent and mean the difference between a successful or unsuccessful application.

Thus, once an application is filed, a firm needs an array of competencies, beyond just the scientific, to successfully navigate the patent examination process. One way firms learn how to patent is by patenting a lot. But firms do not patent frivolously-it simply costs too much. The examination process generally takes at least 18 months and the cost of the application itself can run into the tens of thousands of dollars (Emanuel 1995: 195). Given the difficulty of establishing that an innovation is non-obvious, novel, and useful, as well as meeting the cost of the process, there should be significant returns to knowledge of the patenting process. Powell, Koput, and Smith-Doerr (1996) argue that one reason that ties to large pharmaceutical firms are beneficial for small dedicated biotechnology firms (DBFs) is that those ties allow the DBF to take advantage of the pharmaceutical's hard-won knowledge of the U.S. Food & Drug Administration's approval processes. In much the same way, a biotechnology firm turns to partners for accumulated knowledge of the patent examination process and patent races more generally.

We stress that we are not arguing that a firm with more overall ties or more R&D alliances will apply for more patents. We contend that firms with greater collaborative capital, as evidenced by their network centrality, will subsequently obtain more patents in a timely manner, due to their deeper insights into the knowledge-base of the field and their ability to combine and utilize a broad configuration of partners' expertise in choosing promising lines of research and framing patentable claims. Put formally, we propose that collaborative capital is a resource that builds up through participation in networks, which enables a firm to patent more: HI: The greater a firm's network centrality in a given year (controlling for its

overall number of ties and R&D alliances), the more patents it will obtain in the subsequent year.

Patents as Intellectual Capital Social scientists estimate a patent's value in at least three ways. Patents can be conceived of as: 1) indicators of an unmeasurable variable, 'the accretion of economically valuable knowledge' (Pakes and Griliches 1980: 377); 2) a set of property rights to that knowledge (Griliches 1990); or 3) a unique form of intellectual capital (Grindley and Teece 1997; Podolny and Page 1998). The first two conceptions imply a static view of patents as valuable commodities that firms hold; thus, a patent's value would decrease with use. The third conception takes a more dynamic approach. While patents are still valuable commodities and the


carriers of exclusive proprietary rights, they are also a generative form of intellectual capital that is distinct from the collaborative capital of network centrality. Under this view, patents are valuable because they can be traded. As with other forms of social capital, a patent's value increases with use.

Griliches (1990: 1672) reminds us that it is useful to distinguish between the value of a patent and of patent rights. The distinction is conceptually important. While many economists interested in innovation and technological change are concerned with placing a value on individual patents, common measures of that value depend on patent rights (i.e. Mansfield 1984; Lanjouw 1993; Lanjouw et al. 1996). However, patent law does not confer the right to make, use, or sell a patented innovation. Patents only confer the right to exclude rivals from using the innovation, even if that use is an outcome of independent R&D (Gregory et al. 1994: 8). Thus, the important impact of a patent may not be in the creation of a monopoly on its use but in the implied ability to affect the choices made by rival firms (Waterson 1990).

If a patent is defined as the right to extract rents from a new idea, then the value of a patent is simply a function of its variable maintenance costs and rate of return. Seen in this light, the decision to secure and maintain a patent is based solely on its economic benefits (pakes and Schankerman 1984; Lerner 1995; Lanjouw et al. 1996). Consequently, patents would be viewed as relatively static measures of innovative activity, or as the outcomes of zero-sum innovation races (Dasgupta and David 1987, 1994). Firms would shy away from collaboration in order to secure the exclusive award of intellectual rights. In this view, patents are associated with the restriction of proprietary information, and thus are assumed to deter, rather than foster ongoing collaboration between firms.

Furthermore, under this property-focused perspective, patents can be 'invented around,' infringed, or invalidated (Mansfield 1984; Meurer 1989; Gallini 1992). Meurer (1989: 77) suggests that patents are 'lottery tickets' whose value is uncertain. Patents also 'spill over' technologically and economically from the firm that holds them to its competitors. Jaffe defines 'technological spillover' as the influence of other firms' R&D on a firm's R&D program (Jaffe 1986: 989) and estimates that every million dollar R&D investment by a firm has a return of two patents to that firm but also a return of .06 patents to rivals. To remain valuable under this conception, patents must be held and closely guarded. When they are used publicly their value decreases because the chance that rivals will invent around them increases (Gallini 1992; Mansfield 1984). In industries where technological knowledge 'spills over' and rivals wait to invent around or challenge patents, a 'lottery ticket's' value is predicated on a firm's ability to appropriate rents from an innovation while keeping it from others.

We take a different perspective on patents, however. Recent analyses of the patenting activities of universities (Slaughter and Leslie 1997; Powell and OwenSmith 1998) and its effects on academic culture (Etzkowitz and Webster 1995; Packer and Webster 1996) emphasize instead the mUltiple uses and complex ramifications of patenting. Rather than being solely the end product of an organization's research efforts, patenting may also spur interorganizational collaboration.


Patents represent not only economic value, but the intellectual competencies and capabilities of a firm. As such, they may serve as advertisements to potential collaborators. Dasgupta and David (1987: 533) make this point when noting the near impossibility of transferring tacit technical knowledge through patents: ' ... the purpose of filing the patent may be less that of seeking to deter imitators than of signaling the availability of trade secrets for sale by the patentor.' We take their claim a step further, arguing that patenting is a simultaneously backward- and forward-looking activity. In addition to securing rents from completed research, a firm's patents also advertise technical skills and areas of expertise to potential partners. When a firm obtains more patents, it has more assets to trade or license. Patent activity forms a key strand in the multiplex and ongoing web of relations that characterize the biotechnology industry.

Grindley and Teece (1997) analyze patenting and licensing practices in several high-technology industries, employing an intellectual capital view of patents. They emphasize that rents can be appropriated from patents by licensing-trading the right to infringe on a patent for royalty payments-and cross licensing-trading the use of patented innovations for equivalent access to another firm's patent portfolio. Strong patent portfolios both serve as advertisements-firms with strong portfolios are more appealing partners than firms with weak portfolios-and provide access, through cross-licensing, to innovations that might have remained unavailable otherwise. A compelling quote from IBM's Assistant General Counsel supports this idea: 'You get value from patents in two ways, through fees and through licensing negotiations that give IBM access to other patents. Access is far more valuable to IBM than the fees it receives from its 9000 active [U.S.] patents. There is no direct calculation of this value, but it is many times larger than the fee income, perhaps an order of magnitude larger.' (Grindley and Teece 1997: 15)

Note two points about this passage. First, IBM is not afraid that rivals will 'invent around' their patents. Their stance is not at all exclusionary. Second, the strength and size of the firm's patent portfolio allows them to leverage access to patents they need in cross-licensing negotiations. Here patents are valuable because they can be traded for access to more patents. Access to the same patents can be traded numerous times to multiple partners. Their value increases with use. Furthermore, access to others' portfolios becomes a basis for further R&D, which results in more patents making IBM a more appealing partner and opening avenues to an even wider range of patents.

The fundamental difference between intellectual capital and intellectual property views of patents stems from conceptions of their value. Intellectual property views see patents as things to hold because their value decreases with use. In contrast, an intellectual capital view sees patents as advertisements and things to trade. Under this conception a patent's value increases with use. Where intellectual property views highlight appropriability and proprietary rights we argue that licensing, cross-licensing, and collaboration are the ultimate sources of patents' value. An intellectual capital view of patenting links network connections and collaborative capital with innovative R&D and further patenting. Thus we contend that not only do network ties foster intellectual output, but in tum, network position is enhanced as a result of extensive patenting:


H2: The more patents a firm obtains in a given year, the more centrally connected the firm subsequently becomes (controlling for the overall number of ties and R&D alliances).

Our view that patents are generative and lead to increased firm centrality in networks may apply more to certain types of patents than others. Following Merges and Nelson (1990), we recognize that patents vary significantly in the scope of protection they claim. Thus we differentiate between prospect-defining patents and patents on relatively discrete inventions. We attempt to capture a 'topography of prospects' (Merges and Nelson 1990) by categorizing patents as principles, processes, or products. Principle patents are those of the broadest scope, such as Cetus' early patent for polymerase chain reaction (PCR) amplification. Rabinow (1996) characterizes the development of peR at Cetus as a bricolage of multiple competencies, technologies, and scientists. Kary Mullis received a Nobel Prize largely on the strength of his work at Cetus on PCR, demonstrating the importance of that principle patent. Process patents protect rights to particular uses of items, such as the use of specific reagents or steps in the treatment of a patient. Processes are considered less broad than principles but more broad than products, the most narrow type of patent (e.g., Alza's patent on an innovative pipette tip). To assess the effects of these three different types of patents, we measure the number of each patent type held by a firm.

Of the three types of patents, we expect that the number of principle patents will be a stronger predictor of future collaborations and patent activity. Principle patents, as prospect defining, may be characterized as having a larger tacit component than more narrow patents. The difficulty in transferring tacit knowledge through patents, suggested by Dasgupta and David (1987, 1994), implies that principle patents may advertise a firm's capacity for broader and more innovative know-how than do other kinds of patents. The breadth of protections claimed by principle patents also suggests that more diverse external organizations will seek to pursue collaborations with the holder of the principle patent. And in an industry that maintains strong reputational affinities with academic science (Powell and Owen-Smith 1998; SmithDoerr 1997), principle patents may have a greater effect on centrality because of the scientific credibility provided by principles relative to other patents. Thus, we hypothesize: 93: The more principle patents a firm holds in a given year (controlling for its

overall number of ties and R&D alliances), the more centrally connected the firm becomes in the subsequent year.

And because we propose that network centrality leads to later patenting activity (from HI), we further hypothesize that: 94: The more principle patents a firm holds in a given year (controlling for its

overall number of ties and R&D alliances), the more total patents it will hold in the subsequent year.


DATA AND METHODS

Data Sources Our data cover dedicated biotechnology firms (DBFs) over the period 1988-1996. DBFs are defined as independently-held, for-profit firms involved in human therapeutic and diagnostic applications of biotechnology. We do not include large pharmaceutical corporations, international multi-business enterprises, agricultural or veterinary biotechnology, or government or non-profit research institutes. These organizations enter our database as partners that collaborate with DBFs. The restricted nature of the firms included reflects our effort to assess the activities of dedicated, independent firms in the most research-intensive sector of the field.

The data on firms and interorganizational agreements are taken from Bioscan, an industry publication that reports on firms and the formal agreements they are involved in. Bioscan covers nearly the entire population of dedicated biotechnology firms in existence between 1988 and 1996, though sometimes it does not pick up newly-formed, privately-held companies with few employees. By having nine years of data, we can, in almost all cases, recapture such missing firms by tracing them back from when they first appeared in Bioscan to their actual year of founding. Bioscan initially published complete firm activity only in its April supplement; we continue to use the data from April for consistency. Patent data are extracted from CASSIS for the years 1987-1996. CASSIS is a government document, made available on CD ROM by the U.S. Patent and Trademark Office, listing patent activity in all scientific and technological areas. We sampled patents in CASSIS by recording those assigned to firms on our list of DBFs. This method captures all patents in which our DBF sample formally holds an interest, and allowed us to develop a relational dataset that contains patent-level data for every DBF in our records.3

By gathering patent data from the Assignment files (CASSIS/ASSIGN), we focus on information about the organizations assigned and maintaining patents, rather than on patent classifications (from CASSISIBIB). Because we begin with a list of biotech firms, we assume that all their relevant patents fall in the area of human therapeutic and diagnostic applications, and related instrumentation. Information from CASSIS/ASSIGN reports was coded for characteristics of assignees, assignors (inventors), and of the patents themselves. ASSIGN includes data on the assignment of inventor's interest in patented innovations to other individuals or organizations. In choosing assignment data, we may miss patents retained by individual scientists, which are not assigned to their biotech firm employer. This potential problem is not a key concern because its effect would mean we undercount patent activity, thus weakening our findings rather than artificially inflating them. Our control of fixedfirm effects insures that a particular firm with less of a propensity to obtain patent assignments in the organization's name will not unduly influence results. We suspect that if an idea is potentially valuable enough, a firm will file for a patent. This data collection strategy captures every patent formally assigned to a biotech firm. Our analyses focus on industry-level networks of firms, rather than on networks of individual scientists employed by firms. Another potential problem with patent assignment data is that firms may buy patent assignments from others rather than doing the science in-house; but our suspicion is that this activity is rare in biotechnology because of the emphasis on maintaining scientific credibility in the


industry (Smith-Doerr 1997). Moreover, if a firm becomes capable of knowing where and how to acquire important patents because it is well-connected in the industry, then the essence of our argument remains supported.

Measures Patent productivity is measured by the number of patents granted to a DBF in each year, 1987-1996. While a simple patent count is not a perfect measure of an organization's intellectual output, it is a widely accepted proxy (Schmookler 1966; Griliches 1990; Trajtenberg 1990). In their chapter in this book, Han and Breiger classify volume as one dimension of social capital. Their focus is on the volume of ties between organizational dyads; relatedly, we argue that the volume of patent activity is an important dimension of intellectual capital in the biotechnology industry. Other scholars argue for measurements of the 'value' or 'quality' of patents (Lanjouw 1993; Lanjouw et al. 1996), but our focus on the generative aspects of patent activity allows us to avoid this thorny data issue. By measuring the volume of biotech firm patent activity we emphasize the 'advertising' role patents can play in attracting potential collaborators. And in de-emphasizing patent economic value and technological distance, our analysis avoids the necessity for either renewal data (Lanjouw et al 1996) or co-citation4 and utility class data (Podolny, Stuart, and Hannan 1996). In addition to volume, or number of patents in each year, recall that we measure type of patent to capture variation in the effects of different types of biotechnology patents. Above we described three types of patents: principles, processes, and products. Patents with the broadest and most open prospects are principles, while patents with the narrowest and most specific prospects are products. The intermediate category, processes, captures patents that protect specific methodological or therapeutic processes.

In computing centrality, we need to account for the fact that we do not have a dosed network. In this respect, our measure of interfirm networks is somewhat unconventional. We wished to examine the structure of the network linking our sample of DBFs, but we need to define a closed set of firms to compute measures of connectivity. Yet many of the ties that structure the field involve parties outside the scope of our definition of a DBF-the overall universe of partners is open, diverse, and expanding. We counted a connection between two DBFs: 1) when there was a direct tie (degree one), and 2) when the DBFs were linked (at degree distance two) through a common partner, to capture the information that may flow between DBFs through a non-DBF partner. In measuring centrality we do not differentiate among connections involving different business functions, i.e., R&D, investment, clinical trials, manufacturing, licensing, distribution, joint venture, and complex ties (see Powell, Koput, and Smith-Doerr 1996 for a more detailed discussion of data coding). The various types of collaborative activities each playa comparable role in creating a firm's overall set of relationships. Measures of central connectivity were computed using UClNET IV (Borgatti, Everett, and Freeman 1992). Centrality is a measure of how well connected, or active, a firm is in the overall network. We gauged centrality of a firm locally rather than globally, in network parlance, such that a firm's centrality is the number of other firms connected to that firm, ignoring how well those partners are connected.


Included in the models are other variables that might be expected to have statistically significant effects on patenting and centrality. These controls include size (number of employees), firm age, and other network measures aside from centrality. We controlled for alternative explanations that involve firm age or size as predictors, rather than as outcomes, of network behavior. Age appears as a predictor in ecological and life-cycle theories of organization, while greater size, indicating a more extensive hierarchy, is seen as an alternative to alliances in the transaction-cost literature, and as an outcome of learning in the network literature. Age and size are not logged; skewness is shared by our dependent variables (see methods section for discussion of skewness).

Other measures of network activity include number of R&D ties, network portfolio diversity, and collaborative R&D experience. The number of research and development ties a firm has captures the extent of its involvement in the core activities of the industry, particularly activities driven by discovery efforts. As noted earlier (Powell, Koput, and Smith-Doerr 1996), R&D has a unique status in theories of organizational learning, and so we treat it separately. The range of ties that a firm is engaged in at any given time reflects a firm's portfolio of collaborative activities. Network portfolio diversity is computed for each firm in each year using Blau's index of heterogeneity (Blau 1977).s Collaborative experience at time twas measured as the time since inception of a firm's first alliance. Descriptive statistics for each variable used in the analyses may be found in appendix Table AI.

Methods Our data consist of nine years of cross-sectional records of firm-level variables. To test the predictions, we used a panel-regression model. The selection of this technique involves two primary theoretical considerations and the need to address a number of statistical issues that stem from these concerns. The first theoretical consideration is that social capital resides within firms and develops over time. We argue that while social capital accrues through network relationships, firms are both the actors and the recipients of the skills and expertise that social capital brings. To eliminate any spurious effects due to differences between firms, we included fixedfirm effects by entering a dummy variable for each firm. Thus we avoid, for example, potential effects resulting from firms perhaps having more R&D ties and more diverse portfolios because of patent activity prior to the period investigated. We use fixed rather than random effects because we essentially have the population of dedicated biotech firms over our observed time period, and not a random sample.6

We wanted to estimate a dynamic model, in which the explanatory variables are lagged one year to allow us to make causal inferences. But an important factor to consider is that changes over time within firms will result in autocorrelated errors and may bias estimates of the parameters in which we are most interested. One way of breaking the correlation over time, so as not to overestimate the effects of our hypothesized independent variables, is to include a lagged dependent variable, Yi,t-l' as a predictor. When a lagged dependent variable appears as an explanatory

variable, however, the fixed effects estimator of the parameter on the lagged dependent variable may not be consistent, because our dataset involves a large


number of firms observed over a short period of time. We used an instrumental variables estimator (Hsiao 1986: Ch. 2) to remedy this potential problem.

The dynamics of social capital accumulation involve the co-evolution of firms and networks. This process leads to an additional source of statistical nonindependence across our observations: For each firm we measure properties of its position in the industry's network. but each firm's position can only be assessed in terms of the network positions of others. For instance. a particular firm might have a high centrality score because a number of outside firms act as partners to multiple firms in the network. Hence. we need to control for effects that vary over time but are constant across firms. such as the overall number of outside partners. the density of the industry's network. government budgets for medical research. or the economic circumstances of pharmaceutical companies. To do so. we included fixedyear effects: a dummy variable for each year.

In using what is termed a fixed-effects specification (Judge 1985) for both the firm and year controls. with a lagged dependent variable as one of the predictors. the dependent variable. Yi.t' is modeled as:

J

Yi,l = ai + YI + ;"(Yi,l-t)+ L f3 j (Xi,t-t,j)+ Ei,l . j=t

In this equation. a. is the effect of firm i: i=I .... N; Y is the effect of year t: 1=1 .... 9; I t

f3. is the within-firm slope for x .• pooled over all firms and years; and E. is a J J I,t

normally distributed error term. The strict assumptions of the normal regression model are violated. because our

primary dependent variables are skewed and only approximately continuous. The number of patents and centrality can only take on non-negative integer values. Their ranges are fairly large (approximately 25 and 130. respectively). however. making the continuity approximation reasonable. While the truncation and skewness are potentially problematic. these features are shared by the explanatory variables. allowing us to make the a priori working assumption of symmetric disturbances. We confirmed the validity of this assumption with diagnostic plots in a post-hoc residual analysis (not reported here). Also. correlations among predictor variables are displayed in the appendix.

We checked the robustness of our results in several ways. First. we applied a square-root transformation to the dependent variables and re-estimated the normal regression model. Second. for those dependent variables that are integer numbers. we also estimated panel models for count data. Both of the additional models confirmed the findings of the normal regressions. We report the results of the normal regressions for all variables to ease interpretation and allow comparison of effects across the models. We also conducted logistic regressions for the onset of network ties and initial patents. The goal of the logistic models was to tease out the temporal priority of explanatory variables in HI (network centrality) and H2 (patent obtainment). In keeping with our learning argument. we expect that collaborative capital increases success in navigating a patentable stream of research. and that patents are not required for initiating alliances. but enhance network position through a recursive feedback loop.


Table 1. Panel regression models of the effects of network centrality and patenting activity

Dependent variables at time t+ I

Explanatory variables at t Number of patents Degree Centrality

Degree Centrality .0288"'·· (.0052) .8356··· (.0285)

Number of patents .6553··· (.0273) .1375· (.0695)

Control variables at t

Total number of ties .0461·" (.0094) .2791·" (.0659)

Number of R&D ties .0071 (.0299) .3370··· (.1100)

Diversity of network activity .0251 (.2473) .0843··· (.0173)

Collaborative experience .1388· (.0629) .2486 (.1506)

Size .0735·· (.0221) .1945 (.1598)

Age -.0074 (.0101) -.3383·" (.0709)

Full R2 (within-firm R2) .6372 (.3535) .7871 (.2031)

N 993 992

·p<.05, ··p<.ool , ···p<.OOOI Note: all models include fixed firm effects (dummy variables for firm ID), and dummy variables for year effects as controls.

RESULTS

The results of regression models are displayed in Tables 1-3, which present unstandardized regression coefficients, with standard errors in parentheses.

Table 1 presents results of panel-regression models addressing the hypotheses that greater centrality leads to more patents (HI), and that more patents lead to greater centrality (H2). Both hypotheses are supported. Centrality at time t increases the number of patents obtained in time t+ 1, and receiving more patents at t leads to increased centrality at 1+ 1. Note that the statistically significant effects 7 of centrality and patenting occur while controlling for the total number of ties held by a firm, as well as number of R&D ties, diversity of collaborative activity, experience with network ties, size, and age. Recall that the models in Table 1 also control for fixedfirm and year effects, thus ensuring that unmeasured firm or time characteristics are not skewing the results.

While we do not look at controls as dependent variables,8 we found some interesting implications in comparing the effects of control variables on patenting and centrality. Experience and size have statistically significant effects on patents, but not on centrality. Age is negatively related to centrality so that more central firms are younger, a finding consistent with our earlier work (Powell, Koput, and Smith-Doerr 1996). Holding constant earlier patent activity, other network measures, size, and age, centrality predicts subsequent patenting at the .0001 level. The effects of patenting on later centrality, taking controls into account, are also notable.


R&D Ties

Centrality

Network Diversity

Figure 1. Visual representation of the relationship between network measures and patent activity

The effects of network control variables speak to a more complex relationship between network activity and patenting than might be expected. The number of research and development ties does not predict patents at a later time, but does predict increased centrality. Likewise, diversity of network activity leads to increased centrality, but is not meaningfully related to patent activity. These findings suggest the relationship between variables illustrated in Figure 1.

While R&D ties and diversity of activity lead a firm to a more central position in the industry, they do not lead directly to patent activity. Prior to increased patenting, a firm needs central connectivity, which indicates that social capital from collaboration precedes the propensity to patent. Collaborative experience mediates between centrality and patent activity, in that experience has a statistically significant effect only on number of patents. Experience with ties increases patenting, but only imperceptibly affects centrality, perhaps because centrality often precedes extensive collaborative experience.

Table 2 provides further evidence of network centrality'S temporal priority to patent activity. In a set of logistic regression models, we examine whether centrality better predicts a first patent or whether patenting is a better predictor of an initial tie. Models 1 and 2 in the Table clearly show that patents scarcely predict formation of an initial tie, while network ties do, with statistical significance, predict obtaining a first patent, even controlling for the size and age of firms. But as Table 1 and Figure 1 show, patenting provides intellectual capital that enhances a firm's collaborative capabilities.

Table 2. Logistic regression models predicting initial patents, and initial network ties

Dependent variable at time t

Explanatory variables at time t-I

firm has tie(s)

firm has patent(s)

Size

Age

Chi-Squared

N

patent assigned (1) tie initiated (2)

.8223** (.3277) 5.5660**** (.3865)

1.6034**** (.1475) .7146 (.4527)

.2932**** (.0547) .1343 (.0842)

-.0301 * (.0139) -.0630** (.0240)

266.97**"'*

1086

401.71 ****

1086 *p<0.05, **p<O.01, ***p<.OOI, ****p<.OOOI Note: models include dummy variables to control for flflD effects and year effects.


Table 3. Panel regression models including the effects of patenting scope

Dependent variables at time t+ I

Explanatory variables at t Number of patents Degree Centrality

Degree Centrality .0292*** (.0053) .8385*** (.0299)

Number of Principle patents .6203*** (.1541) .2471*** (.1153)

Number of Process patents .1348* (.0701) -.9254* (.4585)

Number of Product patents .5666*** (.0451) .6103*·* (.2952)

Control variables at t

Total number of ties .0461**· (.0094) .2791*·* (.0659)

Number of R&D ties .0071 (.0299) .3370**· (.1100)

Diversity of network activity .0251 (.2473) .0843·*· (.0173)

Collaborative experience .1388* (.0629) .2486 (.1506)

Size .0735** (.0221) .1945 (.1598)

Age -.0074 (.0101) -.3383*** (.0709)

Full R2 (within-firm R2) .6378 (.3692) 7877 (.2111)

N 993 992

·p<.OOI, ··p<.OOOI, ·**p<.OOOOI Notes: All models include fixed firm effects (dummy variables for firm 10), and dummy variables for year effects as controls. Partial models were run, but as each statistically significant variable adds to the overall R2, only full models are presented.

In fredicting number of patents, prior patents of each type have positive effects. Table 3 indicates that the effects of types of patents act additively, that is, there are no interactive or multiplicative effects. Principle patents are the best predictors of greater subsequent patent activity, and product patents have the next largest effect. Degree centrality is also predicted by greater numbers of prior principle and product patents. Process patents have a negative effect on subsequent centrality, however. This finding seems to be due to a small group of firms that have only process patents and no other types. These 'isolated' specialists also have a higher than average failure rate, underlining the importance of network centrality to biotech success.

As evident from the R2 values reported in the tables, the models are explaining much of the variance in DBFs' collaborating and patenting. Each statistically significant coefficient, when added hierarchically, adds to the model fit so that the change in R2 is statistically significant at the .05 level. For all of the results presented in tabular form, it is important to note that it is the within-firm R2 that is driving the between-firm (full R2) effects rather than vice versa. The between-firm collaborative and intellectual capital differences appear because within-firm learning occurs at different rates. Thus our contention that the firm is still an important unit of analysis even in a highly connected industry is supported by the importance of within-firm variation to the models.


DISCUSSION

All told, we expect that collaboration and patenting constitute complementary activities rather than cross-purposes in biotechnology. Of course, keep in mind that we hypothesize the importance of patents as intellectual capital in the context of a knowledge-intensive, expanding field, rather than one dominated by incumbent organizations vested in a technological paradigm half a century old. Our results show that, as hypothesized, centrality predicts subsequent patenting, and patenting predicts subsequent centrality in the biotechnology industry. When controls for other measures of network activity are entered into the relationship between centrality and patenting, an interesting wrinkle appears. In earlier analyses (Powell, Koput, and Smith-Doerr 1996), we had found that R&D ties are critical to accessing networks of learning in the industry. While R&D ties continue to affect centrality here, they are not a direct predictor of patent activity, suggesting that R&D ties may still be the admission ticket to learning races, but collaborative capital (assessed by centrality, controlling for R&D and other ties) provides the fuel for patenting activity. In order to choose research projects that can be developed to the stage of obtaining a patent, a firm appears to need relationships that place it at the center of the industry's activities, rather than just R&D ties. 10

To some extent, this effect of centrality on patents suggests the importance of other skills besides the capacity to do research in the patent process. For example, a central firm may have access to more savvy information about the state of research in the field as a whole. Our focus on data that look at patent activity in the year following a firm's centrality score suggests that a central firm is more likely to have the necessary connections to organizations with knowledge on how to get patents through the examination process in a timely manner. Thus, patent activity is a result not just of fragmented scientific knowledge and R&D, but of a broader awareness and ability to combine knowledge-from otherwise unconnected collaborative activities-about what to study and what to claim.

The results of the effects of patent types on subsequent centrality and patent activity are generally supportive of our predictions. Principle patents have the greatest magnitude of effect on centrality and further patenting. The finding that process patents have a statistically significant negative effect on centrality is somewhat puzzling, but a closer look at the data provided some initial answers. Apparently, process patents are actually more closely related to specialization than product patents, contrary to our expectations. A small group of firms whose only patents are processes have a high failure rate, and not surprisingly, low centrality. Firms that have a majority of their patents as processes also appear, based on their annual reports, to be more specialized in their project collaborations than other DBFs. But having a process patent alone does not result in less centrality-those firms with the most process patents have the most patents generally, and are centrally connected. A specialization in only process patents appears to detach a firm from core networks. We plan further analyses of this issue and will use more fine-grained coding of our process patents in future work.

To summarize, we stress the learning aspects of our argument and results. We developed an organizational aspect of social capital which we termed collaborative capital. Whereas individual social capital is often characterized as the ability to


connect otherwise disconnected social actors, collaborative capital includes the ability to combine, in novel ways, the activities of diverse parties in an interorganizational network. We assessed collaborative capital by measuring an organization's network centrality, controlling for its number of overall ties and R&D collaborations, time in the network, and the range of activities in its network portfolio. Hence, collaborative capital is distinguished from collaborative capacity, research intensity, experience, or diversity-it is an extra dimension that both combines and extends these other aspects. Collaborative capital builds over time as a firm participates in research, gains experience, develops capacity, and assembles a diverse profile of activities that moves it toward the center of the network and gives it greater access to knowledge spillovers. Once centrally-placed, collaborative capital enables an organization to create opportunities with the greatest potential for timely impact and payoff. Our results demonstrate a link between network centrality and patent obtainment, which we take as evidence of collaborative capital in action. We document that patents are a generative form of intellectual capital that, in knowledge-intensive fields such as biotechnology, once obtained bring scientific visibility to further enhance a firm's collaborative social capital.

Viewed more broadly, our results suggest that, in industries where knowledge is either expanding rapidly and/or distributed broadly, firms that adopt a view of patents as an admission ticket to an information network will be rewarded. As we have argued previously (Powell and Smith-Doerr 1994) access to knowledge and benefit-rich networks is a critical resource in an information-intensive economy. Collaborative capital, a critical component of social capital, enables organizations to expand their reach well beyond the capabilities and limitations of a single firm. Collaborative capital provides not only deeper resource pools, but opens up the possibility of combining ideas, people and resources in novel ways.

APPENDIX

Descriptive Statistics and Correlationsll

Table AI. Means, standard deviations, and range of variables analyzed

Mean Standard Minimum Maximum Number of

deviation Cases

Age 7.65 6.28 0 46 2005

Centrality 15.49 22.65 0 132 2062

Size 144 360 1 6000 1396

R&D ties 1.23 2.09 0 16 2063

Diversity .44 .28 0 .85 1562

NPatents 1.05 2.53 0 25 2733

Principle patents .06 .33 0 5 2733

Process patents .40 1.07 0 12 2733

Product patents .57 1.63 0 22 2733

Total ties 7.38 9.84 0 82 2063

Experience 3.78 3.80 0 32 2063


Table A2. Between-finn correlations of predictor variables. including patent types

Age Centrality Size R&D Diversity Total Exper. Patents Principle Process

ties ties patents patents

Age

Centrality .1574

Size .3752 .4480

R&D ties .2559 .67% .3002

Diversity .1745 .5091 .2834 .4593

Total ties .2190 .6651 .5699 .6214 .5568

Experience .3793 .2721 .3517 .2466 .1634 .2920

Patents .3049 .4812 .6054 .3680 .3044 .5877 .2216

Principle .1354 .4573 .3980 .2672 .2595 .5042 .1519 .5055

patents

Process .2918 .4393 .5994 .3119 .2992 .5788 .2097 .9044 .5270

patents

Product .2752 .3907 .4981 .3272 .2404 .4597 .1837 .9363 .3050 .7117

patents

Table A3. Within-finn correlations of predictor variables. including patent types

Age Centrality Size R&D Diversity Total Exper. Patents Principle Process

ties ties patents patents

Age

Centrality .1224

Size .0725 .1127

R&D ties .0862 .5278 .0540

Diversity .0890 .2967 .0348 .2353

Total ties .1171 .6042 .1286 .6180 .3979

Experience .2787 .3287 .1255 .1403 .1408 .2942

Patents .1547 .0730 -.0908 .0757 .0811 .1343 .0667

Principle patents .0571 -.0037 -.0l2l -.0038 .0222 .0423 .0256 .1839

Process patents .1199 .0920 -.0307 .0560 .0459 .0765 .0380 .2830 .2299

Product patents .1671 .0428 .0865 .0268 .0454 .0473 .0587 .4446 .1269 .3502

Research support provided by NSF grant #9710729. W.W. Powell and K.W. Koput, Co-P.I.'s. Direct correspondence to Laurel Smith-Doerr at: Department of Sociology. Social Sciences 400. University of Arizona. Tucson. AZ 85721. USA; or bye-mail: [email protected].


NOTES

1. For example, in the index to the Handbook of Economic Sociology (Smelser and Swedberg 1994), under social capital the reader is directed to 'see also networks.' 2. These three interorganizational definitions of social capital illustrate common usage of the concept but are not meant to constitute an exhaustive list of the fonns of social capital. See the chapters by Gabbay and Leenders and by Pennings and Lee in this volume for further definition of social capital at both individual and organization levels. 3. CASSIS data are counted at year end. Hence, between 9 and 24 months separate year t centrality and year t+ I patents, for example. (See methods section for further details on statistical issues). 4. See Stuart's chapter in this volume for an analysis of patent citation data. 5. Blau's index is computed for each finn in each year as follows. For finn i in year t, denote the

number of ties of typej as nit,j and the total number of ties aggregated over all types (j=J •..• 8) as nit'

The proportion of finn i's ties of type j, out of the total number of ties. is denoted as Pit.j and given by

P. . = n. . / n. . Each p. . is squared and then the sum is taken over all j and subtracted from I, It,} It,} It It.)

8

resulting in the index of heterogeneity. y. , as so: y . = 1-"" p2 . . II It -'-' ",)

j=1

6. DBFs are our physical population. while our statistical population is all possible realizations of the stochastic processes of patenting and collaborating-thus the use of probability statistics is warranted. 7. Statements on statistical significance refere to a .05 level. 8. Thus, the following discussion of control variable effects is meant as suggestive rather than definitive. 9. Dummy variables to indicate whether patents of each type were held or not were also run in models to predict further patenting and centrality, but had statistically nonsignificant effects. 10. Also recall that explanatory variables are only lagged one year. perhaps not allowing enough time for the measure of R&D ties to affect patent activity in the data analysis. The fact that collaborative experience is positively related to patenting would tend to support this interpretation. II. The between-finn correlations are presented for inspection. Our models are within-finn. however. The within-finn correlations are all modest, posing no coIlinearity problems.

•

Supply Network Strategy and Social Capital

ABSTRACT

22 Christine Harland

The focus ofthis chapter is on interorganizational supply networks and, specifically, the social capital and liability within them. Supply networks are sub-networks nested within interorganizational networks that include the actors, resources and activities associated with value adding processes converting resources to goods and services. To describe the nature of supply networks and the concept of supply, the literature review within the chapter traces the development of the concept through the different systems levels of supply within the firm boundary, within dyadic supplier-customer relationships, within chains of relationships, and within whole focal firm supply networks, providing examples of social capital and social liability.

A normative, rational view of supply network strategy is presented, using examples such as Benetton and Toyota to support the argument that social capital may not be an emergent feature of supply networks but may actively be pursued by firms seeking to increase competitive advantage; this view is in opposition to much of the work of the Industrial Marketing and Purchasing (IMP) group that maintains the coping nature of players in interorganization networks. Firms such as Benetton and Toyota appear to have taken on the role of supply network hub, facilitating and coordinating more effective and efficient flows of materials and information. They also appear to have focused strategically on end customers and their requirements.

Findings are also presented to support the view that social capital in networks, rather than being a tangible, measurable asset, is a function of actors' expectations and perceptions of performance and is therefore a subjective notion.


INTRODUCTION

There has been increasing interest in interorganizational networks notably since the early 1980s when original work by the Industrial Marketing and Purchasing group (IMP) provided models of interaction and a language to discuss aspects of networks (Hakansson 1982). One of the main contributions of the IMP work on networks is the actor, resource, activity model that inspired the specific notion of economic resources discussed by Araujo and Easton (this volume). However, to date there has been only limited empirical work that might provide managerial guidance. Most of the field research evident has been conducted in dyadic relationships, not networks.

Within interorganization networks exist other subnetworks; the focus of this chapter is on supply networks. Supply is defined here from the operations perspective as the value-adding transformation processes involved in the conversion of input resources to provide goods or services.

Supply processes occur within firms and between firms. Different subject areas have undertaken research into these supply activities within the firm, in supplier-customer relationships, in chains of firms, and, more recently, in interorganizational networks of firms. Findings to date at each of these system levels are presented here, with particular attention to benefits gained (social capital) and disadvantages suffered (social liability).

Before supply networks, and any consideration of them (such as social capital within them) can be addressed, the concept of supply, including its boundary and content, requires explanation and clarification, as do the observed trends in business practice that have lead to the development of the concept. The next section, therefore, examines the concept of supply.

THE CONCEPT OF SUPPLY

Research in supply integrates various existing bodies of knowledge and concepts, to form an holistic, strategic perspective of management of operations, stretching across interorganizational boundaries. Core to the concept of supply is the procurement, use, and transformation of resources to provide goods and service packages to satisfy end customers. End customers are defined as the ultimate decision makers that take a product/service differentiation decision (Harland 1996a).

The concept relates to the integration of supply activities within firms, in dyadic relationships, in chains of firms, and in interorganizational networks. These have been expressed as different systems levels of supply (Harland 1996a) as shown in Figure l. Common to all these levels is the flow of supply and the activities and decisions associated with that flow.

The development of the concept and the business practice of managing supply strategically has progressed over time through these levels whereby there is now emerging research and documented business practice relating to supply strategy in interorganizational networks; therefore, it is useful to discuss theory and practice progressively through each of these levels.

Supply Network Strategy and Social Capital - 411

LEVEL 1: Supply within the firm boundary o LEVEL 2: Supply in a dyadic relationship

LEVEL 3: Supply in an interorganizational chain

LEVEL 4: Supply in an interorganizational network

Figure 1. Levels of supply (adapted from Harland 1996a)

SUPPLY WITHIN THE FIRM BOUNDARY

The management of supply within the firm boundary involves the integration of the processes from the input side of the firm to the output side. In a manufacturing organization this requires integration across the functions dealing with physical and information flows, as shown in Figure 2.

This relates closely to the preexisting concepts of materials management (Ammer 1968; Lee and Dobler 1965) and the value chain (Porter 1985; Johnston and Lawrence 1988; Kogut 1985). Early work in supply chain management (Oliver and Webber 1982; Houlihan 1984; Stevens 1989) identified logistics benefits of reduced lead times and costs through integrating of the internal chain.

The span of the supply activities internal to the firm are decided through make-or-buy decisions that determine which activities will be vertically integrated or owned within the boundary of the firm and which will be procured through transactions with other firms. The new institutionalists theory of transactional economics (Williamson 1985; Van de Ven et al. 1975) provided a rationale for the make-or-buy decision. Ellram (1991) related these decisions specifically to supply by identifying advantages and disadvantages of vertical integration (Table 1).

A trend toward vertical disintegration has been highlighted in various industries (Thackray 1986; Porter 1987), partly to reduce risk of being locked into inappropriate technologies (Abernathy 1978; Harrigan 1983; Miles and Snow 1987) as markets changed to demand more variety of products and services. To meet this


Table 1. Supply chain management: the industrial organization perspective (derived from Ellram 1991)

Advantages of Vertical Integration

Control: uncertainty reduction of costs, quality

and quantity of supply, convergent

expectations, reduced probability of

opportunism, reduced probability of

externalities caused through dependence on

monopoly suppliers and ability to protect

important proprietory or competitive

knowledge, ease of conflict resolution.

Communication: improved coordination of

processes and greater goal congruence.

Cost: economies of scale through avoidance of

intermediaries-notably procurement, sales

promotion and distribution, process integration,

avoidance of switching transaction costs.

Disadvantages of Vertical Integration

LimitIng competition: inability to replicate

market incentives, and distortion of internal

infonnation.

Diseconomies: balancing scale economies,

inability of management to control large

organization effectively, limits on span of

control, and increased difficulty in

communication.

Risk: asset concentration, exit barriers,

perpetuation of obsolete processes, and

exaggerated synergies.

Business management

Business support

Operations management Human resource management Financial management Marketing Information systems management

Materials Purchasing Inventory Operations planning Sales-order Physical rnngt. management and control management distribution

management

Physical Goods inward Stores Operation WIP Finished goods Transport

SUPPLY • Figure 2. Internal supply cham


change in market demand, supply activities within firms increasingly concentrated on a limited, manageable set of tasks based on the principles of focus (Skinner 1969). These focused tasks were designed to meet the order winning criteria of customer groups (Hill 1985; Christopher 1992). The trend of focusing and vertical disintegration is continuing through outsourcing. Outsourcing noncore activities, such as catering, security, and information systems, has been performed by many manufacturing and service organizations. Recently, this has extended to some supply activities such as logistics, purchasing, and sales. Reve (1990) argued that only core skills with high asset specificity that significantly contribute to competitive advantage should remain as internal activities. However, the less tangible these core skills and assets are, the more difficult it is to quantify them to decide what is core. Furthermore, what is viewed as noncore today may become core in the future; the dynamics of business are such that the boundaries of core will ebb and flow (Macbeth and Ferguson 1994). Therefore, significant change has occurred in supply within firms, and this change has brought with it complexity of decision making relating to what should remain internal to the firm and what should be supplied through dyadic relationships.

SUPPLY IN DYADIC RELATIONSHIPS

Original work in industrial economics by Marshall (1923) and Coase (1937) identified the existence of alternative types of relationship to vertical integration and competitive, arm's-length market transactions. The intermediate types of relationship were later defined by Richardson (1972) and Blois (1972), differentiated by the closeness of the bonds between actors in terms of equity and commitment. The shift to relationship rather than vertical integration has lead to the industrial organization and contract view of the firm as a nexus of contracts (Aoki, Gustafsson, and Williamson 1990; Cox 1997). Relating specifically to supply, Christopher (1992) and Ellram (1991) positioned supply chain management as the management of these intermediate types of relationship.

Authors and practitioners from many different disciplines and functions have highlighted the increasing dependence on relationships with suppliers (see, for example, Sabel et al. 1987; Christopher 1992; Slack 1991; Schonberger 1986). Closer, longer-term relationships are evident in some industries, reported notably in the Japanese automotive industry (Lamming 1993; Womack et al. 1990), the Japanese textile industry (Dore 1983), craft-based Italian industries (Lorenzoni and Ornati 1988), and various Swedish manufacturing industries (Hakansson 1987). These closer relationships were enabled through the movement away from multi-sourced, adversarial trading and toward single or dual sourcing. For example, Rank Xerox reduced its supply base from almost 5000 suppliers in 1981 to 300 by 1987 (Morgan 1987). Lamming (1989) reported that Japanese lean producers involved less than 300 suppliers in the development of new products, compared to their Western counterparts, who dealt with 1000-2500. By transacting with substantially reduced supply bases, these particular firms were able to collaborate more deeply with suppliers to achieve improvements, notably in speed and reliability of delivery, quality and cost through waste reduction, thereby deriving much social capital through supply relationships. Smith-Doerr et al. (this volume)


discuss an upward spiral effect of collaborative capital glvmg rise to greater innovation, giving rise to greater collaborative capital, and so on.

Frazier et al. (1988) and Lascelles and Dale (1990) identified that traditional, adversarial transacting with competing suppliers was not conducive to generating good quality, thereby giving rise to social liability. The embracing of waste elimination concepts such as just-in-time (lIT) and total quality management (TQM) necessarily involved and depended on suppliers. What was common to all these developments in collaborative relationships is evident in Mowery's (1988) definition of interfirm collaboration in that they all sought to provide mutual medium- or long-term benefit and that they involved substantial contribution by partners in capital, technology, know how or other assets. However, the fact that some long-term, interfirm collaborations, notably in relatively high-volume, low-variety situations, gave rise to social capital should not be extended, without substantially more empirical evidence, to other contexts.

Recently these close, collaborative relationships with suppliers have been termed partnerships. Partnership sourcing has been defined as 'where customer and supplier develop such a close and long term relationship that the two work together as partners to secure the best possible commercial advantage. The principle is that teamwork is better than combat. If the end customer is to be best served, then the parties to a deal must work together-and both must win. Partnership sourcing works because both parties have an interest in each others success.' (Hornby 1992) This introduced the end customer, and better serving the end customer, as a driving force for forming collaborative relationships within which partners would increase their commercial advantage over and above that derived through independent working. Contractor and Lorange (1988) asserted the superiority of a strategy of cooperating over independent operating. Some authors have expressed the motivation for collaboration more strongly than a strategic choice taken for competitive advantage, but rather as mandatory to compete in global markets (Kogut 1988; Ohmae 1989; Porter and Fuller 1986).

However, the nature of partnerships is more than just teamwork. Teece's (1986) view of collaboration is one of supplier and customer providing complementary assets, which Contractor and Lorange (1988) stated would provide synergy. A substantial part of this synergy has been identified as learning (Hamel et al. 1989; Macbeth and Ferguson 1994; Imai et al. 1985; Lamming 1993) providing social capital to the relationship. However, the social exchange leading to social capital requires an appropriate atmosphere (Hakansson 1982). Social exchange within routinised, long-term collaborative relationships reduces uncertainty between the partners leading to an interlocking of supplier and customer through developed trust (Ford 1978). Thus trust is important in the building of social capital in relationships.

Trust in relationships relates in part to expectations (Hakansson 1982) and perceptions of performance. Ring and Van de Ven (1992) suggest that trust evolves partly from each party's perception of the other's fairness. Research on expectations and perceptions of performance in interorganization relationships has been contributed, notably, by the fields of service management and consumer behavior.

Supply Network Strategy and Social Capital- 415

Expectations Both parties to dyadic relationships may have different expectations (Voss et al. 1985); this may be due to real differences in opinion of supplier and customer or because they receive different information (Sheth 1973). Expectations may change over time as the parties learn through the process of interaction and as the relationship develops; satisfaction relates more to present needs, wants, and desires than to those prior to interaction (Cadotte et al. 1987; Miller 1976; Swan and Trawick 1980).

Expectations may be of different types. Boulding et al. (1993) classified expectations as ideal, should, and will expectations. Ideal expectations reflect enduring, possibly ambitious wants and needs that are relatively stable over time. Should expectations are those assumed should happen. Will expectations are tempered by experience and are a realistic assessment of what will happen and are affected by recent or critical incidents (Bitner 1985).

Therefore, to achieve the required trust that enables the synergistic acquisition of social capital in supply relationships, clarity and agreement on expectations is vital.

Perceptions of Performance Perception of performance replaces reality in the eyes of each party to a relationship (Sasser et al. 1978; Parasuraman et al. 1985; Gummesson 1987; Berry and Parasuraman 1991; Cronin and Taylor 1992). Perception is influenced by capability to judge tangible and intangible aspects of performance (Haywood-Farmer and Nollet 1991). Perceptions of performance have been identified as determining success or failure in the development of long-term relationships (Carlisle and Parker 1989).

Satisfaction and dissatisfaction arise as a result of comparison of expectations and perceptions of performance (Berry and Parasuraman 1991; Brogowicz et al. 1990; Gronroos 1990; Haywood-Farmer and Nollet 1991). Relationships where the gap between expectations and perceptions of performance is small are more likely to yield social capital in the form of satisfaction; however, where the gap is large, this gives rise to dissatisfaction or social liability. The social capital within relationships is vulnerable to critical incidents (Bitner et al. 1985), but distinction can be made between short- and long-term satisfaction. Attitudes in longer term relationships are more of a global nature and are a function of previous attitudes and satisfaction and dissatisfaction with current transactions. Boulding et al. (1993) argue that behavioral outcomes such as loyalty and positive word of mouth are a function of overall cumulative perceptions rather than short-term satisfaction with individual transactions.

Harland (1995) investigated gaps in perceptions of requirements and performance in dyadic relationships between players in the automotive aftermarket. A mismatch tool (see Figure 3) was developed to investigate these gaps.

In the relationships studied, customer dissatisfaction was highest where the supplier and customer had the largest difference in perception about performance in the relationship (mismatch 2). Substantial effort and investment was being made by suppliers in understanding what ·was required. However, little attention or


investment was devoted to understanding and communicating with customers about operational performance; this questioned the appropriateness of a marketing focus in favor of a supply focus. A further finding related to the closeness of the relationships and its impact on satisfaction and dissatisfaction. Dyadic relationships conducted through very close interpersonal relationships with depth of social exchange were compared to those conducted in more distant, nonfriendly relationships with more systematized information exchange than social exchange. The close relationships did not yield any higher satisfaction than the arm's-length relationships. This finding reflected differences in expectations-in some relationships the parties wanted closeness, in others they did not. These expectations then set the benchmark against which performance was compared by the parties. The different expectations typified cultural, territorial variation; parties to UK-based relationships expected more distance than those in Spanish relationships studied where friendship between supplier and customer was expected. These findings put into perspective some of the social capital literature through highlighting that closeness in relationships, per se, is not always a provider of social capital; if closeness is · not expected or welcomed by one of the parties to the relationship, actions by the other party to become closer may even give rise to social liability. Conversely, if closeness is not appreciated by a partner, nonclose relationships will yield more social capital to the parties than will close ties.

It is evident, therefore, that social capital is an important feature of longer-term, cooperative dyadic supply relationships though its presence is affected by parties' expectations, perceptions of performance, and resulting satisfaction and dissatisfaction and trust.

SUPPLY IN INTERORGANIZATIONAL CHAINS

A supply chain is defined here as a connected string of dyadic relationships that form a route for material-product-service flow. There are various terms to represent the same route. Hayes and Wheelwright (1984) termed this a commercial chain involving flow from raw material source, through manufacture, distribution, and ultimate sale to a consumer. Farmer and Ploos von Amstel (1991) referred to a supply pipeline. Marketing provides the term channel to refer to the route from a manufacturer downstream to consumer through distribution (e.g., Stock and Lambert 1987).

Supply chains have been studied in some research merely as a convenient unit of analysis of interorganizational features, notably logistical features. Studies in industrial dynamics (Forrester 1961; Burbidge 1961; Towill 1991) have observed what has been termed the Forrester or bull-whip effect. These studies have observed information flows and ordering patterns in supply chains that have given rise to cumulative error, or noise, that distorts real demand at the end of a supply chain as it is manipulated and transmitted in dyadic relationships and within firms in the chain. The distortion is amplified as demand is progressively transmitted up the supply chain. This cumulative error-arising, in part, from second guessing, lack of trust between dyadic relationships in the chain concerning forecasts of anticipated demand, batching of information, and delay in its transmission-is a form of social liability.

Supplier's perception of requirements


mismatch 1

mismatch 2

Customer's perception of requirements

Figure 3. Mismatch tool (Harland 1996a)

The principles of the Forrester effect have been tested in relation to behavioral factors in supply chains. Harland (1996a) showed that satisfaction and dissatisfaction increased upstream in supply chains and was positively correlated to the size of gaps in perceptions of performance. This provides evidence of other forms of social liability in supply chains that varies according to physical position (upstream or downstream) in the chain of supply.

However, supply chains have been considered more strategically than mere convenient units of analysis. Hayes and Wheelwright (1984) and Hill (1985) included strategic consideration of the direction, extent and balance of vertical integration in the commercial chain and subsequent sourcing decisions within their manufacturing strategy formulation approaches. Eberling and Doorley (1983), Jarillo (1993), Kogut (1984), and Porter (1985) discussed the strategic design of the value-added chain to implement different firm strategies, notably low-cost or differentiated strategies. Kogut (1985) examined international location of links in the value chain to gain comparative and competitive advantage. In the field of operations strategy, Shi and Gregory (1994) differentiated between a domestic and multidomestic orientation to international operations strategy and a global orientation and their impact on domestic, regional, multinational, and worldwide configuration of operations.

Evidence of normative, rational design decisions relating to differentiated strategies in supply chains is apparent in some documented case examples. Johnston et al. (1997) described how differentiated demand for blood products had lead to purposefully designed supply chains to satisfy that demand. For example, neonatal patients requiring fresh blood were served through the bleeding of donors located close to the point of use and a rapid testing and delivery route to the patient. The infrastructure of their supply chains was also differentiated; blood was collected in different designs of blood bags that matched the requirement of the procedure that would use the blood. In the automotive sector, Harland (l997b) identified that


supply chain design differed in some instances but not in others to meet the differentiated demands of particular original equipment manufacturers compared to aftermarket spares providers. Where a rational, differentiated approach to designing the supply chain had not been taken, certain critical links in the chain were less effective at coping with varying demands from downstream.

It appears, therefore, that individual supply chains can be designed and managed to meet differentiated demands of end customers. The nature of this differentiation gives rise to the need to understand and manage the complexity of the total set of chains that any firm supplies in. Extending the concept of focus (Skinner 1969) greater complexity and variety of these chains will impact on a firm's (and its dyadic relationships) ability to learn and benefit from the social capital that this learning provides. Therefore, a higher system level-the total supply network-has to be understood. Harland (l996b) proposed that strategic supply decisions should be considered at the level of the interorganizational network .

. SUPPLY IN INTERORGANIZATIONAL NETWORKS

Every firm is a nexus in its own unique network of upstream and downstream chains, forming a structure. Within that structure supply activities use the resources of the internal supply chain within each node and of the dyadic supply relationships that connect the nodes; this has been termed the infrastructure of the network (Harland 1996a).

Supply network infrastructure will not be discussed here but instead the structure of supply networks and strategies for them will be examined using theory and practice.

Supply Network Structure Nodes within supply networks contain those resources associated with supply. Bounding supply networks in this way distinguishes them from interorganizational networks, which become their network context; a supply network can be viewed as a slice, or plane, through the interorganizational network. Supply network structures differ according to their shape, where a firm is physically positioned in the network (which relates to type of player) and the size, power, and number of players contained within. There are other differences such as the international location of players and the capacity of the network, though these are not examined here.

Supply Network Shape Supply networks differ in shape, reflected in the breadth and length of the network, as shown in Figure 4.

In this volume, Uzzi and Gillespie highlight that the overall structure of the network within which a firm is embedded influences both the behavior of the firm and of the network as a whole. This chapter deals more specifically with supply network structure and its impact on supply performance.

Breadth o/the Supply Network The trends in dyadic supply relationships described earlier, notably supply base reduction, have resulted in the breadth of many supply networks decreasing.

Supply Network Strategy and Social Capital- 419

Network length

Network breadth

.... 1------ (number of levels or echelons) ------t~~

Figure 4. Supply network shape

Turnbull and Valla (1986) and Hakansson (1987) suggested that most interorganization networks are now more concentrated and structured.

There have also been changes in breadth of the downstream supply network structure in many cases. Womack et al. (1990) reported Japanese automotive distribution and retailing supply networks contained fewer dealers compared to Western counterparts-about 1700 in Japan compared with 17600 in the U.S. in a market only 2-3 times as large.

Operational performance benefits have been reported in narrower networks. Closer, longer-term relationships present in Japanese distribution networks provided ample social capital as they allowed more efficient sales and distribution and captured important customer feedback on design and manufacture (Womack et al. 1990), transmitting this knowledge upstream to facilitate learning and improvement. Easton and Quayle (1990) proposed that single-sourcing networks would be more rigid and strong in that there would be dense flows of exchanges within them. Also it would be easier to retain confidentiality in narrower networks.

However, it has been argued that there may be liabilities in narrower supply network structures. Concentrating sourcing on a smaller number of suppliers can increase risk and reduce learning from other networks (Sabel et al. 1987); Podolny and Castellucci (this volume) discuss the negative effects of being involved with too few others vis-a-vis learning. Easton and Quayle (1990) identified that narrower networks reduce the ability to adapt to changes in the environment through switching. Where uncertainty existed in the upstream supply market, Puto et al. (1985) advocated mUltiple sourcing as an important strategy.

Therefore, it can be seen that, while there is support in the literature of a trend toward narrower supply networks providing social capital benefits, there are possible social liabilities. The relative merits of broad and narrow networks are summarized in Table 2.


Table 2. Relative merits of broad vs. narrow networks

Advantages of Broad Networks

Adaptable to change More switching opportunities Wider access to knowledge Hedge against uncertainty Cost competitive

Advantages of Narrow Networks

Collaborative innovation Rigid and strong Dense flows of information Higher confidentiality Shared destiny

To date most of the empirical research on supply network breadth has been performed in the automotive industry, and its particular operational characteristics should be understood. Most automotive production is relatively high volume, low variety, as reflected in the documented cases on supply chains and networks, such as Toyota (Womack et al. 1990) and Nissan (Nishiguchi 1994). Product life cycles are such that a reasonable amount of stability can be maintained in the supply networks, which assists learning through collaborative working. Product design can be modularized, enabling automotive manufacturers to focus operations on assembly, and immediate suppliers on assembling modules, thereby causing breadth reduction through a tiered structure. Most automotive production competes primarily on price and quality; the previous discussion identified that having fewer, closer, longer-term relationships is more suitable for collaborative improvement of costs and quality.

However, recent research has provided an initial indication that supply network breadth varies across industries and across supply networks with different performance criteria, which may well imply that different types of supply network give rise to different forms of social capital and social liability. Harland (l997a) reported the findings of structured interviews of 200 firms across a range of sectors and at various positions in their supply network (see the later discussion on supply network position). Regarding supply network breadth, questions were asked about concentration of immediate suppliers and customers and about future plans for deliberate change to the supply network structure. Using Ward's method, the findings were clustered with other variables relating to the category of operation (type of operation, sector, volume, variety, size, and purchase spend), the business strategy of the firm, and supply strategy (whether one was perceived to exist, who was driving it, and upstream and downstream operational requirements). The following significant clusters were identified.

Tightly focused niche players (including manufacturers in yarns and fabrics, electronic components, cigarettes, soft furnishings, office and shop furniture fixtures and fittings, household goods, computers and communications, and toys/sports! leisure goods) had highly concentrated upstream supply networks, with 1 to 3 immediate suppliers representing 80 percent of purchase spend, whilst the downstream customer network was less concentrated, with 250-1000 customers representing 80 percent of revenue. These supply networks were driven by delivery speed, delivery reliability, and range requirements from the customer end of the network. In this cluster, limiting suppliers to such a small number increased social capital in learning how to respond quickly, through closer collaboration and dependence.


Manufacturers of food ingredients, printing and packaging, workwear apparel, and medical and surgical items with survival strategies were driven by service and flexibility by their customer network. The upstream supply network was relatively unconcentrated and in a state of continuous change, responding to informal, short notice demands from the downstream customer network. The shape of the supply networks of food processors, wood component manufacturers, and fashion apparel manufacturers were almost a mirror image of the tight, focused niche players. These networks were highly customer dominated, with 1-3 customers representing 80 percent of revenue. These customer dominated networks, while possibly deriving social capital benefits in the downstream, customer end of the network, appeared to give rise to social liability in upstream relationships, in terms of inefficient flexing to meet frequent, informal and unplanned changes, as well as stress within the operations of those relationships.

An example of social capital derived in broad networks was provided by the electrical goods manufacturers surveyed. These were in broad networks both up and downstream; they were driven by demands for constant innovation by customers. In order to satisfy the requirements of constant change from the customer end of the network, it suited these firms to retain breadth, and therefore choice and opportunity, in the upstream part of the network.

Variation in network breadth across industries requires more investigation before a taxonomy of network shapes can be developed. However, these initial indications of substantial differences do alert attention to the need for more research of supply network shape outside the automotive sector, particularly where different operational performance requirements exist. These differences are likely to impact nature and location of social capital and social liability in the total supply network.

Having examined issues of network breadth, the next section considers the number of echelons or levels in a network that determine its length.

Length of the Supply Network Nishiguchi (1994) reported that Toyota and other Japanese companies organized their suppliers into hierarchies; first-tier or primary suppliers provided systems rather than components. This significantly reduced the number of suppliers dealt with on a direct supply basis, but reorganization of the echelons in itself did not necessarily reduce the number of supply sources in the network in total, but instead imposed more levels in the network. Tiering made each firm more dependent on its suppliers, whereas traditional broader networks attempted to reduce dependency to suppress prices through competition to aid switching. Increasing the number of echelons necessarily placed intermediate relationships between any player and other players in the network. This may have reduced their visibility of operations in the network and may have impacted on learning resulting from direct interaction.

Therefore, the motivations for supply base reduction must be understood before the likely impact on social capital and social liability can be assessed. If supply base reduction, and hence a narrowing of the network, occurs through the removal of inefficient competition and adversarial bargaining across suppliers, this may provide social capital through more intense development of the same parts or services. However, if supply base reduction occurs through tiering, and hence lengthening of


the supply network, the remaining relationships no longer transact for the same goods and services but rather for modules or complete packages previously assembled in house. In this situation it is possible that social liability may occur through increasing the number of intermediaries involved in the operational flow.

Physical Position in the Supply Network and Type of Player In supply networks for manufactured products the types of players contained in the network relate to their physical position. At the upstream end of the network there are raw material extractors and processors whose materials are manufactured by component manufacturers; these components are assembled and then pass through distribution channels to reach their destination with the end customer. The supply network context for each of these players therefore varies according to their physical position. Each type of player has its own different role to play in the network; for example, most of product quality value is added in the upstream part of the network, whereas range, service, and delivery required by the end customer is usually satisfied by the downstream part of the network (Harland 1997b). Raw material extractors and processors often serve diverse downstream networks; for example, a foundry may serve a large range of industrial sectors including automotive, construction, capital equipment manufacturing, and any other sector where steel is used. These sectors may exhibit very different operational performance requirements. Relating this back to the idea of focus and role in networks (Harland 1997b), upstream players may have to learn about a wide spread of sectors and their requirements compared to players further downstream. It is not yet understood whether social capital in supply networks is most evident between the players in the network contained within one sector, rather than being derived through relationships with the more generalist upstream players.

Most of the research to date in supply networks has been conducted in networks for the provision of manufactured products. Service provision is conducted in very different network shapes. Typically a service provider is surrounded by a cluster of immediate suppliers of component services, people, and information used to provide a service package. For example, a travel agent deals with tour operators, transport providers such as airlines and railways, insurance brokers and currency providers, each providing a service element that the agent assembles into a service package. To date little research attention has been paid to service supply networks. Documented examples, such as in health care (Harland 1996b), tend to concentrate on the supply network for the tangible elements of the service package (in the case of health care provision, all the equipment and consumables required to perform health care services). As service networks appear to be shorter in length than most supply networks, there are fewer intermediaries to act as barriers to the transfer of knowledge and learning in the network; therefore, greater social capital may be evident in more of the network than in long, complex manufacturing networks; however, there is little evidence of empirical research to support this, as yet.

Size, Power, and Number of Players in the Supply Network There are only a limited number of documented cases on supply networks and their structures; interestingly they tend to be of large, powerful, high-volume operations


residing in the center of the total supply network-for example, Toyota (Womack, Jones, and Roos 1990; Fruin 1992), Nike (Lorenzoni and Baden-Fuller 1995), Benetton (Jarillo and Stevenson 1991), Coming (Lorenzoni and Baden-Fuller 1995), Nissan (Nishiguchi 1994), SweFork (Dubois 1994), Volvo (Kinch 1992) and Apple (Jarillo 1993; Lorenzoni and Baden-Fuller 1995). Each of these large players controlled the upstream and downstream parts of the network, apparently benefiting from the social capital they derived in these closely managed relationships.

Sako (1991) highlighted that while the Japanese manufacturing industry employed just over twice the number of people compared to British manufacturing, the industry had nearly four times the number of firms, with significantly more smaller businesses. However, Mitsui (1990) drew attention to structural forms in Japan other than the highly pyramidal vehicle networks which tended to have larger first-tier suppliers dealing with smaller lower-tier subcontractors, again highlighting that care should be taken in generalizing findings from limited cases in one sector. Highly flexible and organized networks of very small engineering companies existed in textiles, printing, construction, and information technology sectors where the smaller firms were indispensable partners to the large companies. Currently there is inadequate research to identify if there is any correlation between the number and size of players in supply networks, their resulting performance, and evidence of social capital or social liability. However, the observed differences indicate potential for such research.

Research into the size, power and number of players in downstream parts of supply networks has been provided in the marketing literature on channel management. Stock and Lambert (1987) described how goods are physically transported between the place they are produced and the place they are consumed through channels. If direct contact is made between every manufacturer and every customer, a large number of contacts occur. Selling through an intermediary reduces the number of contact points made in the market and also improves service to the customer as they deal with one point of contact. Channels have therefore evolved to fulfill this intermediate role between manufacturer and consumer, reducing transaction costs in the downstream part of the supply network; therefore, the creation of new supply network players, such as an intermediary, may provide social capital in some circumstances because of specialization of relationship management.

Not all channels have evolved in the same way. Different channel structures are apparent in different markets and for different products. Stock and Lambert's (1987) figure of alternative structures of channels of distribution of industrial goods is shown in Figure 5.

These variations of structure showed differences in the number, ownership, and role of intermediaries. It does not appear that anyone structure, or the presence of absence of an intermediary, necessarily gives rise to the presence of social capital or social liability. Rather, it appears that it is more an issue of 'horses for courses,' where different structures are appropriate to different contextual situations.

The traditional marketing literature, based on an industrial economics perspective of channel structure with the underlying belief that producers organize channels to minimize costs, has been criticized (Gattoma 1978), leading to the emergence of behavioral models. These suggest that channels and channel structures


Manufacturer Manufacturer

Manufacturer Manufacturer

Figure 5. Alternative channel structures (Stock and Lambert 1987)

evolve as a consequence of the actions of all the parties involved-namely manufacturers, agents, distributors and retailers. Behavioral models lead to the conclusion that channel behavior is based on power, conflict, leadership and cooperation. Notable studies on power in marketing channels have been provided by Etgar (1976), El-Ansary and Stern (1972), and Lusch and Brown (1982); these studies concentrate on identifying the nature and sources of one channel member's power over another. The impact of power on conflict in channels has been identified, notably by Rosenburg and Stem (1971), Lusch (1976), Etgar (1979), and Wilkinson (1981). It appears from these studies that abuse of power may give rise to social liability in terms of detrimental impact of conflict within the network.

More recently, models of channels have combined both economic and behavioral theories; these have been termed managerial models. Managerial models ambitiously attempt to incorporate a wide variety of variables including consideration of the channel as a political economy affected by individual firm dynamics and the economic environment of the channel. Extension of managerial models into strategic models (see, for example, Knee and Walters 1985) support the belief that channel evolution is a consequence of strategic choices made by manufacturers, distributors, agents and retailers within the channel.

While the research on supply network structure has been conducted by academics in different fields, evidence has been provided that shows differentiation of supply network structure according to industrial sectors, operational performance priorities, and territory. The examples provided in the discussion appear to indicate that some firms have taken a rational, strategic approach to design of supply networks or part of them. The next section examines the current state of knowledge on supply network strategy.

Supply Network Strategy To date work in supply network strategy has taken a rational, normative approach to strategy formulation following Chandlerian principles that structure follows strategy (Harland and Wensley 1996). A supply strategy formulation approach provided by Harland (1996b) extended the rational operations strategy formulation process provided by Hayes and Wheelwright (1984) and Hill (1985) and is shown in Table 3. This approach is in contrast to the IMP approach to interorganizational network


Table 3. Extension of operations strategy elements to supply network strategy

Competitive Priority

Structure

Infrastructure

Manufacturing Strategy

Of the operation:

Price (cost)

Dependability

Flexibility

Quality

Capacity-size, volume, timing

Facilities

Supply Network Strategy

Of the end customer and each

supply network actor:

Price (cost)

Delivery speed

Flexibility

Product quality

Innovation

Range

Service quality

- reliability

- responsiveness

- competence

- access

- courtesy

- communication

- credibility

- security

- understandinglknowing

the customer

- tangibles

Capacity-size, volume, timing

Supply network actors

configuration

Production equipment and systems Supply network facilities

configuration (e.g., fleet, buildings,

materials handling systems)

Internal or external sourcing

Human resource policies

Quality systems

Production planning and control

New product development

Organization

Performance measurement

Do-or-buy

Supply network human resource

policies

Supply network quality systems

Supply network operations

planning and control

New product or service

development

Network organization

Performance measurement


strategy, which takes a more emergent, Mintzbergian approach, claiming that players in the network 'merely cope' (Hakansson and Snehota 1995). The differences in approach might be explained by the tangibility in supply networks that enables a more planned approach to supply strategy. A further justification for the continuance of rationality lies in the role of operations, and supply, in meeting the focused, measurable range of operational competitive priorities.

Some firms appear to have designed and managed their supply networks strategically and successfully, embracing the rational principles of focus and segmentation (both upstream and downstream). For example, Benetton retained control over color quality and design within its firm boundary but formed relationships for most of manufacturing and retailing. Toyota focused its manufacturing and distribution operations according to models; model type reflected differences in competitive priorities.

However, their success may lie in their management of information in the supply network and the resultant social capital in the form of flexibility and speed of supply network learning, much of which was conducted through social exchange. Benetton was the first fashion retailer in its market to be able to respond to variations in demand within the season. To do this they ensured quick response to information on what was being sold in the franchised retail outlets by connecting point of sales information back to the factory. They changed their production operations to leave dyeing as the last process to be conducted, thereby increasing their speed to response. But their upstream supply network was quite distinct; it consisted of a large number of contractors and sub-contractors who were connected through familial and social relationships with Benetton employees. The subcontractors only worked, and were paid through informal means, when there was demand. In this way Benetton did not have to take the risk or responsibility for the upstream supply network and were able to use it flexibly and quickly when required. Toyota's use of door-to-door selling maintained social contact with end customers. Not only did this provide information flows about likely future demand, but, more important, it provided access to design and quality feedback from customers over a vehicle's life; this information was fed back to Toyota manufacturing and to the upstream supply network (Y'fomack et al. 1990).

Benetton and Toyota appear to have acted as hubs in the supply network, facilitating and coordinating quick flows of information, connecting the upstream supply network with downstream demand. It is in this hub role that they differed to their competitors. Burt (1992) identified the importance of access, timing and referrals relating to information flows in networks. Burt identified that information does not flow easily through interorganization networks; the network helps to screen the large volume of information held by players, providing some focus on key pieces of information. He highlighted that benefit-rich networks 1) have contacts established in the places where useful bits of information are likely to air and 2) provide a reliable flow of information to and from those places. It is this establishment of contact with the right parts of the supply network, the filtering of relevant information and focus and communication of critical pieces of information that constitutes the core role of a supply network hub.


CONCLUSIONS

Supply networks are a special case of inter- and intraorganizational networks, containing value-adding flows of materials, products, and services. To date there has been limited research on supply networks; most of the research has concentrated on automotive networks and attempts to generalize from this work should be viewed with caution. Recent research has provided initial indications of the substantial differences between supply networks relating to whether they contain flows of products or services, which sector they are in and the operational competitive priorities driving demand in the network. These differences are evidenced in different network structures.

While the feasibility of interorganization network strategies and their implementation is contended, it appears that the tangibility of supply networks may enable a more planned, rational approach enabling implementation of supply network strategies. Increasingly it appears that these strategies address differentiation to meet different operational competitive priorities of end customers groups.

A feature of successful interorganizational supply networks that warrants more research attention is that of coordination, management, and communication of critical information. Firms that appear to have strategically managed their supply networks have created social capital by acting as hubs, capturing important information from one part of the network and transmitting it to where it required.

• SECTlONY

Structure at the Firm Level social capital and financial capital

•

Choosing Ties from the Inside of a Prism: Egocentric Uncertainty and Status in Venture Capital Markets

ABSTRACT

23 Joel M. Podolny

Fabrizio Castellucci

This chapter draws an analytical distinction between altercentric and egocentric uncertainty. Altercentric uncertainty refers to the uncertainty that buyers face about the product quality of a focal producer (ego). Egocentric uncertainty refers to the uncertainty that the producer itself faces about the resource allocation decisions that will result in a product that is regarded as high quality by buyers. This chapter then argues that the value that a firm derives from its own status is positively related to altercentric uncertainty and negatively related to egocentric uncertainty. That is, status is valuable when buyers can use it as a signal of quality, but status is not valuable when a producer does not know how to 'spend' its status to produce quality. As a consequence, high status producers should seek out markets or market segments where egocentric uncertainty is low. This argument and hypothesis are tested in an examination of the venture capital markets.

INTRODUCTION

Sociologists have traditionally regarded networks as the plumbing of the market. That is, networks are the channels or conduits through which 'market stuff flows. 'Market stuff encompasses information about exchange opportunities as well as the actual goods, services, and payments that are transferred between buyer and seller. Granovetter's (1995) work on job search is one example of research emphasizing the flow of information about exchange opportunities (see also Flap and Boxman, this volume), as is recent research on the role of interlocking directorates in the dissemination of information and perceptions of legitimacy (Haunschild 1993; Davis 1991). Perhaps the most noted example of work stressing the role of networks in the


transfer of goods and payments is Burt's (1992) work on intersectoral flows in the US economy. Recently, however, sociologists have started to perceive a second function of network ties within the market. Network ties between a delimited set of actors are not only relevant as pipes for the flow of market stuff, but as informational cues from which others outside that set make inferences about qualities possessed by those actors (also see Nooteboom, this volume). As an empirical example, consider the syndicate relations formed among investment banks when underwriting corporate securities. When one bank agrees to be in an underwriting syndicate led by another, the former invariably must give the latter the right to list its name in the 'tombstone advertisement' announcing the offering in prominent trade publications, such as The Wall Street Journal. While a number of scholars have discussed in detail how the location of a bank's name in the advertisement reflects and determines its status (e.g., Hayes 1971; Eccles and Crane 1988; Podolny 1993), it is sufficient here to note that a low vertical placement is indicative of low status and a high vertical placement is indicative of high status. In accepting a subordinate position in the advertisement, a bank exhibits deference to those banks that are listed higher. A syndicate relation between bank A and B thus has a dual significance. On the one hand, it is a conduit of resources or information passed between A and B. On the other, it is an informational cue affecting third party perceptions of the relative quality of the services that A and B offer in the market. Raised to the level of a metaphorical abstraction, this example illustrates that networks are not only pipes carrying the stuff of the market; they are prisms, splitting out and inducing differentiation among a set of economic actors.

An increasingly broad array of sociological work can be grouped into this prismatic perspective. For example, Baum and Oliver (1992) show that potential consumers regard a day care center as more legitimate if it possesses a tie to some prominent organization in the community, such as a governmental agency or church group. Similarly, Stuart, Hoang, and Hybels (1999) demonstrate that the investment community is more receptive to biotechnology firms that possess an affiliation to a prominent alliance partner (also see Stuart, this volume). At a theoretical level, both pieces of work show that the affiliation with a high-status or trustworthy third party signals quality. As a third example, Zuckerman (1999) illustrates that a firm's pattern of exchange relations with financial analysts affects how favorably that firm is evaluated by the investment community. In all of these pieces of work, the presence or absence of an exchange relation between two actors constitutes an informational cue that relevant third parties use in evaluating those actors. Considered at the more macro-level, exchange relations between some subset of market actors yield perceptual distinctions along some dimension, and these perceptual distinctions inform the market decisions of some other subset.

In this chapter, we build on the research within this broad, prismatic perspective that has employed the sociological conception of status to understand this differentiation. Prior research (Podolny 1993; Podolny, Stuart, and Hannan 1996) has argued that there exist a diverse array of ties among competitors on one side of a market with the same differential significance that we noted in the tombstone advertisement. For example, patterns of alliance formation, personnel transfer, resource flows, and directed technological imitation between competitors often

Choosing Ties from the Inside of a Prism - 433

convey deference and, in so doing, induce a status ordering among those competitors. A firm's position in this status ordering is a signal that the firm's constituencies-such as potential buyers, possible alliance partners, and third party observers (e.g, financial analysts, investors, media, governmental agencies)-utilize in making inferences about the firm's quality (Podolny 1993; Han 1994).

If a firm's relevant constituencies rely on status to make inferences about the quality of the firm's product or service, then the value of status should vary inversely with those constituencies' ability to observe and assess quality directly. For example, as a firm's buyers become more certain of the quality of the firm's service or product, then the value of the firm's status should decline. Conversely, as the buyers become less certain of the quality of service or product that the firm offers, the value of the firm's status should rise. There is some evidence that the value of a firm's status does hinge on the difficulty that the firm's constituencies have in directly perceiving or evaluating quality (Podolny 1993; Podolny 1994; Podolny, Stuart, and Hannan 1996).

However, constituency uncertainty regarding quality is not the only form of uncertainty that a focal firm faces. As numerous organizational scholars have emphasized (Thompson 1967; Cyert and March 1963; Stinchcombe 1990), a firm also confronts uncertainty in making its own resource allocation decisions. In fact, from the perspective of a focal firm, we can distinguish between two types of market uncertainty. The first type is the uncertainty that the firm has regarding market opportunities and the set of resource allocation decisions that will best enable the producer to realize those market opportunities. We label this type of uncertainty 'egocentric uncertainty' because the focal firm is the actor who is uncertain. The second type is the uncertainty that consumers or possibly other constituencies such as potential alliance partners have about the quality of goods or services that the firm presents to them in the market. That is, a producer of a good or a provider of a service may be very aware of the quality of that which it provides on the market, but potential exchange partners can still be unaware of this quality due to lack of previous contact, knowledge of, or experience with the firm offering the good or service. We label this type of uncertainty 'altercentric uncertainty' because the firm's alters-the consumers or potential alliance partners-are the actors who are uncertain.

We know that the value of a firm's status increases with altercentric uncertainty, but how is a firm's status affected by egocentric uncertainty? In the next few pages, we argue the following: whereas the value of status increases with altercentric uncertainty, the value of status declines with egocentric uncertainty. Though we justify this claim in some detail in the following pages, we can at this point layout the broad contours of the argument: the greater the egocentric uncertainty confronted by a firm, the less that the producer knows how to leverage its status to acquire resources in the pursuit of superior resource allocation decisions. Stated in terms of social capital, status manifests itself as a productive asset and hence as corporate social capital to the extent that a producer's constituencies are uncertain about the quality of the good that the producer brings to market. However, the more that the producer faces uncertainty about what resource allocation decisions yield a


product that will be regarded as high quality, the less value the producer derives from status.

If the value of status declines with egocentric uncertainty, high-status firms should attempt to move away from markets or segments in which there is a high amount of egocentric uncertainty and toward markets or segments in which there is a low amount of egocentric uncertainty. After developing this argument in more detail, we test it in an examination of the venture capital markets.

STATUS AND EGOCENTRIC UNCERTAINTY

To understand why there is a negative relation between a firm's status and egocentric uncertainty, we begin with a simple observation: a firm's quality is a function of the feasible set of resource allocation decisions that a frrm can make, and the feasible set of resource allocation decisions is largely determined by the exchange partners to which the firm has access. Concretely, a firm will not be able to make a high quality product if it lacks access to suppliers of high quality labor or raw materials.

The set of exchange partners to which a focal firm has access is, in turn, a function of the firm's status. There are two reasons why this is so. First, a highstatus firm will generally have greater access because potential exchange partners will be more likely to believe that the firm's output is above some quality threshold. For example, the higher a focal firm's status, the more that potential alliance partners will positively evaluate the prospects of an alliance with the focal firm. Second, a high-status firm will generally have more access because at least some potential exchange partners will value the association with status either as an end itself or as a means toward some other end. Consider consulting firms recruiting MBAs. High-status consulting firms will presumably find it easier or less costly than low-status firms to hire the MBAs of their choice. The MBAs may either value the association with status as an end in itself or value the high-status affiliation as a resource that can be leveraged to pursue future opportunities.

Some research demonstrates that if a high-status firm enters into exchange relations with lower-status actors, then the firm risks the dilution of its status (Podolny and Phillips 1996). Our point is, therefore, not that the high-status firm is completely unconstrained in its ability to choose exchange partners. Rather, our point is simply that at any particular time, a high-status firm has more exchange opportunities than a lower-status counterpart. The high-status firm will find it easier to enter into exchange relations with other high-status actors. And while the highstatus firm may not wish to risk the dilution of its status by entering into exchange relations with lower-status actors, the high-status firm at least has the opportunity to pursue such relations if it believes that the tangible benefit from the exchange outweighs the potential loss of status.

Since part of the value of status lies in the breadth of exchange relations to which it grants a firm access, the value of status necessarily declines when a firm does not know which exchange partners it should be pursuing. Or, stated in terms of egocentric uncertainty, the higher the egocentric uncertainty confronted by a firm, the lower the value of status.


Were the discussion of status and egocentric uncertainty to end here, then we would be left with the following conclusion: high-status firms should make better resource allocation decisions when there is low egocentric uncertainty, and they should make approximately the same quality of resource allocation decisions in a situation of high egocentric uncertainty. That is, the foregoing discussion suggests no reason why high-status firms should actually make worse resource allocation decisions than low- status firms in a condition of high egocentric uncertainty.

However, if we adopt a behavioral conception of the firm (Cyert and March 1963; see also Uzzi, this volume) and assume that firms engage in problemistic rather than profit-maximizing search, then it is possible to conclude that low-status firms may make better resource allocations decisions than their higher-status counterparts in a condition of high egocentric uncertainty. The reason is the following: status is a source of positional advantage in markets, and by definition, positional advantage insulates a firm from the competitive pressure that would lead a firm to engage in problemistic search and therefore learning (Barnett 1997). When egocentric uncertainty is low, this learning disability is not problematic because there is, by definition, little need to learn. However, as egocentric uncertainty increases, this learning disability becomes much more problematic for the highstatus firm.

Such a learning disability does not mean that a high-status firm will obtain fewer market rewards than a low-status competitor. The positional advantage deriving from status may still outweigh or at least counterbalance any weakness in capabilities resulting from inferior learning. Rather, the learning disability of status only means that when egocentric uncertainty is high, there is a sharp trade-off between the positional advantage afforded by status and the capability-based advantage that can be afforded by learning. As a consequence of this trade-off, highstatus firms may make worse resource allocation decisions than their low-status counterparts.

Yet regardless of whether high egocentric uncertainty causes high-status firms to make worse resource allocation decisions than low-status firms, the fact remains: the competitive advantage of status declines with egocentric uncertainty. As a result, we expect that high-status firms will avoid market segments in which there is high egocentric uncertainty and move toward market segments in which there is low egocentric uncertainty. That is, high-status firms will avoid market segments or niches in which there is considerable uncertainty about which exchange partners will help yield a product or service valued by consumers, and they will actively seek out niches and segments in which the opposite holds. For example, when Drexel Burnham Lambert-a relatively low status-bank in the early 1980s-pioneered the market for non-investment grade or ~unk' debt, the high-status banks were initially reluctant to enter this market despite the extremely high margins to that went to underwriters. Undoubtedly part of the reason that high-status banks avoided this segment was that they simply did not wish to be perceived as affiliates of the lowstatus corporate issuers that populated this market (Podolny 1994). But part of the reason may also have been that they did not see how they could leverage their status in a market in which there was such high uncertainty about which issuers should be sought out as exchange partners.)


Or, as another example, consider cultural organizations, such as film studios. As Hirsch (1972) observes, the markets in which cultural organizations find themselves are characterized by relatively low altercentric uncertainty, but relatively high egocentric uncertainty. That is, after a movie is made, a typical movie-goer has relatively inexpensive ways to gauge whether she will like the movie. She can rely on the opinion of either trusted critics or trusted friends; that is, a relatively small number of opinion leaders can effectively bear the search costs for a large number of movie-goers. However, before the movie is made, the studios face considerable uncertainty as to what exchange relations with writers, actors, and directors will yield a product valued by the consumers. While the movie-going public may not perceive salient status distinctions among film studios, investors clearly do. Based on the foregoing discussion, we would expect that the high status studios avoid film genres or projects in which there is relatively high uncertainty about which combination of artistic inputs will yield a product valued by the movie-going pUblic.2

Thus, our central hypothesis is the following: the higher a firm's status relative to its competitors, the more that it will retreat from market segments in which there is high egocentric uncertainty and seek out market segments in which there is low egocentric uncertainty. To test this hypothesis, we now move to a consideration of the empirical context for our analysis: the venture capital market. 3

VENTURE CAPITAL

In the market for venture capital, venture capitalists occupy a broker role between investors and entrepreneurial companies in need of financial capital (see Freeman, this volume for a more extensive discussion of the venture capital industry). A venture capital frrm enacts this brokerage role by first raising money from the investors and placing the money into a fund. The fund is organized as a partnership, with the senior members of the venture capital firm serving as general partners and the investors having the role of limited partners. The venture capital firm invests the fund's money in entrepreneurial companies in exchange for an ownership stake. At the end of a fixed period of time, usually 7 to 10 years, the fund is dissolved. The venture capital firm takes a fraction of the proceeds, usually about 20%, and distributes the remainder to the limited partners in proportion to their original investment in the fund.

Venture capital firms are evaluated on their ability to generate high returns for their limited partners. A venture capital firm may manage more than one fund at a time, but the returns to the limited partner derive solely from the performance of the fund in which she or he invested. The greater the limited partners' return on the money that they invested in the fund, the easier it is for a venture capitalist to raise money for subsequent funds.

For the purpose of this analysis, we regard investors as 'consumers,' venture capital firms as 'producers,' and the allocation of funds to entrepreneurial firms as the most critical resource allocation decision made by venture capital firms. While this particular resource alIocation decision is the most critical, it is not the only resource alIocation decision that venture capital firms make. In addition, venture capital firms often assist the entrepreneurial firms they back in the selection and


recruitment of board members and critical managers, such as the CEO or the CFO. More generally, venture capital firms must decide how to best allocate the time and energy of their own employees across different activities.

From the perspective of the venture capitalist, the venture capital markets are characterized by both high egocentric uncertainty and high altercentric uncertainty, where investors constitute the critical 'alters.' Neither the investors nor the venture capitalists have a high degree of certainty about the benefit that they will derive from any particular investment. Even though some of the uncertainty associated with investments in this market can be 'priced away' and thus transformed into risk, there still remains a large amount of uncertainty in these markets that is difficult if not impossible to price. Venture capital emerged as an institution largely because traditional financing institutions were unwilling to invest in entrepreneurial firms lacking collateral. The origins of venture capital are thus indicative of the generally high level of egocentric and altercentric uncertainty in this particular market. In personal conversations with venture capitalists, we have found them to be quite open about the fact that there exist no reliable quantitative formulas for evaluating the risk associated with their investments.4 The quality of a venture capital firm is a function of its ability to make superior investment decisions in the context of this uncertainty. The venture capital firm must learn to identify which attributes of an entrepreneurial firm make that firm more likely to emerge as a success and go public or be acquired at a high price. In making an assessment of an entrepreneurial firm's chances for success, the venture capitalist will often consider a diverse range of factors. For example, the venture capital firm must evaluate the entrepreneurial firm's technology, the managerial ability of the firm's founders, the dynamics of the market(s) in which the entrepreneurial firm hopes to compete, and the potential responsiveness of the financial markets to a public offering of the entrepreneurial firm's equity.

While it should now be clear that there is a high level of altercentric and egocentric uncertainty in the venture capital markets, it is equally important to note that there is variance in the level of egocentric uncertainty across investments. To understand one basis for this variance, we must discuss in slightly more detail how entrepreneurial firms raise capital from venture capitalists. Entrepreneurial firms generally raise money in 'rounds.' That is, entrepreneurial companies do not receive a continuous stream of payments from venture capitalists. Rather, they seek financing over some discrete time period. At the end of the time period, the process is then repeated. In some rounds, a firm may require more money than anyone venture capital firm is willing to invest. In these cases, multiple firms may invest in a given round.

Rounds can be categorized into stages. For the purpose of this analysis, we divide rounds into three stages. Start-ups that do not have a viable product are regarded as being in the first stage. At this stage, the entrepreneurial firm may have little more than an idea or concept for a product. At least some of the financing in this stage is referred to as 'seed' or 'start-up' financing. Once an entrepreneurial firm proves that its product is viable, it may seek financing for the commercial manufacturing and sales of its product. We denote such financing as second stage financing. Second stage financing takes place after a company has started to produce


its product but typically before the company has become profitable. Finally, the company enters third stage financing when it is profitable but is seeking capital for further expansion. As a company progresses through the various stages, its value increases. From the perspective of the venture capital firm, that is from the perspective of an investor purchasing equity in a company, the highest returns are to be found when a company that was financed in the early stage demonstrates a promise of high future earnings and goes pUblic. Because there is less uncertainty in the later stages, the returns from a late stage investment that goes public are generally smaller than the returns from an early stage investment that goes public.

An entrepreneurial company does not necessarily begin to receive financing in a seed or a start-up round, and a company need not go through all stages of financing before being acquired or going public. For example, because they have high capital needs in the product development stage, biotechnology firms will typically go public long before they have a marketable product. Given this categorization of stages above, biotechnology firms frequently go public before they conclude first stage financing.

For our purposes, what is important about these stages is that they provide a basis for categorizing investments in terms of the egocentric uncertainty faced by the venture capitalist at the time of the resource allocation decision. The later the stage of the investment, the less uncertainty that the venture capital firm confronts regarding the outcome of an investment decision.

Whereas the altercentric uncertainty of investors may also vary with the stage of company in which their money is invested, we believe that it is reasonable to assert that the variance in altercentric uncertainty within the venture capital markets is less than the variance in egocentric uncertainty. There are at least two reasons why this is so. First, as noted above, investors do not typically make decisions to invest in particular companies; rather, they make the decision to invest in a fund. While some funds may be targeted to the investment of companies at a particular stage, most are not. Therefore, an investors' decision to place money in one fund rather than another is associated with less variance in uncertainty than a venture capitalist's decision to invest the fund's financial resource in start-ups at one stage rather than another. Second, investors invariably have only part of their financial resources in the venture capital markets; they diversify their risk by including the equity of more established firms in their investment portfolio, and they often invest in other financial products. As a result, investors will generally devote less time and energy than the venture capitalists to being knowledgeable participants in these markets. The less knowledgeable one is about existing entrepreneurial firms in general, the less that one's insight into the future performance of a firm can vary with the firm's stage. Thus, whereas egocentric uncertainty will be strongly associated with the stage of an investment, altercentric uncertainty is more constant across these stages.

While this distinction between the variance in egocentric and altercentric uncertainty may seem to represent an analytical digression, it is critical to our empirical analysis. Because egocentric uncertainty varies across market segments more than altercentric uncertainty, we can now move from our general hypothesis in the previous section to a more specific prediction in this particular context. To the extent that high-status firms seek out market segments in which there is high


egocentric uncertainty, we expect the following: as a venture capitalist's status increases, it should gradually shift its portfolio of investment from the early rounds to the later rounds.

If altercentric uncertainty revealed considerable variance across segments, then it would not be possible to develop a straightforward prediction. While status would still be of less benefit in making resource allocation decisions in the early rounds, status would potentially be more valuable as a signal. If the signaling value of status were noteably higher in the early stages than the later stages, then there would be counterveiling effects on a high-status venture capitalist's decision to invest in early or late stages. On the one hand, the higher egocentric uncertainty would push the high-status venture capitalist to invest in later stages, but the higher altercentric uncertainty and thus signaling value of status would push the venture capitalist back in the direction of early round investments. However, because altercentric uncertainty displays little variance across market segments, it is possible to make the straightforward prediction that an increase in status should lead a venture capitalist to invest more of its financial resources in later stage start-ups.

Model To test the hypothesis, we propose the following model:

m

R il+1 = BISil + LBkXk +O'i +1", +£il+l

k=2

where Ril+1 denotes the average stage of firm i's investments during year 1+1, Sit

signifies the status of venture capitalist i in year t, X2 through Xm represent a set of control variables, 0'; indicates a firm-specific effect for venture capital firm i, and 1",

reflects a time-specific effect for year t. The model can be estimated using conventional OLS techniques for the analysis of panel data. Our central hypothesis is that BI > o.

DATA AND MEASUREMENT

To test this hypothesis, we obtained data from the Securities Oata Corporation's (SOC) Venture Economics data base. The SOC collects information on numerous financial markets and then sells this information to the various financial communities. The SOC obtains the information in the Venture Economics data base from both public and private sources. Venture capitalists use this data to obtain benchmarks for their performance, and entrepreneurs use this data to better understand venture capitalists' investment preferences.

Though the SOC has information on entrepeneurial start-ups dating back to the early 1970s, data from the 1970s is extremely scant, leading us to believe that much of the data from this time period is missing. Moreover, it was not until probably the mid-1980s that venture capital emerged as a clear industry whose provision of financial resources was regarded as a critical resource for entrepreneurial start-ups, especially in the high technology sectors. Because we wish to isolate the egocentric uncertainty that is due to variance in the stages of investments from the variance that may be due to the maturity of the venture capital markets and because of our


concerns about the scantness of the data during the early period, we limit our analysis to the time period between 1986 and 1996, inclusive.

In addition to deciding on an appropriate time window for the analysis, one also needs to decide which actors are to be considered venture capitalists. The SOC venture capital database includes information not only on the investments of venture capital firms, but also information on the investments of key individuals and organizations, such as universities, that make investments in entrepreneurial firms. These individuals and organizations might often be prominent actors in the venture capital community; in fact, the partners in venture capital firms wiIl not infrequently make investments in entrepreneurial firms as private individuals apart from the firm for which they are partner. These individuals and organizations should be able to derive the same access benefits from status that a venture capital firm would derive, and these individuals and organizations should therefore experience similar pressures to focus their investments in particular segments. Therefore, it seemed important not to exclude them from the popUlation of venture capitalists that make investments in entrepreneurial start-ups. That being said, there are hundreds of individuals and organizations reported in the data that may make only a couple investments over a several year period, and it seemed unreasonable to consider these transitory investors as venture capitalists with pretensions to having some status in this industry. As a selection rule, we exclude all individuals and/or organizations from the population of venture capital firms that make less than 10 investments over the 5 year period prior to the year for which the status matrix was constructed. After employing this selection rule, we obtained a population of 387 venture capitalists over the 1986 to 1996 time period. These 387 venture capitalists yield 2386 firmyear observations. The average number of years that a firm was in the data was 2386/387 = 6.16 years.

Dependent Variable: Average Investment Stage Measuring the average stage of a venture capitalist's investments in year t+ I is straightforward. The SOC Venture Economics database categorizes investments into the three investment stages discussed above. To reiterate, an entrepreneurial firm is considered a first stage investment when it does not yet have a viable product. It is identified as a second stage investment when it has a viable product but is not yet able to profitably manufacture and distribute the product. Finally, the entrepreneurial firm is categorized as a third stage investment when it is generating a profit. We assign first stage investments a value of I, second stage investments a value of 2, and third stage investments a value of 3. We then add up the values associated with a firm's investments in a given year and divide by the total number of investments. The range of this variable is accordingly between 1 and 3, inclusive.

Explanatory Variable: Status To measure the status of the venture capital firms in year t, we construct a matrix based on the joint involvement of venture capitalists in financing entrepreneural start-ups. That is, we construct a matrix R, in which cell Rij, denotes the number of times venture capitalist i and j jointly financed a start-up between years t and t-4, inclusive.


Given R" we then calculate status scores based on Bonacich' s (1987) measure (see the chapters by Doreian and by Han and Breiger, this volume for alternative ways to measure status). A venture capitalist's status is a function of the number and status of the firms with which it jointly finances start-ups; the status of these financing partners is in tum a function of the number and status of their syndicate partners, and so on. Particularly given that we include the number of deals in which a venture capital firm has participated as a control variable in the analysis (see below), this status measure reflects the extent to which a firm has financing partners who are 'players.'

Conversations with those in the industry provide at least anecdotal support for the use of this particular measure. Lower status venture capitalists express a strong desire to be included on deals financed primarily by higher-status firms, and higher status venture capitalists occassionally refuse to finance a venture if that venture is receiving financing from a lower status venture capitalist. In effect, to be high-status is to be an insider; to be low-status is to be an outsider, and joint financing constitutes a symmetrical form of deference in which each venture capitalists acknowledges the standing of the others.

Control Variables In addition to status, we include two control variables in the analysis. First, we include the number of investments made by the venture capitalist between years t and t-4, inclusive. There are several reasons to include ths variable as a control variable. First, to the extent that we find evidence consistent with our hypothesis, we would like to disentangle the effects of relationally-based status from simply the volume of activity in which a venture capitalist is engaged. That is, we would like to distinguish the effect of status from the effect of size. There is a second reason for including this variable. The number of deals in which a firm is engaged is a measure of a firm's experience, and to the extent that a firm learns from experience, a firm's total amount of investments should reduce the uncertainty that it confronts in making subsequent investment decisions. Since there is more egocentric uncertainty and thus more to learn about investing in the early rounds than in the late rounds, we would expect that such learning from experience would have a greater effect on a firm's ability to make judicious early stage investments than judicious late stage investments. In effect, whereas status should push firms to make more late stage investments, experience should push firms to engage in a high proportion of early round investments.

Of course, it seems reasonable to expect that there is a strong positive correlation between status and the number of investments made by a firm. In fact, as indicated in Table 1, the correlation is .79 if one includes both within-firm and cross-firm variance and .66 if one includes only within-firm variance.5 Thus, in arguing that one of these variables has a particular effect, we are implicitly assuming that the value of the other variable is held constant.

A second control variable is the number of funds from which the venture capital firm makes investments in year t. As noted above, a venture capital firm can have more than one fund from which it makes investments at any particular time. There


Table 1. Descriptive statistics for explanatory variables

Variables Mean Standard Correlations

Deviation (1) (2)

(l) Status, t to t-4 0.685(0) 0.971 (0.228)

(2) Number of Deals, t to t-4 91.876(0) 111.600(45.982) 0.793(0.645)

(3) Number of funds in year t 2.198(0) 1.683 (0.754) 0.608(0.233) 0.720(0.370)

Values outside of parentheses are based on within-firm and between firm-variance; values within parentheses are based on within-firm variance only.

are three reasons for including this control variable. The first two reasons are similar to the reasons for including the number of deals over a five-year time interval as a control variable. As with number of deals, number of funds is likely to be an indicator of both size and experience. To the extent that number of deals is an indicator of size, we wish to distinguish the status effect from the size effect. To the extent that number of funds is an indicator of experience, number of funds should have a negative effect on the average stage of a venture capital firm's investment.

The third reason for including this control variable is to rule out an alternative explanation for our hypothesized effect of status. We have argued that high-status firms should be more likely to make late-round investments because status is most valuable in those market segments in which a firm knows how to best leverage its status in forming exchange relations. However, if higher-status firms generally have access to more financial resources than lower-status firms, high status firms may be forced to make more late-stage investments because-with the noteable exception of biotechnology start-ups-early stage investments generally have minimal capital requirements. For example, the capital requirements for a first-stage software company, which is engaged in the development of a product, are typically much less than the capital requirements of a second- or third-stage software company, which must either put together or at least outsource manufacturing and distribution functions. Put simply, the high-status firms may simply have too much money to invest in early stage firms. We thus include number of funds as a measure of the amount of financial capital to which a venture capital firm has access. If the effect of status is spuriously related to access to financial capital, then the effect of status should disappear when number of funds is included in the analysis. This interpretation of number of funds as a proxy for access to financial capital suggests that the variables should have a positive effect on average stage of investment. On the other hand, to the extent that number of funds is simply another indicator of market experience, it should have a negative effect. Since we include this variable only as a control, we do not hypothesize as to whether the negative or positive effect should be stronger.

Finally, in addition to these two control variables, we include firm-specific effects, O'i, and period-specific effects, 't'" to denote effects that should be either firmspecific or year-specific. These effects simply control for unobserved heterogeneity that would be common either to all of the observations drawn from a particular firm or all of the observations drawn from a particular year.


Table 2. Determinants of average investment round

Variables

Status Number of dealS/IOO, t through t-4

(I)

0.054 (0.031)*

(2)

0.016(0.017)

Number of funds -0.026(0.010)**

(3) (4) 0.098(0.042)** 0.098(0.043)** -0.033(0.020) -0.016(0.021)

R2 for within-firm variance .048 .050 0.049 -0.026**(0.010) 0.052

N= 387; NT = 2386 * = p < 0.1; ** = p<0.05 Fixed-effects for fnms and years are not reported.

ANALYSIS

Table 1 presents descriptive statistics for the explanatory variables in the analysis, and Table 2 presents the regression analyses. The reason that we report correlations based on within-unit variance only is that the inclusion of fixed-effects for firms essentially removes cross-firm variance when estimating the effects of the other coefficients. Accordingly, in trying to interpret the substantive effect of the coefficients reported in Table 2, it is helpful to have the descriptive statistics based on within-firm variance only.

As we see in Table 2, the effect of status is positive and statistically significant,6 as hypothesized. Interestingly, the effect becomes stronger-almost doubling in magnitude-when the number of deals is included as a control variable. There is thus evidence of at least a mild supression effect; the status effect is weaker when the measures of learning through experience are not included in the model. This suppression effect lends at least tentative support to the view that higher-status firms are on average worse learners.

Comparing the effects of the two control variables, we see that number of funds making investments in year t has a stronger effect than number of deals undertaken between years t and t-4. Moreover, the negative effect for number of funds clearly indicates that scarcity of early-stage investments does not seem to be an issue for the firms with more financial resources.

The positive effect for status is particular noteworthy in light of the fact that each model includes firm-specific effects. Because these firm-specific effects remove all cross-firm variance, the regression actually shows us that within-firm changes in status are accompanied by within-firm shifts in the average stage of investment.

CONCLUSION

In this chapter, we have tried to distinguish between two types of market uncertainty-the altercentric uncertainty that a producer's constituencies confront when evaluating the producer as a potential exchange partner and the egocentric uncertainty that the producer confronts in making its own resource allocation decisions. Previous research has demonstrated that the value of status increases with altercentric uncertainty.


While this work does not provide direct evidence that social capital induced by status necessarily declines with egocentric uncertainty, it does show that the higher a producer's status, the more it seeks to avoid market segments that would be associated with high egocentric uncertainty.

A related paper (Podolny and Feldman 1997) takes the next step and analyzes how variance in egocentric uncertainty affects the ability of high-status firms to make resource allocations that are superior to those made by low-status firms. This related paper investigates the outcomes of the investment decisions made by the venture capitalists. Specifically, it considers the investments of venture capitalists at each stage and examines how a firm's status relates to the likelihood that the entrepreneurial firms in which it invests go public. The paper finds that the higher a venture capital firm's status, the higher the likelihood that its third-stage investments go public, but the lower the likelihood that its first-stage investments go public. In other words, in the latest stage, when egocentric uncertainty is low, the higher-status firm make resource allocation decisions that are superior to those made by their lower-status counterparts. But in the earliest stage, when egocentric uncertainty is high, the higher-status firms apparently make resource allocation decisions that are worse than their lower-status counterparts. By demonstrating that status yields the greatest social capital when egocentric uncertainty is low and that status seems to create social liability, as reflected by a learning disability, when egocentric uncertainty is high, the findings in Podolny and Feldman (1997) complement and reinforce the findings in this one.

Before concluding, we would like to briefly note some implications of this chapter for future research. First, because new markets or new market segments are likely to be characterized by especially high egocentric uncertainty, the results of this chapter suggest that high-status producers will generally be more conservative than their lower-status counterparts in moving into new markets or market segments Earlier we briefly alluded to some evidence consistent with this conjecture: highstatus investment banks were extremely reluctant to enter the junk market despite the apparent profitability of doing so. However, more systematic evidence is obviously necessary in order to validate this conjecture.

More broadly, the results of this chapter suggest a new direction for research that falls into the prismatic perspective on market networks. At the beginning of the chapter, we invoked the metaphor of a prism to characterize work that highlights how the pattern of ties among a set of actors on one side of a market induces a differentiated social structure. A number of scholars (Baum and Oliver 1992; Stuart, Hoang, and Hybels 1999; Zuckerman 1999) have illustrated how the affiliations and positions that arise as part of this structure affect the decisions of those seeking to evaluate and discriminate between the actors within this structure. However, we know much less about how an actor's location within the prism affects the actor's own ability to make decisions. Work following from the behaviorial theory of the firm (e.g., March 1988) has identified a number of dysfunctional learning behaviors in which firms engage. Moreover, there seems good reason to believe that decisionmaking in some organizations is more political than decision-making in others, where the term 'political' refers to the realization of individual self-interest through bargaining and informal social exchange within the organization. Given variance


across organizations in the efficaciousness of learning and the political character of decisions, it is valuable to ask how an organization's location within the prism affects the decision-making process. For example, how does the presence or absence of ties to legitimate actors affect the nature of decision-making within organizations? In pursuing research along these lines, sociologists will provide further insight into the functionalities and dysfunctionalities of the social structural underpinnings of the market.

We wish to thank Michael Hannan and Ezra Zuckerman for helpful conunents on this chapter.

NOTES

I. Following Knight (1921), it is important to distinguish the risk associated with the underwriting of securities that have a higher default rate than investment-grade debt from the uncertainty associated with underwriting in a new type of security market. Investment banks can in a rather formulaic way set the prices of the bonds to reflect the risk associated with underwriting and owning bonds with a higher default rate. However, because the junk market was a new securities market in the 1980s and the demand for and performance of bonds in these markets was not well-understood, investment banks also confronted some uncertainty in these markets that could not be simply 'price away' as risk. 2. Of course, in conducing an empirical test of this hypothesis, it is essential to control for the fact that higher-status firms will typically have larger reserves and thus are bener able to take on high-risk projects. 3. Importantly, a high-status firm need not be aware of its learning disability in order for it to act on the basis of that disability. Simple trial-and-error in segments with high and low egocentric uncertainty should lead the high status firms to those segments in which their learning disadvantage is less problematic. 4. Along these lines, it is worth noting that an interesting distinction between the way that economists and sociologists study markets is that economists will typically seek to expand the set of market decisions that can be framed in terms of risks, whereas sociologists will generally seek to expand the set of market decisions that can be framed in terms of uncertainty. 5. Arguably, the within-firm correlation is more relevant to the analysis because the inclusion of fixed-effects for firms removes cross-firm variance from the analysis. 6. Statements on statistical significance refer to a .I level.

•

Corporate Social Capital and the Cost of Financial Capital: An Embeddedness Approach

ABSTRACT

24 Brian Uzzi

James J. Gillespie

Using a structural embeddedness approach, we present argument and evidence on the ways social capital affects the operation of financial capital markets in the context of the small business loan market. We posit that the quality of a relationship between a bank and a corporate borrower, as well as the network structure of ties between the borrower and its bank(s) influences the cost of capital firms pay on their loans. Specifically we examine two dimensions of structural embeddedness at the dyad level and two at the network level. At the dyad level of analysis, we find that the duration of the relationship and relationship multiplexity are associated with a lower cost of capital (i.e., paying lower interest rates). At the network level, we find that firms that have ego-networks composed of a mix of embedded and arm's-length ties obtain a lower cost of capital then firms with either a ego network composed of arm's-length ties or an ego-network composed of only embedded ties. We find no effect for simple ego-network size on the cost of capital. The implications of our embeddedness perspective on corporate social capital are discussed.

INTRODUCTION

Economic sociology is concerned with questions of how organizations acquire resources and the mechanisms by which social structure influences the allocation of resources in a market. In a capitalist economy, the key resource is financial capital, and consequently, the connection between a firm and its lender(s) is equivalent to an organizational umbilical cord. The purpose of this chapter is to apply a sociological approach to the study of financial market intermediation with special attention to how social structure affects an organization's cost of borrowed capital.

Corporate Social Capital and the Cost of Financial Capital - 447

Currently, financial economic theory has developed a widely used model to explain how the market for capital operates and how the cost of capital is determined for organizations. The theory predicts that, in a frictionless capital market, individuals or firms with a positive net present value of investment opportunities will always have access to funds and that the riskiness of the borrower determines the cost of capital (Mintz and Schwartz 1985; Mizruchi and Stearns 1994; Petersen and Rajan 1994). This approach focuses on the use of objective financial criteria in evaluating the creditworthiness of a borrower and on how financial market characteristics affect lending practices. The theory posits that firm level financial statistics adequately measure the organization's ability to service debt through future revenues, liquidity of assets, or both. Thus, high performing firms or firms with high liquidity are prime candidates for receiving loans at competitive prices. In addition, the age and size of the organization are also viewed as important measures of the firm's ability to bear credit. Old and large firms are expected to receive better financing terms because they have positive reputations and a more diversified portfolio of assets.

The financial economic approach also focuses on how financial market characteristics influence lenders' bargaining power and historical practices. A significant characteristic of the market is the level of bank concentration in a given region: higher bank concentration is thought to increase the cost of capital to borrowers because the decrease in competition among banks can permit each to bargain aggressively for a premium (Petersen and Rajan 1994). Also regional characteristics are important. Certain regions may have structurally embedded financing and production cultures that increase access to resources relative to other geographic regions (Romo and Schwartz 1995). A California software firm may find it easier to acquire capital than a Mississippi software firm. Industries vary in their growth rate, which can provide organizations in these industries with an advantage in acquiring capital. A biotechnology firm may have more favorable cost of capital than a firm in a declining heavy manufacturing industry (Powell et al. 1996).

While evidence in support of this theory has been accumulating, particularly at the level of large banks and large corporations (Uzzi and Gillespie 1998), a recent critique argues that it fails to account for how social structure (e.g., lending relationships, discrimination, and bias) affects the cost of that capital (Mintz and Schwartz 1985; Podolny 1993; Petersen and Rajan 1994; Abolafia 1996). For example, financial economics generally regards relationships as peripheral to the operation of capital markets or as adding inefficiencies to the system (Blackwell and Santomero 1982; Baker 1990; Podolny 1993; Mizruchi and Stearns 1994). Yet, research has long recognized that relationships are an integral part of the banking system and highly valued by entrepreneurs and bankers (Baker 1990; Hoshi, Kashyap, and Scharfstein 1990; MacKie-Mason 1990; Diamond 1984). For example, consistent with the argument that relationships matter, Hoshi, Kashyap, and Scharfstein (1991) found that long-term ties between Japanese firms and banks was associated with fewer liquidity constraints on a firm's investments and a greater capacity to make investments when financially distressed. Petersen and Rajan (1994) found that the number of banks from which a firm borrows and the number of services the firm uses at the bank are associated with a lower cost of borrowing.


They concluded that ties between banks and borrowers can increase information flow and the bank's control over the firm's actions, thereby addressing problems of adverse selection (e.g., high interest rates attracting riskier borrowers) and moral hazard (e.g., applicants choosing higher risk investments).

This chapter uses an embedded ness approach to extend the above work and examine how social structure affects an organization's cost of capital, where cost of capital is reflected in the interest rate on long-term financing. The interest rate on long term financing is an appropriate measure of the cost of capital because 1) it is clearly measurable and 2) it is the cost index most widely used in financial reporting. Specifically, we examine the relationship between social structure and lending practices, with particular emphasis on the quality and structure of ties between small and medium sized organizations and their banks. Small and medium sized organizations have annual sales up to 500 million, with the medium size in terms of annual sales being ten million.

The embeddedness approach gives social structure a central role in explaining lending practices and outcomes (Granovetter 1985; Portes and Sensenbrenner 1993; Uzzi 1996a, 1997a, 1997b). It explicates how the substance of ties, as well as the ego-network of ties in which an organization is situated, affects exchange between organizations. The decisive factor is that particular types of social ties can mitigate opportunism, increase resource pooling, and motivate actors to seek Pareto superior outcomes rather than selfish gains. This theoretical approach extends previous work in economics and sociology by developing more fully the social mechanisms by which relationships benefit the firm and by furnishing more exacting measures of embeddedness than the current literature.

Integrally related to our embedded ness arguments is the role of social capital in the creation of financial capital (Gabbay 1995, 1997). 'Unlike other forms of capital, social capital inheres in the structure of relations between actors and among actors. It is not lodged either in the actors themselves or in the physical implements of production' (Coleman 1988: 98). Social capital consists of the social relationships between actors, and it importantly affects the operation of financial capital markets. Banks are obviously repositories of financial capital. In addition, much like venture capital firms, they are sources of experience, information, and personal contacts for young firms (Freeman, this volume; Podolny and Castellucci, this volume). Banks gain social capital by bridging structural holes and disconnects in the social structure of the financial market, principally serving to connect savers with borrowers. As this chapter shows, loan interest rates are partly a function of the strength of the relationship (i.e., the degree of existing social capital) between a bank and a potential borrower and the architecture of the firm's network of bank ties.

STRUCTURAL EMBEDDEDNESS: THEORETICAL ARGUMENTS

The structural embeddedness approach extends the work of classical sociological theory on the economy and combines it with organization and social network theory (White 1981; Granovetter 1985; Powell 1990; Portes and Sensenbrenner 1993; Romo and Schwartz 1995; Uzzi 1996a). The basic argument is that the nature of relationships between and among firms, as well as the overall structure of the egonetwork within which the firm is embedded, influences individual firm behavior and


the behavior of the network in its entirety. The type of network in which an organization is situated defines its potential store of strategic opportunities, while the quality of its relationships with other actors demarcates its capacity to access and implement those opportunities.

The embeddedness approach assumes that actors' interest and motives are variable and follow predictably from social structural parameters (Granovetter 1985). These differences in the microbehavioral foundations of embeddedness and the macro structural conditions of exchange are what distinguish the logic of embeddedness from other approaches (see Uzzi 1997a: 61). A key feature is that actors operate under what has been referred to as the 'logic of embeddedness' because ongoing social ties shape actors' expectation, motives, and decision-making processes in ways that differ from the logic of market behavior (Portes and Sensenbrenner 1993). According to this logic, actors use heuristic decision rules rather than intensive calculation to make decisions, and they aim to cultivate cooperative ties rather than narrowly pursue self-interest.

Structural embeddedness refers to the concrete social ties between and among actors and focuses on material exchanges of resources and information as the basis of the exchange. The argument posits that different structural conditions set in motion either self-interested or cooperative interests and motives among banks and borrowers, which in tum affect the cost and availability of capital. In addition, our arguments recognize that social structure can either facilitate (social capital) or derail (social liability) economic action. Therefore, 'social capital should be treated as a context-dependent concept calling for the definition of the conditions in which it has productive outcomes' (Gabbay 1997: 13-14). In an extreme case of overembeddedness, social structure can be more constraining than beneficial and constitute 'social liability' (Gabbay and Leenders, this volume).

We begin by discussing four dimensions of structural embeddedness: relationship duration, relationship multiplexity, ego-network size, and network coupling. These four dimensions all affect whether a given configuration of ties turns into social capital or into social liability for to corporate borrower. Working at the levels of dyads and ego-networks, we then examine how these four dimensions of structural embeddedness affect the pricing of loans to small business entrepreneurs. Important dimensions at the dyadic level are relationship duration and relationship multiplexity (Coleman 1988; Baker 1990; Podolny 1994; Uzzi 1996a). Important dimensions at the network level include ego-network size and egonetwork coupling (Baker 1990; Burt 1992; Uzzi 1996a), with the unit of analysis being the set of ties between a firm and one or more banks.

Length of Bank-Firm Relationship Relationship duration refers to the elapsed time in a relationship from the point of its inception. According to financial theory, a firm's age could provide prospective lenders with a gauge of the its ability to service debt by providing a record of the firm's creditworthiness with past employees, suppliers, and lenders (Blackwell and Santomero 1982). In contrast, recent arguments hold that the information learned about a borrower through a long-term relationship may include information about a borrower's creditworthiness that is not contained in past dealings with others. Of


particular importance is that the information may not be readily transferred to new banks that have different lending policies, practices, or experiences with borrowers. Thus, the longer a relationship between a borrower and a bank, the more likely it is that the bank will view the borrower as creditworthy relative to other borrowers (Petersen and Rajan 1994). Similarly, if a borrower has had only short-term relationships with many different lenders, banks could interpret that as a signal of credit unworthiness.

The embedded ness approach concurs with the insights of financial economics but argues that a relationship is not only the source of specific information about an exchange partner, but is the source of unique resources that would not be generated in the absence of the relationship. 'Social capital inheres in the structure of relations between actors and among actors. It is not lodged either in the actors themselves or in physical implements of production' (Coleman 1988: 98). For example, several studies of various types of exchange relationships have shown that ongoing interaction fosters trust between exchange partners as they learn one another's expectations and values (Coleman 1988). As trust accrues, the resources that were dedicated to monitoring an exchange partner can be redeployed, increasing the value of the tie. In a study of interfirm relationships, Macauley (1963) found that, the longer two firms transacted with one another, the less detailed were their contracts and the greater the organizations' ability to work out problems of transacting. Baker (1990) reported that long-term relationships between investment bankers, and investment bankers and their corporate clients, permits both parties to put faith in the contractual pledges of other parties. Repeated interaction has also been found to increase liking and the formation of 'business friendships' which can lower the likelihood of opportunistic behavior and increase the search for mutually beneficial outcomes (Homans 1950; Batson 1990). Baker (1990) showed that investment bankers are more inclined to look for Pareto improved outcomes when business friendships have developed, even when immediate or long-term payoffs are not apparent or guaranteed. The important outcome of these processes is that ongoing ties can lower costs, as well as the threat of opportunistic behavior by either exchange partner, because both parties are motivated to preserve the tie and to first look for Pareto-improved rather than self-interested outcomes.

In the case of banks and borrowers, this may mean that the bank can lower its monitoring costs and contract writing costs for long-term borrowers, and firms with ties can acquire cost of capital advantages relative to those lacking enduring social ties. Indeed, in our conversations with bankers, the motivational benefits of relationships are viewed as more important than the informational benefits because, in many cases, more than enough information on the credit history of the firm and the entrepreneur can be readily accessed through credit raters such as TRW (Uzzi and Gillespie 1998).

Multiplexity of Bank-Firm Relationship Another aspect of dyadic relationships that is important for understanding exchange dynamics is the degree of multiplexity. Multiplex ties are relationships in which persons are linked by more than one type of role (e.g., buyer and seller, business partner, friend, etc.). Coleman (1988) argued that multiplexity increases the overall


level of resources available to actors because resources from one dimension of the relationship can be appropriated for use in others. Multiplex ties develop when transacting partners can enact a set of relationships in addition to the immediate relationship of buyer and seller. In the context of banking-borrower relationships, multiplex ties are likely to form when the borrower relies on the bank for multiple services that span the roles of borrower and lender. Typically, these broader relationships include financial planning, personal credit card issuing, retirement planning, pension or payroll account services, lock boxes, letters of credit, etc. The presence of multiplex ties of this form constitute a type of social capital that should lower the cost of financial capital because resources from one dimension of a banking relationship can support other dimensions either through direct resource allocations or by expanding the possibility for finding compromise solutions to gaining credit. For example, in return for a lower interest rate or access to more credit -the primary resource desired by the firm-a business may use a bank's new service division or expand its use of services already offered by the bank.

Consistent with this argument, there is evidence that when a new relationship is added to a multiplex tie, the new ties rely on self-enforcement rather than external constraints to manage interdependence (Gimeno and Woo 1996; Nooteboom, this volume). This process has the effect of building interpersonal trust in multiple contexts and roles (e.g., norms of reciprocity as benefactor and as beneficiary). For example, research on automaker-supplier relationships suggests that, as US parts suppliers and Big 3 automakers moved towards more cooperative relationships, the level of contractual oversight over new relationships (e.g., supplier as co-designer or investor) decreased (Helper 1990; Dyer 1997). Building on Coleman's (1988) initial insight, Uzzi (1997a) argued that multiplex ties build redundancies that reinforce relationships and reduce the risks associated with exchanges. In risky situations, multiplex ties enable resource pooling and adaptation to random events by creating or increasing the level of slack resources in the relationship. This increases the likelihood of risk taking and investment on the part of both exchange partners.

These arguments suggest two hypotheses: HI: The duration of the relationships between the borrowing organization and

banks is inversely associated with the organization's cost of capital. H2: The degree of multiplexity in the relationships between the borrowing

organization and banks is inversely associated with the organization's cost of capital.

Structural embeddedness also operates at the network level of analysis. Two key measures are network size and network coupling. The logic behind these effects is that an organization's overall ego-network of ties affects the value of each dyadic relationship possessed by the firm.

Network Size Network size measures the quantity of ties possessed by an actor. Several perspectives argue that a large network of contacts is beneficial to an organization because it increases the organization'S bargaining power and access to alternatives. Transaction cost economics predicts that firms increase credit availability and lower the cost of capital by maintaining many ties to many financial institutions. From the


perspective of transaction costs theory, the more trading partners a firm possesses, the greater its probability of finding a prospective lender and the greater its bargaining power vis-a-vis each bank (Milgrom and Roberts 1996). From the perspective of resource dependence theory, a large ego-network of trading partners should lower the power asymmetry between financial institutions and corporate borrowers (Mintz and Schwartz 1985).

While we agree that a large ego-network of contacts may provide more opportunities to acquire capital for the reasons outlined above, we predict that it will have a negative rather than a positive effect on the cost of capital, particularly under the credit rationing conditions that small businesses typically face. Our argument is that large ego-networks work against the building of close relationships between a borrower and a lender for several reasons, and thus lower the bank's incentive to offer attractive rates. First, large ego-networks limit interaction because time and resources are spread across a large set of partners. This reduces opportunities for repeated interactions that can cause arm's length relationships to blossom into business friendships (Granovetter 1993). Similarly, small ego-networks signal to network members that the organization has enacted a strategy of cooperative exchange and problem solving by consciously restricting its bargaining alternatives (Kollock 1993; Jackson and Wolinsky 1996). If the number of banks that a firm uses is small, then it is likely to attempt to maintain a close tie with its bank in order to support the continuity of the relationship (Baker 1990). As Leenders (1995b) has noted, this dependence cuts both ways because there is a mutual interdependence between individual actors and network structure: Firms depend on banks as a key source of financial capital but banks depend on firms to provide a market for the purchase of their capital. We expect this effect to be strong in a well developed banking market like the US because lenders use their knowledge of the going rate of capital to bargain aggressively with borrowers who are shopping around since they know that other lenders will also bargain aggressively to maintain their spreads.

Consistent with this argument, several studies have shown that firms with large ego-networks were more likely to play their partners against one another (i.e., whipsawing) in an opportunistic manner (Helper 1990; Dyer 1997). Second, a large network reduces the economies of time advantages typically found in the close relationships that follow from small networks (Freeman, this volume). Smitka (1991) showed that the development time of new models, as well as speed to market was higher for Japanese automakers than for US auto makers partly because the smaller contractor networks of the Japanese firms enabled tacit knowledge to develop, which was crucial for faster decision making. In the context of our study, we anticipate that small businesses with larger ego-networks of banks will pay higher interest rates on their loans.

Network Coupling Our last argument relating ego-network structure to the cost of capital concerns the effects of the portfolio structure of an organization's network of ties. An egonetwork's portfolio structure differs from ego-network size in that networks of like size can have a dissimilar portfolio structure. An ego-network can have a dispersed, consolidated, or mixed-mode structure (Baker 1990; Uzzi 1996a). A dispersed


portfolio structure occurs when an organization spreads its banking business out among many banks in small parcels. For example, the organization may use six banks, each of which gets about fifteen percent of the organization's business. This would represent a disbursed structure. Conversely, an organization with the same overall size of banking business could use one bank for a sole-source relationship, or it could give one bank 95 percent of its business and another bank the remaining five percent. This structure would represent a consolidated structure. In between the dispersed and consolidated structures is a mixed or dual mode structure, wherein the organization sends the lion share of its business to one or two banks and then distributes the remaining share to two or three banks.

An organization with a consolidated portfolio gains the benefits of close ties, yet runs the risk of becoming insulated from new and novel information that is circulating outside its network (Uzzi 1997a). For example, firms that use only one bank may be unaware of innovations in banking or financial services or of new competitive loan rates or instruments that other banks in the industry are adopting or are first-movers in adopting. Over time, the accumulated effects of the social liability of this weak network position can put the firm at a strategic disadvantage for gaining timely market information about capital availability and cost, even if their primary bank is motivated to find Pareto-improved outcomes, because there is a lack of knowledge of alternatives.

Conversely, an organization with a dispersed portfolio can optimize a firm's access to new and novel information (Granovetter 1973; Burt 1992) but lacks the benefits of collaboration and resource pooling that are associated with embedded ties (Powell, Koput, and Smith-Doerr 1996). Under these conditions, firms are likely to be put at a disadvantage in garnering favorably priced loans because they lack the embedded ties to a bank that promote integrative bargaining and the search for Pareto-improved outcomes. The trade-off between consolidated and disbursed networks suggests that a middling level of network coupling provides the benefits of markets ties and embedded ties, while minimizing their disadvantages (Uzzi 1997a). When a borrower's ego-network contains an integrated mix of arm's length and embedded ties, the borrower is in a position to scan the market widely for innovations in banking and financial services, while remaining in close collaboration with a principal lender or two. For example, a borrower may learn of a new financial service (e.g., processing of credit card receipts, revolving credit arrangements, or employee retirement accounts) through its weak ties and then work with lenders it has close relationships with to develop these services for the firm at competitive rates. Consistent with these arguments, firms that maintain an ego-network with a dual mode structure have been found to gain efficient access to market information and to equalize power differences in the investment banking industry (Baker 1990) and to minimize the probability of failure in supplier-manufacturer networks (Uzzi 1996a). In the context of the capital lending market, we expect mixed ego-network coupling to have a similarly beneficial effect on the cost of capital.

These arguments suggest two additional hypotheses: H3: The ego-network size of the borrowing organization is positively associated

with the organization's cost of capital. H4: Organizations with a dual mode ego-network will obtain financing at a lower

cost than organizations with either a dispersed or a consolidated ego-network.


DATA AND METHODS

We use data from the 1987 National Survey of Small Business Finances, which was administered by the Board of Governors of the Federal Reserve and the Small Business Administration on a one-time basis in 1988-1989. The purpose of the study was to investigate the sources of borrowing of small businesses and how characteristics of the market, the firm, and the lending relationship affect the cost, availability and conditions of credit. The face-to-face administered questionnaire surveyed a random sample of 3,404 non-financial, non-agricultural small businesses operating in the US in 1989. Sample range covered firms with 50 to 500 employees and $1,000 to $154,000,000 in asset value; 1,875 firms were corporations and 1,529 were partnerships or sole-proprietorships. Nearly 90 percent of the firms were owner managed. The response rate was between 70 to 80 percent, depending on the item. This reduced the sample size to approximately 2400 cases. Respondents answered questions about the characteristics of their firms, including the quality, number, size, and duration of their lending relationships, sources of financing, and the conditions of their loans. In addition, the survey administrators collected some financial data on each firm for the previous year where applicable.

Variables Our dependent variable is the cost of capital, which we operationalize as the interest rate on the most recent loan received by the organization. Approximately 1300 firms reported having secured long-term financing from one of their banks in the form of a loan over the period from mid 1987 to 1989, the observation period of the survey. Length of bank-firm relationship is measured as the number of years the business has had done business with the bank. Multiplexity of bank-fIrm relationship is operationalized as the count of the number of services the business engages in with the lender. These include: Brokerage services, capital leases, cash management services, checking accounts, equipment loans, letters of credit, lines of credit, mortgages, motor vehicle loans, night depository, pension fund, processing of credit card receipts, retirement accounts, revolving credit arrangements, savings accounts, supplying money/coins for operations, trusts, and wire transfers.

Ego-network size is a count of the number financial institutions a firm uses for any of the above financial services. Some firms reported non-banks as possible sources of financial capital but did not report having a loan from them. Because the potential for receiving capital was possible from these non-banks, we included these reported non-bank sources in the ego-network size variable because some theories view these potential sources of capital as an important dimension of price competition in the banking market for small businesses (Pfeffer and Salancik 1978).

Ego-network coupling measures the level of consolidation in a firm's egonetwork of ties to banks they do business with. Consistent with previous studies, we operationalize this measure using a modified Gibbs-Martin index of heterogeneity (Baker 1990; Uzzi 1996a). It is calculated by summing three sources of business a firm dedicates to its banks: the amount of cash in checking, the amount of cash in savings, and the size of the line of credit. For each firm, we summed these three sources and then added the sums across all banks. This permitted us to calculate the percentage of each firm's business that is dedicated to each of its banks. For


example, a firm may have a total of $100,000 of 'banking business' to distribute across the three banks it interacts with. Suppose it makes the following allocation: The first bank gets $50,000 of its business (i.e., 50% because $50,000/$100,(0) and the second gets $30,000 (i.e., 30%) of its business, and the third gets the remaining $20,000 of its business (i.e., 20%). Our consolidation index is calculated by taking the sum of the s~uared percentages. For our hypothetical example, the index would be .38 (i.e., 52+.3 +.22). This Gibbs-Martin index ranges from just above zero to one. A value close to zero indicates a dispersed network; a value of one indicates a perfectly consolidated network; and values between .4 and .6 indicate a dual mode network (Uzzi 1996a). Our prediction is that a middling level of coupling will result in the best cost of capital for firms. In our model, we represent this by adding network coupling and network coupling squared into the equation. A negative and statistically significant value on the linear term and a statistically significant and positive value on the squared term would suggest support for the hypothesis.

Following financial theories, we control for important firm, loan, and market characteristics that affect loan interest rates (Petersen and Rajan 1994). Financial ratios are widely used to determine the credit worthiness of a business. Creditors are primarily interested in the firm's short-term liquidity (i.e., the ability to quickly convert assets and other resources into cash) and its long-term ability to service debt. We use two standard financial ratios to operationalize a firm's credit worthiness, the acid (quick) test ratio and the debt ratio (Gitman 1979). The acid test is computed as the firm's current assets minus inventory, divided by current liabilities, and the debt ratio is computed as firm's total liabilities divided by total assets (i.e., the proportion of total assets provided by the firm's creditors). Other firm-level factors controlled for are organization age (log transformation) and organization size (log transformation of number of employees). For the subs ample of firms analyzed, the mean age was 14 years (with a range from 6 months to 105 years), and the average number of employees was 25 (with a range from 1 to 475). We include two controls for the characteristics of the loan. Collateral measures whether a firm pledged physical assets as security in the loan agreement. In case of default, the bank can seize the collateral, sell it, and apply the proceeds towards satisfaction of the firm's obligation. Term spread controls for differences in interest rates attributable to different loan maturities. It is calculated by subtracting the Treasury bill yield from the yield on a government bond of the same maturity (Peterson and Rajan 1994). Finally, we include four controls for financial market characteristics. The first is the prime rate. The prime rate is the interest rate banks charge to their best customers and serves as the pegging rate that banks use in pricing commercial and consumer loans. The second control is the level of bank concentration in the local area: Areas with high concentration contain one or very few financial institutions; areas with low concentration contain many financial institutions. The Federal Reserve provided this variable in an ordinal form (3=high, 2=medium, 1=low concentration). The higher the level of concentration of banks in a region, the less competition there is among banks and the more power they have to set rates (Peterson and Rajan 1995). Finally, we include indicator variables to control for the census region where the small business is located (Northeast, North Central, South, and West) and the


industry in which it operates (using two-digit Standard Industrial Classification codes), since both of there variable are thought to affect interest rates.

We employ a Tobit regression model to analyze the effect of lending relationships on the cost of capital because the interest rate variable cannot take on values below zero percent or above the value set for usury in our sample. Tobit analysis is appropriate for estimating models on this type of limited dependent variable because it will not estimate values out of the range of truncated values as would OLS, and because it produces unbiased and efficient estimates (Maddala 1983; Baba 1990; Roncek 1992).

RESULTS

Table 1 presents the results of our Tobit regression analysis. The overall model was statistically significant at the 0.00 1 level. As expected, we find that many of the firm and market level control variables are predictive of a firm's cost of capital. Older firms received loans at lower interest rates. This finding suggests that banks value older, more established firms. Age of the firm appears to carry important information in a market context that is not fully substituted by the duration of the tie between the organization and its bank.

Not surprisingly, the prime rate was a good predictor of the loan interest rate. The lower (or higher) the prime rate, the lower (or higher) the interest rate on the loan. Firms that pledged collateral with their loan received lower interest rates on those loans. Firms located in areas with a high bank concentration (i.e., areas with less competition among financial institutions) had higher capital costs. Two of the four regional indicator variables were statistically significant; none of the seven SIC indicator variables were statistically significant.

The results from the exogenous variables are broadly consistent with our expectations. Consistent with hypothesis I, relationship duration is inversely related to the cost of capital. The longer a small business and a financial institution have been interacting, the lower the interest rate tends to be on the firm's loan. (The average duration of relationships was 13 years, with a range of I to 95 years). Consistent with hypothesis 2, the greater the degree of multiplexity in the relationship between the bank and the firm, the lower the the cost of capital. Thus, businesses maintaining mUltiple connections with their financial institution performed well in the competition to secure capital at favorable interest rates. (The average number of multi pie x ties was 2.6, with a range of 0 to 14 ties).

Hypothesis 3 was not confirmed: Ego-network size did not have a statistically significantly effect on the cost of capital. (Average ego-network size was 2.2 banks, with a range of 0 to 12 banks). One possible reason for this may be that an average size of just over two banks does not give firms much bargaining power or the ability to shop the market widely enough. It may also be that the effect of network size has no net effect once the quality of the relationship and the distribution of ties with the network has been controlled for (Uzzi 1996a).

Consistent with hypothesis 4, a dual mode network of embedded and arm'slength ties is positively related to a lower cost of capital. Small businesses maintaining either only arm's-length ties or only embedded ties put themselves at a


Table 1. Tobit Analysis Predicting Interest Rate on Most Recent Loan, National Survey of Small Business Finances, 1989

EXOGENOUS VARIABLES Coef. (s.e.)

Structural Embeddedness

Length of bank-finn relationship -.012* (.006)

Multiplexity of bank-finn relationship -.043* (.017) Ego-network size .003 (.122)

Ego-Network coupling of lending ties -3.264** (.985)

(Ego-Network coupling of lending ties)2 2.828** (.874)

Firm Characteristics

Acid ratio -.012 (.011)

Debt ratio -.120 (.117)

Log of age -.154* (.069)

Log of employment .001 (.005)

Market Characteristics

Collateral -.343* (.170)

Term spread .095 (.060)

Prime rate .286** (.031)

Bank concentration in MSA .246* (.100)

Reg I -.332 (.189)

Reg2 -.362* (.185)

Reg3 -.379* (.184)

Sic1 -.882 (.631)

Sic2 .133 (.211)

Sic3 -.310 (.212)

Sic4 .252 (.360)

SicS -.269 (.161)

Sic6 -.445 (.284)

Cons 9.700** (.539)

*p < .05, **p <.01 (all tests two sided) n = 1308

Prob > X2 = 0.000

Log-likelihood = -2890.08

disadvantage in the competition to secure capital at favorable interest rates. These results suggest that the distribution of exchange within a network plays an important role in determining which actors garner the potential benefits of a network of relationships. In comparison to dual mode networks, networks that are overly dispersed or overly consolidated are relatively less effective in shaping market exchanges with trading partners than are dual mode networks.

DISCUSSION

This chapter asked the question: What logic governs economic exchange between financial institutions and small businesses? Seeking to broaden our understanding of capital market dynamics, we proposed an integrated social capital and network


embedded ness approach, arguing that the quality of the relationship between a bank and an organization and the architecture of organization's network of lending relationships shapes the cost of financial capital. In this view, the cost of capital is not based solely on general market conditions, firm-specific financial ratios, or net present values of investment opportunities (although these remain important). Rather, the quality and mix of arm's-length and embedded ties between a bank and a business create new value in the relationship and increase the flow of information.

Specifically, we advanced four hypotheses to test our arguments. We found that small businesses garner loans at lower interest rates by increasing the duration and multiplexity of their relationships with a financial institution. We also found that businesses can most successfully lower their cost of capital by constructing an egonetwork portfolio that includes the proper mix and intensity of ties to financial institutions. Finally, we found that a simple measure of network size had no effect on the firm's cost of capital. These results offer evidence in favor of the embeddedness thesis and suggest that economic exchange is not only embedded in ongoing social ties but that such ties produce outcomes that add to the benefits of market transacting.

The implications for corporate social capital theory are evident. Ties with financial institutions need not yield favorable interest rates. In this chapter we have shown that the mix of duration, multiplexity, and volume of these ties and a medium amount of network coupling derive social capital from these ties (lower interest rates). But we have also shown that, for example, a high level of network coupling can create social liability. This type of research can shed light on the question of under what conditions social structure produces social capital or social liability.

Our results also have implications for the economic sociology of financial markets. Consistent with Podolny (1994), we found that relationships offer an alternative way for firms to manage uncertainty and improve upon pure market outcomes. In comparison to simple market ties, embedded relationships appear to transfer useful information about the firm's ability to service credit. The idea that a mix of consolidated and disbursed network connections may be highly functional is expressed in Baker's (1990) research. He found that 'hybrid interfaces' are most effective at exploiting power advantages and reducing resource dependence because they combine the advantages of'relationship interfaces' and 'transaction interfaces,' without many of their disadvantages. Our work differs slightly from Baker's on the issue of whether parties are motivated more by the tradeoff between power and efficiency or by the informational and bargaining relationship benefits of dual mode ego-network structures. Baker stresses power considerations, as opposed to the value of relationship building. Consequently, we would reverse Baker's order of priority: Dual mode network coupling offers a way to maintain high-quality relationships, while retaining the important benefits of atomistic markets. In our study, this theoretical reversal is logical given the improbability that small firms can ever gain true power advantages over the large banks. Another difference is the strategic intentionality and agency implicit in Baker's idea. Our approach argues that network coupling is an emergent property of actors' attempts to balance social and market imperatives. And that the consequences of attempts to balance ties is always imperfect because actors are in a web of ties, some of which are beyond the actor's


direct control. Nonetheless, creating a network with the proper degree of network coupling requires sufficient knowledge of the market. Yet, mixed coupling is only feasible when lenders trust the borrower enough so that the borrower can maintain a non-consolidated network and when the borrower trusts the lenders enough to not have to rely on disbursed ego-networks (Uzzi 1996b; see Uzzi and Gillespie 1998 on the formation of bank-firm relationships).

This chapter also fits into recent efforts to develop a broader understanding of the social processes surrounding entrepreneurship. The success of small businesses and start-up firms is dependent on more than the personal traits of the entrepreneurs or the financial characteristics of their business (Becker 1964). Success crucially depends on gathering resources and information via networks extending beyond the boundaries of a particular firm or individual entrepreneur (Gabbay 1995, 1997). Similarly, network ties provide informational cues that outsiders make inferences upon. When a struggling small business gets capital at a competitive rate, this can serve as a signal of legitimation to other exchange partners who rely on banks to evaluate the financial wherewithal of firms. Our analysis shows how structural embeddedness plays an integral role in the process by which social capital (e.g., a strong bank-firm relationship) is used to acquire financial capital. We further speculate that the acquisition of this financial capital can contribute to building social capital (e.g., increasing the firm's legitimacy and reputation). Thus, there is a cumulative, reciprocal relationship between financial capital and social capital.

Future research might examine the processes of building and deconstructing lending ties and how these processes are shaped by market characteristics. One might expect that borrowers start with a large ego-network (trying to maximize the probability of securing a loan) and then, as time passes and loans are acquired, they gradually decrease the size of their network. A firm's ego-network size may decrease over time because trust can only be gained through enduring and repeated relationships. To address these propositions, there is a need to track the evolution of specific firm ego-networks over time. Too frequently, social network analysis consists of static snap-shots where network structure is assumed to be unchanging (Leenders 1995b). Our chapter suffers from this weakness, so future research should examine how bank-firm networks change over time. In addition, because we examined lending relationships from the perspective of small businesses, as opposed to financial institutions, future research could profit from a more in-depth analysis of the supplier side of the loan market.

In the past twenty years, the pace of change in the financial services industry has been without parallel-typical forms of market exchange and bank control have diminished in importance or been supplanted by other exchange logics (Davis and Mizruchi 1997). Although more research is needed on the economic sociological dimensions of capital markets, this chapter has attempted to specify the mechanisms and forms of embeddedness that shape lending relationships.

We express appreciation for the valuable comments offered by Roger Leenders and Shaul Gabbay on an earlier version of this chapter. We also gratefully acknowledge the assistance of Mitchell Petersen, Woody Powell, John Wolken, and the Board of Governors of the Federal Reserve System. Please direct correspondence to Brian Uzzi.

•

Venture Capital as an Economy of Time

ABSTRACT

25 John Freeman

Entrepreneurship is a social activity in which resources used to build new organizations are acquired through the social relations that combine to create a community structure. An important participant in this community is the venture capital firm. These organizations provide funding for new ventures and also help build the new ventures' social capital. They do this by making social connections to other important actors and by providing advice. Venture capital organizations are constrained by shortages of time that rise in severity as the venture capital firm's centrality in the community increases. As a result of these shortages of time, there is an inverse correlation between central location in the community and willingness to work with the entrepreneur whose venture is struggling. The most central venture capitalists are expected to display greater impatience with their portfolio ventures leading to different outcomes for those ventures defined in terms of three liquidity events: acquisitions, failures, and initial public offerings.

THE SOCIAL NATURE OF ENTREPRENEURSHIP

The purpose of this chapter is to present the beginnings of a sociological analysis of entrepreneurship. Starting a business organization, the defining act of an entrepreneur in this treatment, is a social act. Organizations by definition include more than one member, so of course starting an organization involves attracting these participants and organizing them. A second social aspect of the phenomenon is a consequence of the resource gathering activities of the entrepreneur. These resources usually come from other organizations.

Venture Capital as an Economy of Time - 461

Such observations are commonplace, yet many writers ignore the social context of entrepreneurship, taking an heroic, individualistic view of the subject. The problem with this approach is that when people do things in groups, and rely on actors outside those groups for support, individual acts of heroism (and villainy) are difficult to discern. So the social nature of the activity tends to distract attention from a romantic story.

This individualistic perspective on entrepreneurship has led many scholars to focus on the psychological properties of entrepreneurs. They are described as less risk averse than most people, as having higher needs for achievement, and they are often found to have a more internal locus of control (Brockhaus and Hurwitz 1986). This is consistent with the treatment one usually finds in the business press, which tends to focus on how some entrepreneur overcame various obstacles to build a great enterprise. A sociological treatment of entrepreneurship does not deny that entrepreneurs do great things. Rather, sociologists tend to look elsewhere for causal explanations in order to preserve the falsifiability of the explanation. Much apparently depends on the social capital members bring to those teams and on how well people work together. The second observation noted above is that cooperative activity spans emerging organizational boundaries. The entrepreneurial team almost always receives support from non-members who themselves usually work for other organizations. All business organizations require such mundane resources as raw materials, power, water, waste disposal, mail and banking services. They also require customers or clients. Most also need sponsorship and endorsement. All of these are provided by other organizations and by individuals who are members of those organizations. So the entrepreneurial effort is embedded in a network of social relations that provide material support, information and legitimacy to the nascent organization (Birley 1985; Aldrich and Zimmer 1986). The resources that comprise this support are not randomly allocated in society and there is much to be learned by studying the allocation and accumulation processes of this social capital (Stinchcombe 1965; Aldrich 1979).

The channels of allocation are important mechanisms structuring opportunity at the societal level and producing varying rates of social mobility. They are especially important for ethnic minorities and immigrant groups. They channel capital and human activity into areas of economic expansion. They affect the speed with which society responds to changing technology and demography. Finally, of course, they affect the pattern of organizational variability. Entrepreneurs struggle to solve problems of organizing so as to commercialize new technologies, or market new products or services. Successful organization invites imitation. New forms of organization are thus created and their populations proliferate.

Successful new organizations grow but how much they grow and the time path their growth follows is not well understood. Some become giant corporations. Most fail. Some organizational forms include smallness as a design characteristic. Such organizations often reach their maximum size at opening (e.g., diners) while others grow slowly but may achieve very large size (e.g., insurance companies). At some point, organizations outlive or outgrow the state of disorganization that usually follows startup, or they fail. Sooner or later they stop being entrepreneurial in the


sense of being 'young' or 'new.' When this happens is difficult to define but almost certainly varies between organizational forms.

Bounding the trailing edge of entrepreneurship is difficult no matter what definition one employs. This is partly because defining when a pattern of change slows down, and how much it has to slow down before one decides to stop paying attention, is problematic. It is also difficult because different forms of organization become institutionalized at different ages and at different developmental stages. For some kinds of organizations, an initial public offering of stock (IPO) constitutes a sea change that is irreversible and organizationally significant. After an IPO, companies must be audited and report the results publicly. This constrains their internal information and decision processes. They are forced to become more orderly. But some kinds of organizations go through an IPO when they are quite young, before they begin to sell products much less generate profits. Biotechnology companies are such organizations. These companies often retain their useful disorganization, looking very much like Mintzberg's (1983) adhocracies long after they have 'gone public.' They may grow quite large without developing the managerial functions that are fundamental to most corporations. For example, since biotechnology companies usually have no products in the market, they have no marketing function. Their financial officers may be scientists working as part time fund raisers. Financial controls are often primitive. This is because their organizational models are university research laboratories. The point here is that the problem of defining ending dates for the period of entrepreneurship is theoretically challenging, and this is true whether or not the research defines entrepreneurship as we do here or in some other way.

In the research reported below, three terminal events are used to bound the phenomenon. These events would probably be deemed worth studying by most scholars even if they do not accept the definition of entrepreneurship we have chosen here because they are the most common paths to liquidity. That is, they are events through which investors are able to recapture invested funds: acquisitions (ACQ), initial public offerings (IPO), and failures (FAIL). The first two are theoretically important because they involve the sale of the company, not just its assets. The organized entity has value above and beyond the desks, computers, inventories and other tangible assets. This value is partly dependent on having previously solved the organizational problems that confront anyone trying to employ the technology those tangible assets were assembled to exploit. When those assets are sold off, and the organization disappears, we simply refer to these organizations as failed. These three events are ordered according to their usual payoff to those who own equity in the company. An IPO produces higher returns than an acquisition, which produces higher returns than a failure.

VENTURE CAPITAL FIRMS AND THE ENTREPRENEURIAL COMMUNITY

As organizations move through time, from a gleam in someone's eye, to an established company, an internal process occurs through which roles gain definition, routines for operating the business are developed, authorities for acting on the organization's behalf are specified. Physical plant and equipment are acquired and


put into operation. Nelson and Winter (1982) call this 'learning by doing.' Along the way, with hard work and good luck, the company generates revenues from its operations (rather than purely from equity financing and borrowing), begins to generate positive cash flow, and finally, turns profitable. Of course, many business organizations never get this far, but if they last, most go through some internal evolution of this kind. Since this is not the main subject of study here, we will continue noting as an aside that the speed with which this process occurs, the stages that can be discerned and the problems managers and others encounter along the way vary from one form of organization to another. The more technical and organizational innovation is involved, the more challenging this process is for managers and the greater the hazards associated with failing to negotiate the process successfully. Starting an innovative organization is more difficult, and hazardous, than starting an organization following a well-established form (Hannan and Freeman 1984; Freeman 1986).

At the same time, organizations go through a process of developing relations with their environments. No organizational theorist would deny this. They would certainly debate how it is done, who does it, and why. These relationships vary in longevity. Given the time scale that is appropriate for the organizational form in question, these relationships usually last long enough to take on the properties of relational contracts (Williamson 1985). Following the language of social network analysis, any business entity, which we can call Ego, develops around itself a set of relations with Alters. In real time, these relationships develop serially. Through them flow money, information, and access to still other relationships. When strung together, these ego nets form a network that is the structural basis of a community. The flows of money, information and access constitute channels of opportunity. The structure of the network, then, constrains opportunity (Larson 1992). It makes entrepreneurial activity easy for some and nearly impossible for others (Van de Ven 1993). The advantages these social structures provide to an entrepreneur are 'social capital' in Coleman's (1990: 300-321) sense of the term.

It is almost certainly true that the structures of some of these communities facilitate the process of starting new business organizations while other structural arrangements impede it. Indeed, Silicon Valley in California is famous for providing the support that technology-oriented entrepreneurs require (Saxenian 1994). When these structures operate efficiently, the internal process described above can be expected to work faster as well (Florida and Kenney 1988). Some of the information that flows through this network describes the routines for organizing and managing new ventures. Some of the access involves finding key people who know what routines to employ and how to make them operate when required (Bygrave 1988).

This chapter focuses on just one form of organization found in such communities-Venture Capital firms (VCs). VCs are management firms set up to invest capital in young ventures. This capital is provided by investors who contribute to a venture capital 'fund.' The fund is organized as a limited partnership that liquidates at a known date, distributing the proceeds to the partners. Since the funds have a planned life cycle, the VC often raises multiple funds, staggering them over time so as to balance the payouts and manage the workload they generate. The classic definition of a VC is based on their investing activities in young, rapidly


growing companies (Bygrave and Timmons 1992; Barry 1994). But some VCs finance corporate restructuring processes such as Management Buyouts. In this chapter, we focus on the more classical definition and confine attention to their activities with entrepreneurs.

Young ventures often develop through a sequence of investment events called 'rounds' of financing. The VCs involved in these rounds often change. This serial quality to investment practices encourages careful and sometimes sober reviews of the company's progress, or lack of progress. One of the important contributions of venture capital investors is this periodic review. It forces the entrepreneurs to look back to their previous claims and promises, and focuses attention on the dynamics, not just the current state of affairs.

Venture capitalists frequently state that their consulting services are just as important to the new venture as is the funding. Typical is the following statement, taken from a promotional brochure provided by one of the most famous venture capital firms, the Mayfield Fund:

We Offer A Global Network of Contacts

Mayfield has close working relationships with technology leaders, universities, other venture capitalists, financial institutions, consultants and corporations throughout the United States, as well as an international network ... Mayfield's contacts provide a key resource for developing the relationships critical to a growing technology-based company, including potential corporate partners, both here and abroad.

Because of our long association with a large number of successful companies and entrepreneurs, a relationship with Mayfield is highly regarded. It can enhance the credibility of a young company with potential customers, vendors and employees, and with other financial institutions.

In other words: Mayfield not only provides financial capital, but also provides social capital to the entrepreneur.

Community structure develops like any social structure out of the differential tendencies for some to associate with others. These relationships are implied when venture capitalists and other actors participate in rounds of financing together. In the vernacular, they 'do deals' together. When such actors provide services to an entrepreneurial venture, they usually seek information on the identities of the other parties. Syndicates of venture capital firms are formed in much the same way Stuart (this volume) finds in the alliances developed among semiconductor firms. For example, some venture capitalists have strong preferences for working with certain accounting firms and will refuse to invest if another accounting firm provides consultation and auditing services to the entrepreneur. Venture capital firms select other venture capital firms. Those currently investing in a young company often recruit and select other VCs for participation in subsequent rounds of financing. Since their currently illiquid investments are made more or less valuable by the actions of subsequent investors, they care who these investors are and they care about the terms of the deals that are negotiated for their entry to the group of investors.

One of the organizing principles underlying these associational preferences among VCs and other members of the community is faith in the reliability of others


to perform, one of the important functions identified by Nooteboom (this volume) as characteristics of individuals and organizations acting as go-betweens. This attribution of reliability is a form of trust, and is based in part on having done business together in the past. This is not the same thing as endorsing the other's ethics, though some minimal level of ethical behavior is surely implicit. The point here is that having been involved in the creation of a large number of ventures establishes any actor in the community better than if that actor has only been involved in a few. There are obvious status distinctions among venture capital firms much like those Podolny has studied among investment banks (Podolny 1993). If it is true that the community structure channels opportunity, then the more centrally located actors should provide better access to entrepreneurs they support. Implicitly, they deny easy access to others. The chapter by Podolny and Castellucci (this volume) argues that a high status organization has more and better access to resource providers than does a low status organization. A point completely in harmony with their view is that being connected to a high status ally has this same effect. Below, we will argue that when the high status other is a venture capital firm, its ability to provide such access is limited by the time it takes to do so. For these organizations, time is a very scarce resource.

In addition to these relationships between organizations, it is useful to note that the entrepreneurial community is a very personal society. Managers, board members and other individuals working for or with these ventures bring their own resources to the organizations (Fried and Hisrich 1995). In particular, the venture capitalist brings personal experience and contacts to the young venture. This is an example of Granovetter's embeddedness concept (Granovetter 1985). These personal contacts are a form of social capital that provides value to the entrepreneur by making it easier to move the venture through the organization-building process. So we would expect better connected venture capitalists to provide services that improve the performance of the portfolio ventures. We need to qualify this statement, however, because the strategic position of the venture capital firm affects its relations with entrepreneurs.

THE ECONOMY OF TIME

Venture capital firms are highly competitive organizations. They are structured to provide very powerful incentives to their managers, incentives that encourage them to seek out young ventures that have the potential to grow rapidly. These incentives also encourage the venture capitalist to seek out investors who are willing to take risks and remain illiquid for lengthy periods of time. An early name for venture capital was 'patient money.' In the industry's early years (from 1946 to about 1975), VCs received the bulk of their funds from wealthy individuals and families such as the Rockefellers and Mellons. More recently, they have drawn from pension funds, insurance companies, and endowments. They succeed in drawing new ventures, called 'deal flow; on the basis of their prior records of success with other young ventures. These great successes compensate for the many unsuccessful ventures they also back, deals that draw substantially less attention from the press.

The other resource that forms the basis for competition is the capital invested in the funds VCs raise and manage. Investors evaluate the VCs on the internal rate of


return (lRR) , the annually compounded rate at which the proceeds from liquidity events among their portfolio companies accumulate. The more famous the VC, the higher its IRR tends to be. Investors willingly invest in the funds managed by people who have been highly successful in the past. VCs whose last fund produced a low IRR often fail to attract capital to their next fund.

Both of these resource flows depend on the company's past performance which in turn depends on a minority of investments in very successful young ventures (Bygrave and Timmons 1992). A successful VC firm produces an IRR of about 30% per year. Performance tends to be approximately log-normally distributed. Many ventures fail or are barely profitable (from the VC's point of view). A few are enormously successful. When such successes occur, they make the person who managed the investment famous. The attribution of a Midas touch draws the two resources under discussion here. Such enhancements to market position are sometimes called the 'Mathew Effect.' (Merton 1964; Podolny 1993) Working against such advantages are the limitations of time available to the ventures capitalists themselves. The more famous the VCs are, the easier it is to attract money, but generating very high performance requires time for analysis, time for monitoring the young venture's progress, and time to provide the mentoring and contact-building that are so essential to the nonmonitory contributions VCs make.

The younger the venture, the more extensive these services are likely to be and the greater the risk. Risk emanates from the fact that many strategic and organizational problems wait to be solved and also from the illiquidity of the investment. Failure to solve these problems can lead to slow growth or even to outright failure. Slow growth is as bad as failure for the VC because of the fixed time cycle for their funds (described above) and because their performance is evaluated in annualized terms.

Only some of these activities can be delegated to staff members. Investors are paying for the judgement of the prominent ventures capitalists who lead the venture capital firm. Salaried staff can work through the analysis required by due diligence but the judgment required to chose high potential ventures is not subject to routinization. It often depends on extensive personal contact with the entrepreneurs. Similarly, the investors conduct their own due diligence analysis on the VC and its senior managers. Such analysis often consumes great amounts of the VC's time. So the size of the venture capital firm is limited by the time of its senior managers.

The VC is paid in two ways, neither of which involves fees from the entrepreneur. First, there is a management fee charged on capital invested, usually between 1.75% and 2.25% per year. When a fund is raised, limited partners commit funds. As ventures are identified, those commitments are drawn upon and the VC starts charging a management fee on them. If there is no flow of deals, there is no management fee. So there is pressure to invest the fund quickly. The second way in which the VC makes money is by collecting a share of the profits when the fund liquidates, called the 'carry rate,' usually about 20%. Notice that operating expenses for the VC do not figure in these reward mechanisms. Costs are deducted from the revenues generated by these two forms of payment and are not captured in IRR calculations. So venture capital firms do not win competitively by being more cost efficient. Rather, they win competitively by being more time efficient.


The VC looks for entrepreneurial ventures that have the potential to grow rapidly, seeking liquidity that precedes the terminal date of the fund. The preferred source of liquidity is an IPO, a public sale of stock. Often, the IPO is followed by a secondary offering in which the VCs liquidate their holdings. So the VC is not interested in many of the ventures seeking funding, even when they are quite viable businesses. The venture has to have the potential to grow big enough, fast enough to warrant a public offering. The investment banks who organize those offerings by underwriting the stock sale have fixed costs that virtually prohibit offerings of less than $35 million. So the venture has to be big enough to justify such a valuation.

The younger the venture, the more difficult it is to assess its growth potential. Consultation to support the growth at young companies is also more time consuming. Once the VC is committed, illiquidity creates dependence and the VC responds by doing whatever is required to make the venture grow. This includes providing managerial services, such as negotiating with lenders and major customers, recruiting management and technical talent, and finding facilities (Bruno and Cooper 1982). The VCs often serve on the young venture's board of directors. Sometimes, they fire the entrepreneurs. When venture capitalists do this, they have to find replacements. So most VCs shun ventures with glaring management weaknesses. Fixing such weaknesses costs them time they would rather spend identifying high potential prospects.

Venture capital firms are not expensive organizations to run. Their capitalization is not high. They do not require costly equipment. Their staffs are often modest in size. Since the profits of the firm are divided amongst the partners, there is an incentive to expand the funds relative to the number of general partners. This process is limited by the time these general partners have for analyzing new ventures and helping them grow. Therefore, it is not money that is in short supply for most of them, it is time. The larger the fund, the more critical this problem is because the number of deals is correlated with fund size.

The evaluation of venture capital performance by the internal rate of return exacerbates the time problem because the IRR is time-valued. That is, the return is compared with the rate of return that would have been produced if the same funds had been committed to some easily managed highly secure investment such as government bonds or bank certificates of deposit. Given the risks involved, venture capitalists are expected to produce much higher returns. So evaluation of the VC's performance is not based simply on whether the fund cashes out at the end of its time with a handsome surplus, evaluation depends on how long funds were under management and what the return was over that time. So there is intense pressure on venture capitalists to bring ventures to liquidity quickly.

There are two ways to deal with this shortage of time. The VC can restrict its investments to older ventures that are closer to an IPO. Because the risk is lower and the information is better, such an investment strategy allows fewer, larger investments. Of course, such investments produce more modest returns. The second way of dealing with this shortage of time is to follow the lead of other VCs who have spent the time required to perform the due diligence. This strategy requires that the focal VC trust the judgment of the lead VC. Trust here means faith in the ability of this other VC, or set of VCs, to perform such analysis. In this sense, the followers


use exactly the same logic as their investors employ-they commit capital based on the sagacity of some other investor. Why doesn't everyone just do this? The answer is that not everyone can get a piece of every deal. One VC informs another about ventures that need more funding, and they have influence over the prices that are charged for stock in that venture. This argument follows those developed by Podolny and Castellucci (this volume) and Stuart (this volume) who both argue that superior social status increases the choices organizations have for establishing relationships with others. So it is with deal flow. VCs that are central in the community and have high status can reciprocate with access to ventures for which they have performed the due diligence. This involves technical expertise as well as time. So one VC may specialize in biotechnology, and include in its ranks general partners who have advanced training in fields such as biochemistry, but still invest in ventures that are not biotechnology companies by trading access to deals with VCs knowledgeable in other areas. In other words: social capital generates social capital.

Because of the greater time pressure on VCs managing multiple large funds, we can expect less patience from them. They are looking for 'home runs,' huge successes. Modest successes are not disasters, but these central players in the community cannot maintain their elite status earning modest rates of return. To earn the 30%, compounded annually, that is expected of the elite, they need great successes. So ventures that are moving forward at a slow pace are often abandoned. It is the time to monitor the ventures' activities and guide them through the entrepreneurial community that is the main cost to the VC. The more deals the VC does, the more connections it has with other organizations in the community. If this is true, we can expect network centrality of VCs to accelerate the liquidity process. This leads to the following hypothesis:

The more central a venture capital firm is in the entrepreneurial community, the more impatient it is likely to be with its portfolio ventures.

Impatience is displayed in a higher rate of acquisition and failure among the ventures in the portfolios of the most prominent VCs.

MODELS AND ESTIMATION

As indicated above, we are interested in three kinds of events that constitute outcomes for entrepreneurial ventures: initial public offerings, acquisitions, and failures. Ventures do not experience more than one of these simultaneously, but they can experience them sequentially. That is, a company may sell stock to the public and then be acquired or fail. Of course, the scope of the study changes fundamentally if one focuses on the entire life courses of such organizations. Our interest here is on their period as entrepreneurial ventures. In addition, it is also important to note that when any of these events occurs, the relationships of the venture with its associates in the community are altered. If a company is acquired, its relationships become subordinate to its parent corporation. If it conducts an IPO, it must be audited and its strategic relationships with key customers or suppliers must be disclosed. In general, then, these events endogenize the community


relationships of greatest interest here. So terminating the analysis when one of the three events occurs makes sense. The three events represent changes between discrete states and changes can occur between these states at any point in time. Accordingly, we study the probability of their occurrence using continuous time models. We want to estimate the time-dependent process by which young ventures move from startup to a liquidity event. We do this by estimating the hazard rate, as a function of duration (organizational age). The methods for conducting such analysis, often called event history methods, are by now well known (Tuma and Hannan 1984). We note here that the hazard rate can be defined as

( ) . Pr~ ~ T < t1T ~ t) r t = hm--=----.,,-...l--..:..

1'-1 t -t

the ratio of the transition probability to the length of the duration in a state such as illiquidity. It can be described (following Blossfeld and Rohwer 1995) as the propensity to change from the state of illiquidity to the state of liquidity at time t. We will estimate such a rate for each of the three liquidity events conceptualized as 'competing risks. ' For each kind of event, the other two are treated as censored observations.

To assess the effects of exogenous variables on such a rate, we need to specify a parametric model that considers time observed as duration. organizational age in this formulation. Such an analysis requires the assumption of some form of the timedependence in the process. When previous research and theory produces good bases for specifying this time dependence. a wide variety of models with different patterns of time dependence is available. Since we do not have such a body of research to fall back on for the rate of IPOs and the rate of acquisitions. we begin with a piecewise exponential model. This model allows us to treat organizational age nonparametrically by estimating a constant for each of a set of time periods. So if the effect of age on the rate is irregular or nonmonotonic. we can evaluate the causal factors of interest without fear that observed effects will be artifacts of the time dependence. The piecewise exponential model performs this estimation with a structure similar to the use of dummy variables in OLS regression.

The time span over which organizations age is divided into a set of separate periods

tp = 'f. +'f2 +'f3 + ... +'fL;

and the rate from origin state 0 to destination state d is

In this expression A(od) is a row vector of covariates and a.(otl) is a row vector of coefficients and 1/ is an index of time periods tp. After exploring the nature of the duration dependency. we then estimate the simpler exponential model which substitutes a single constant for the constants that are specific to each time period in the model above.


The regressors present in the hypotheses are presented in three groupings: venture capital firm centrality measures, measures that describe the venture itself, and control variables that describe economic circumstances. Some of these variables change over the time period under study. In order to allow for such changing regressors, we employ the common tactic of spell splitting (Tuma and Hannan 1984) in which the time period over which a venture is observed is divided into quarterly subspells. Each subspell is treated as censored on the right except for the subspells ending with a liquidity event.

Venture capital centrality measures describe client-service provider contacts: CjNV investment firm degree centrality C_ VC venture capital firm centrality C_NVC centrality of non-VC investors. These centralities are the number of entrepreneurial ventures in which the focal

investors have put capital at the study's end (1995). To understand how this relates to the usual network degree centrality measure (see Knoke and Burt (1983) for a review) we note, first, that venture capital firms never invest in each other nor do entrepreneurial ventures invest in each other (or at least this is very uncommon). Our data, described in detail below, were drawn from the ranks of venture capital-backed ventures. We use the term 'investment firm' here because venture capital funds are managed in a variety of ways. Some are large firms as described above. Others are venture investing arms of corporations. Still others are small investors with funds contributed only by a few wealthy investors. They may call themselves venture capitalists, or they may simply refer to themselves as private investors. We begin this analysis assuming that the 'impatience' argument applies to all investors and that the shortage of time is a function of network centrality. Then we expand the analysis to see if VCs classically defined and other investors differ in conformity with the impatience hypothesis.

Venture characteristics-business strategies and locations. Because we are not interested in the effects of these variables in their own right, and the theories prompting their inclusion in the model are obvious, we simply list them below with their expected effects.

Log Age: Any of the three events can occur at any organizational age, but since all are stochastic, and we follow them over time until one of the events occurs, the probability that each will occur cumulates. So we would expect each to be a positive function of the organization's age (here the natural logarithm of age);

CAPITAL: The data tell us how much capital was raised in each round but not how much each VC invested; this variable is the sum of the amounts previously raised as of any date.

Some of the ventures in our data are not young, but small businesses acquired by entrepreneurs when they seek venture capital to support a strategic change. We should take this difference into account and do so with a dummy variable,

ENTRY: Assumes a value of 1 if the venture started before January 1, 1981; 0 otherwise.

Financial performance is notoriously difficult to measure for entrepreneurial ventures. They do not have to disclose the data that one would use for such analyses and their structures and operating procedures are so volatile that such data would be


hard to interpret anyway. We use stage of operation instead. When ventures start out, they are usually developing their product or service. Then they make prototypes and put them in 'beta test' mode, distributing them to potential customers for tryouts. If such tests are successful, the product or service goes into distribution. Finally, if the market responds well and production can be expanded to efficient levels, the venture becomes profitable. For this analysis, we compare those that are not yet shipping product, those in development or beta test mode, with those that are shipping or are profitable. Performance should be positively correlated with the probability of an IPO, and negatively correlated with the probability of a failure. It is difficult to predict the effect on the probability of an acquisition since acquisitions sometimes are fire sales, in which the investors are simply recouping whatever they can from their investments. Sometimes, however, the company is acquired at a premium for its technology and other high potential assets.

BETA: Assumes a value of I if the venture has a product in beta test, 0 otherwise.

SHPG: Assumes a value of I if the venture is shipping a product, 0 otherwise; PROF: Assumes a value of 1 if the firm reports that it is profitable, 0 otherwise.

In addition, we want to control for the following fixed regressors: SV Silicon Valley Location within region running from Menlo Park to San Jose,

California. Most observers would expect a high success rate for Silicon Valley ventures. So the probability of an IPO should be higher, and the probability that the venture will FAIL should be lower when the venture is in Silicon Valley.

Industry: The events we are studying are affected by a wide variety of industryspecific factors such as competitive conditions. We control for industry with the following dummy variables:

CDPR Data processing C CSO Computer software CCOM Communications 1_ CEL Consumer electronics I_IEQ Industrial equipment CMED Healthcare and biotechnology C TST Test and analytical instrumentation CCMP Components CSEM Semiconductors CRET Retailing CENV Environmental CCHM Advanced specialty materials and chemicals.

See Table 1 for the proportions of observations in each industry. National level economic conditions R_IPO Number of IPOs occurring each month R_BFAIL Number of business failures each month R_BINCORP Number of incorporations each month A V _TBILL Average treasury bill yield A V_PRIME Average prime interest rate.


DATA

The data used to test these hypotheses were provided by the San Francisco company, VentureOne, which has tracked venture capital activity in the United States since 1987. VentureOne provides data to venture capital firms that the ves use to make investment decisions. The ves provide information to VentureOne about the ventures in which they invest. VentureOne also monitors the business press looking for information on new ventures. They then send questionnaires to the ventures seeking information on the time of founding, markets in which the ventures are active (or markets in which their management intends to be active), sources of funding, location, et cetera. It is important to note that both the ves and the ventures have an interest in reporting as VentureOne's data is often used by investors to identify companies that might be seeking funding in the future. VentureOne follows these ventures over time, sending questionnaires to update their files yearly. This process stops when one of the liquidity events described above occurs.

Data were provided in the form of tables describing the ventures themselves, the rounds of financing they have conducted, and the investors in those rounds. There are 4,073 records on entrepreneurial ventures and 22,068 records on investors in which each record represents an investor funding a venture in one or more rounds of financing. The data tell us how much was raised in each round of financing, and which investors participated, but not how much each investor put into the round. These data were examined to remove name misspellings and obviously erroneous records. For example, investors named 'Individuals' or 'blankco' were treated as missing values. A record was deemed incomplete when there was no start date or city of location. When an investor could not be matched with a venture, the record on the investor was deleted. This yielded an effective sample size of 4,064 ventures, along with statistics describing the investors and service providers with which they are associated.

Table 1 presents descriptive statistics on this sample of ventures. The variable AGE is reported in quarters. It ranges from January of 1907 to October of 1995. The last date of observation is October of 1995. There were 83 ventures that were born in the month before the data were provided by VentureOne. Of course, these are all right-censored (still in existence at the time of the study). Approximately two-thirds of the cases are either shipping product or are profitable as of their last observation.

RESULTS

We begin with a simple version of a piecewise exponential model that allows for changing regressors. Our objective here is to see if the time dependence is monotonic. The time periods are defined from 0 to 7 quarters, 8 to 15, 16 to 23, 24 to 31, 32 to 39, 40 to 47, 48 to 55,56 to 63 and 64 and above. The effects of these period constants are negative, increasing toward zero the older the venture is. In the last period or two, it turns negative again. Further analysis shows this is because of a small number of left-truncated cases. When a dummy variable representing ventures begun prior to 1981 is introduced, the effects of duration appear monotonic.


Table 1. Descriptive statistics

Variable N* Mean Std Dev Minimum Maximum

ACQ 4064 0.16 0.37 0.00 1.00

FAIL 4064 0.09 0.28 0 .00 1.00

IPO 4064 0.24 0.42 0.00 1.00

Starting Time 145133 30.02 36.39 0.00 354.00

Ending Time 145133 31.01 36.39 0.01 355.00

AGE (in quarters) 145133 31.02 36.39 0.01 355.00

CENS 145133 0.99 0.12 0.00 0.01

SV (Silicon Valley) 145133 0.12 0.33 0.00 0.01

CAPITAL 145133 6.51 16.71 0.00 438.00

BETA 145133 0.03 0.16 0.00 0.01

PROF 145133 0.35 0.48 0.00 0.01

SHPG 145133 0.35 0.48 0.00 0.01

C_INV 145133 302.63 330.79 0.01 2.14

C_VC 145133 240.07 260.65 0.00 1.60

C_NVC 145133 62.56 117.74 0.00 775.00

CDPR 145133 0.21 0.40 0.00 0.01

CCSO 145133 0.14 0.34 0.00 0.01

CCOM 145133 0.10 0.31 0.00 0.01

CCEL 145133 0.01 0.11 0.00 0.01

UEQ 145133 0.06 0.24 0.00 0.01

CMED 145133 0.07 0.25 0.00 0.01

CTST 145133 0.02 0.15 0.00 0 .01

CCMP 145133 0.04 0.18 0.00 0.01

CSEM 145133 0.03 0.16 0.00 0.01

CRET 145133 0.07 0.26 0.00 0.01

CENV 145133 0.02 0.12 0.00 0.01

CCHM 145133 0.01 0 .12 0.00 0.01

COTHER 145133 0.23 0.42 0.00 0.01

R_IPO 138903 110.91 65.82 0.00 283.00

R_BFAIL 138903 5148.00 2043.00 458.00 9143.00

R_BINCORP 138903 55434.00 7505.00 21034.00 65691.00

AV_TBILL 138903 8.47 1.67 0.06 01.4

AV_PRlME 138903 8.96 2.64 0.05 02.0

EARLY 145133 0.30 0.46 0.00 0.01 . . . . * For the first three vaflables thIS IS number of orgamzattons; for the rest It IS the number

of firm-year combinations.

In Table 2, and those that follow it, we report various probability values associated with the estimated effects. The critical value in use is .05, and this is what is meant when the effects are treated as 'statistically significant.' We can see that the effect of being in Silicon Valley does not have the expected statistically significant positive effect on the rate of IPOs. Surprisingly, its positive effect on the rate of


Table 2. Piecewise exponential model with competing risks

ACQ FAIL !PO

Coef. (s.e.) Coef. (s.e.) Coef. (s.e.)

Period I -9.0947*** (0.5049) -8.3369*** (0.4156) -7.4850*** (0.2224) Period 2 -6.8071 *** (0.1837) -6.5876*** (0.1993) -6.3934*** (0.1431) Period 3 -5.8340*** (0.1355) -5.6463*** (0.1502) -5.4136*** (0.1043)

Period 4 -5.2968*** (0.1226) -5.4457*** (0.1527) -5.1257*** (0.1009)

Period 5 -5.1188 *** (0.1266) -5.2196*** (0.1547) -4.8286*** (0.1004)

Period 6 -4.4314*** (0.1120) -5.1199*** (0.1700) -4.6288*** . (0.1054)

Period 7 -4.1078*** (0.1177) -4.7285*** (0.1763) -4.1445*** (0.1073)

Period 8 -4.3035*** (0.1617) -4.3019··* (0.1932) -4.4892*·· (0.1558)

Period 9 -4.9382*·* (0.1350) -5.4591 *** (0.2107) -5.1686*** (0.1240)

CAPITAL 0.0023 (0.0018) 0.0019 (0.0027) 0.0124*·· (0.0003)

SV 0.3562··· (0.1067) 0.5179··* (0.1347) 0.1058 (0.0960)

BETA 0.3006 (0.2882) 0.1588 (0.3146) 0.8915**· (0.1882)

SHPG 0.4256·** (0.0937) 0.0274 (0.1132) -0.0990 (0.0948)

PROF -0.4928*** (0.1119) -2.6390·*· (0.2895) 0.4596··· (0.0830) Events 639 343 955

.** p < .001, X2 = 2452.2 df = 42.

acquisition and failure is statistically significant. Even in the most complex model, reported in Table 5, the positive effect of SV on FAIL remains positive and statistically significant. Apparently, the many famous success stories of the area, the 'home runs' sought by so many VCs, mask a lower batting average. The three performance dummies, BETA, SHPG and PROF, produce results that make more sense when the model is fully specified, so we will discuss them below.

Table 3 presents a simpler, exponential model, employing the dummy variable for left-truncation, EARLY, and dropping the piecewise specification, and adding the logarithm of AGE. The main effects of interest are the centrality measures. VC centrality has a positive effect on the rate of IPOs, but it also has a positive effect on the rate of acquisition, consistent with the impatience hypothesis. The effect on the rate of failure is positive but not statistically significant. These results stand up to controls as we complicate the model (below).

The effects of centrality of other investors are not statistically significant for ACQ and FAIL, but are negative for the rate ofIPOs (the latter effect does not stand up to the addition of more detailed controls, however). So the impatience hypothesis is supported when centrality of VCs is analyzed, but not when centrality of other investors is analyzed. An interesting question is whether these effects reflect the superior ability of central VC firms to choose winners when they invest, or whether they nurture their portfolio companies better. We return to this issue in the next section of this chapter, but at this point we simply note that if these empirical patterns were generated by higher perspicacity of centrally positioned venture capital firms, rates of acquisition and failure would be lower for their portfolio companies.


Table 3. Exponential model with competing risks estimating effects of investor centrality

ACQ FAIL IPO Coef. (s.e.) Coef. (s.e.) Coef. (s.e.)

Const -11.2846*** 0.2914 -10.3697 *** 0.3696 -9.8033*** 0.2270

L_AGE 1.9051 *** 0.0808 1.6125*** 0.1051 1.5170*** 0.0637

CAPITAL -0.0082* 0.0031 -0.0057 0.0042 0.0117*** 0.0003

SV 0.2627* 0.1089 0.4984*** 0.1375 -0.0020 0.0983

BETA 0.1388 0.2885 0.0585 0.3154 0.6867*** 0.1879

SHPG 0.3173*** 0.0941 -0.0736 0.1138 -0.2187* 0.0953

PROF -0.5601 *** 0.1136 -2.6900*** 0.2905 0.3914*** 0.0834

C_VC 0.0006*** 0.0002 0.0003 0.0002 0.0004*** 0.0001

C_NVC -0.0002 0.0003 -0.0008 0.0005 -0.0011 *** 0.0003

EARLY -2.7616*** 0.1582 -2.7187*** 0.2204 -2.5509*** 0.1298

Events 639 343 955

* p < .05; ** P < .01; *** p < .001, X2 = 3159.6 df= 30.

The analyses reported in Table 4 add controls for the industry and economic conditions. The effects of the VC centrality measures are the same, positive for both IPO and ACQ, not statistically significant for FAIL. All three effects are not statistically significant for the non-VC investors.

There are a number of interesting changes among the other variables comparing these estimates with the previously presented tables. First, CAPITAL now has statistically significant effects for each of the three events. Its effects are negative for both acquisitions and failures, but they are positive for IPOs. This makes sense if investors can forbear from throwing good money after bad. Location in Silicon Valley continues to show no statistically significant effect on the rate of IPOs, and the effect on failure is positive. Now, however, there is no statistically significant effect on ACQ.

The three performance measures have effects that make sense. Looking first at the rate of acquisitions, the only statistically significant effect is profitability, which is negative. When we shift to the next column, that for the rate of failure, both SHPG and PROF have statistically significant negative effects. So if the venture is shipping product or turning a profit, it is less likely to fail. Being in beta test has a negative effect, but it is not statistically significant. For IPOs, the pattern strengthens. This time profitability has a positive effect on the rate of IPOs as it should. So ventures that are profitable are more likely to experience an IPO and less likely to fail or be acquired.

The economic macro variables and industry identifiers are present for control purposes. Some of the effects are curious, such as the positive effect of A V _ TBILL yields and negative effects of A V _PRIME. It is not obvious why the number of IPOs in the previous quarter should have negative effects on all three rates.

The industry dummies show some interesting effects. Being in data processing (CDPR) exposes a venture to a very high failure rate, a high probability of being acquired, and a large, negative effect on the probability of an IPO. A similar, though less pronounced pattern can be seen in the effects of CCSO (computer software) and


Table 4. Exponential model with competing risks: economic conditions, industry and early founding controlled

ACQ FAll.. IPO

Coef. (s.e.) Coef. (s.e.) Coef.

Const -72.8680*" 3.3974 -84.3657**· 5.2624 -59.3934***

L_AGE 1.5062*·* 0.0829 1.1354*** 0.1012 1.0999***

CAPITAL -0.0208**· 0.0038 -0.0187*** 0.0053 0.0110·**

SV 0.2133 0.1100 0.4433·* 0.1405 0.0311

BETA -0.3939 0.2930 -0.4135 0.3215 -0.0048

SHPG 0.0883 0.0961 -0.2968* 0.1168 -0.2867**

PROF -0.6276·*· 0.1154 -2.7948··· 0.2924 0.3919***

C_VC 0.0009·*· 0.0002 0.0005 0.0002 0.0003*·

C_NVC 0.0006 0.0003 0.0001 0.0005 -0.0002

R_IPO -0.0109*** 0.0012 -0.0101*** 0.0016 -0.0149***

R_BFAll.. 0.0010·*· 0.0001 0.0012*** 0.0001 -0.0001

R_BINCORP 0.0008*** 0.0000 0.0009*·* 0.0001 0.0008***

AV_TBILL -0.4502*** 0.0704 -0.1832 0.0999 -0.0169

AV]RIME 1.4468**· 0.0664 1.5717*** 0.0989 0.4233**·

_DPR 0.4567*** 0.1318 0.9743*** 0.1826 -0.3974***

_CSO 0.5865*** 0.1229 0.5910** 0.1954 -0.4908***

_COM 0.4412** 0.1405 0.6593** 0.2076 -0.0968

_CEL 0.0896 0.3898 0.3207 0.5941 -0.0291 _IEQ 0.3420 0.1827 0.9518*** 0.2428 -0.6015**

_MED 0.1821 0.1771 0.3748 0.2431 0.2734*

_TST 0.9586*** 0.2122 0.6053 0.4023 -0.3966

_CMP 0.2247 0.2399 1.1727*** 0.2663 0.0093

-SEM -0.4640 0.3184 0.6257* 0.3119 -0.2219

_RET -1.0667*** 0.2812 0.3070 0.2444 0.3084

-ENV 0.4634 0.3284 0.5387 0.5183 -0.0357

_CHM -1.1977 0.7151 0.1585 0.7198 -0.4266

EARLY -1.3541*** 0.1452 -1.6581*** 0.2221 -1.4462·**

Events 639 343 955

* p < .05;·· p < .01 ; •• * P < .001. X2 = 8044.3 df= 81

(s.e.)

2.2090

0.0637

0.0004

0.1012

0.1906

0.0947

0.0860

0.0001

0.0003

0.0013

0.0001

0.0000

0.0604

0.0588

0.1191

0.1177

0.1105

0.2835

0.1894

0.1214

0.2840

0.1953

0.1748

0.1637

0.3220

0.3609

0.1277

CCOM (communications). Since many of the more famous recent entrepreneurial successes come from these industries (e.g., Cisco Systems, Oracle), it is interesting to note that success is not the rule.

A PORTFOLIO MANAGEMENT PERSPECTIVE

Suppose the impatience hypothesis is true and works primarily through the superior deal flow of the more central VCs. Perhaps all that is happening is that more central VCs, searching for the extraordinary performing investments, behave like options traders. They seek highly risky investments, knowing that risk and expected return are inversely related. To economize on time, such investors might make many high


risk investments, walking away from the losers. Of course such behavior might very well undermine the deal flow advantages that centrality provides. It would also damage the VC's maintenance of its network position and, consequently, the VCs stock of social capital. In addition, Of course, for this to be true the market would have to be characterized by long term imperfections. Otherwise, the rate of return would exactly compensate for the heightened risk and a strategy specializing in high risklhigh return investments would have no greater expected return than a random investment strategy. Nevertheless, such a view should be explored because it renders the impatience hypothesis less interesting. One would expect to see a positive effect of centrality on the rates of acquisition and failure simply because the central actors choose to invest in higher risk ventures.

To explore this possibility, we change unit of observation to the investment spell, rather than studying the venture's duration. As noted above, ventures often raise capital repeatedly in 'rounds' of investment. If an investor is pursuing a high risk and abandonment strategy, the standard deviation of the duration of its investment spells should be large. Very rapidly growing young ventures will either experience an IPO quickly, or raise successive rounds of financing in quick succession. Very poorly performing investments lead to quick exits. On the other hand, an investor pursuing a more cautious approach will tend to choose ventures that display steady but slower growth. Rounds come farther apart. The probability of abandonment is lower. If it is true that centrally connected investors choose higher risk investments, then we should observe that the standard deviation of the time between investments in a venture, or between an investment and a liquidity event should be greater. The mean time may not be different, but the standard deviation should be higher.

Shifting the unit of observation in this way poses numerous challenges. In particular, we can anticipate problems with autocorrelated disturbances, reflecting a lack of independence between observations. For instance, if there are multiple investors with funds in a venture, and that venture experiences a liquidity event, then all investment spells terminate at the same time. One is not free to end independently of another. Clearly, this problem is sufficiently different that a satisfying analysis will require a new modeling effort. For present purposes, we simply calculate the product-moment correlation between investor centrality and the standard deviation of investment duration.

Since we are interested in manifestations of impatience, and a single investor may participate in repeated rounds, we should look for the correlation in question controlling for the patience signified by repeated investment. So we measure the time duration for each investor putting funds into each venture, and count whether this is the first, second, third, etc. round in which that investor has participated.

The results do not support this portfolio interpretation. That is, the correlation between investor centrality and the standard deviation of investment duration is negative, not positive as one would expect from this portfolio view of the problem. It is worth noting that this is, at best, a suggestive result. It does not solve the autocorrelation problem. Such a task, while worth the effort, goes well beyond the purposes of this chapter or the space available in this volume.


Table S. Investor centrality and duration of investment

Zero-order r* Round 1 Round 2 Round 5 Round4 Round 5

Centrality and SD of Duration -0.139 -0.177 -0.159 -0.173 -0.194

Centrality and Mean of Duration -0.195 -0.202 -0.191 -0.204 -0.203

Mean of Centrality 118 131 145 158 167

Number of Observations 12546 6677 3157 1206 354 * P < .0001 for all coefficients

Investors participating in multiple rounds are likely to be more central than those participating in only one or two rounds. Both the mean duration and the standard deviation of investment duration are inversely correlated with investor centrality. One might be tempted to think that more central investors have access to higher quality investment opportunities or that they simply pick winners more often, but we should recall that the effects of centrality reported in Table 2 through 4 do not suggest such a simple explanation.

DISCUSSION

This research shows that the connections ventures have with investors do seem to matter for the liquidity events under study. Receiving support from centrally connected venture capitalists has the expected positive effect on the probability of an IPO. This effect is consistent with intuition and also fits the arguments advanced here that the superior network connections of prominent VCs help move the young ventures to become public companies. The high quality social structure the VC has built, thus yields social capital for the starting company. Of course, the superior deal flow that these more prominent VCs see allows them to pick the most promising ventures. In fact, the ability to choose well is one of the qualities that leads investors to put capital into their funds. On the other hand, venture capital firm centrality also increases the probability of acquisition. This suggests that the advantages of working with a well-connected group of investors are tempered by their likely higher impatience. This is consistent with the common view presented in the business press that venture capital works for the entrepreneur only so long as that entrepreneur's interests are aligned with those of the Vc. After that, the ties with the VC, that brought the entrepreneur social capital in the past, may start to produce social liability to the entrepreneur. This alignment depends on rapid growth. Sadly, firms sometimes grow faster than their managers can learn.

The centrality effects of non-VC investors are generally not statistically significant. This further supports the arguments advanced earlier in the chapter, particularly the impatience hypothesis. If the effects of VC centrality were artifacts of some variable such as size of fund, we would expect to see similar effects among the funds run by universities, corporations and state governments. An important difference is that such organizations are not driven by the same liquidity demands as are the classically defined VCs. This is because their capital is not organized in funds with fixed time horizons.

It is obvious that this analysis has employed only the simplest of network techniques. There are many other centrality measures. In addition, we have not yet


explored the similarities in network connections among the investors. The effects of industrial specialization have not been explored here, for example. One particularly intriguing subject for further study is the tendency for some VCs to invest in early rounds rather than later rounds of financing. There is much discussion in the business press and the scholarly literature to the effect that such tendencies affect the ability and Willingness of the VC to serve as a guide through the community, to provide consulting services as well as money. In addition, some firms act as the 'lead investor' and this affects their willingness to invest in subsequent rounds, and the ability of others to invest in the ventures for which they have taken leadership.

Finally, the principal unit of analysis in this study is the focal venture. The community itself may be an equally promising subject for study. In some social systems, the mechanisms for allocating opportunity depends heavily on kinship and ethnicity. In some circumstances, technical training is important while in others work experience is the key to admission (Fiet 1991). The efficiency of this system may explain why entrepreneurship happens more often, and perhaps with greater probability of success, in some communities rather than others.

If we ask why we have entrepreneurs at all, part of the answer is bound up in the economy of time that drives the working activities of venture capitalists. A highly efficient opportunity structure can allocate capital, create social connections, provide expertise, and link a new venture with suppliers and customers very quickly, probably faster than the same functions can be performed in a large corporation. Social structure then efficiently breeds social capital. Viewed in this way, entrepreneurship is a set of activities that links individuals, organizations and social context. The entrepreneurial community, an important part of this social context, produces social capital and social liability in the form of barriers and constraints and operates through a set of intensely personal connections. Money talks. But so do trust, reputation, access and legitimacy. All of these factors add value, leveraging the hard work, intelligence and good luck that lead to entrepreneurial success.

I received help from several sources in conducting the research reported here and in preparing this report. The San Francisco company VentureOne provided the data and allowed me to use it for research purposes. Lindy Archambeau contributed research assistance. Jerry Engel and Mario Rosati made suggestions about the institutional realities of venture capital. Bill Barnett, Glenn Carroll and Mike Hannan read an earlier draft. Thanks are also owed to the members of Stanford's joint organizations/strategy workshop. Their comments helped me make numerous improvements. Of course, my ability to generate ignorant substantive assertions and stupid technical errors remains undiminished.

• Final Issues

csc: An Agenda for the Future

•

INTRODUCTION

Roger Th.A.J. Leenders Shaul M. Gabbay

Corporate social capital refers to the set of resources, tangible or virtual, that accrue to a corporate player through the player's social relationships, facilitating the attainment of goals. In the opening chapter to this volume, we suggested that social structure and social capital are related, but distinct, entities. Both reside at various levels of analysis-in particular, at the levels of the individual and the firm-and interact with each other at these various levels of analysis.

The overarching agenda of the chapters in this volume thus pertains to the relations between social structure and goal attainment of corporate players. Social networks in this framework are discussed in the context of their functional roletheir positive or negative effect. We are only at the initial stages of the development of an encompassing theory of corporate social capital; this chapter is directed towards future steps. What the effect of corporate networks is, how to manage them and how to avoid social liability and create social capital instead, are all questions grounded in the study of corporate social capital. In the current chapter we will discuss the research and practical applications for further developments. We will highlight critical questions that, in our view, should be resolved and, in so doing, we will open more new questions for future discussions. At its initial stages, CSC is an emerging research agenda. This, of course, presents a wide open window of opportunity for future and further contributions.

AN AGENDA

The study of corporate social capital is an exciting area. Using the 'lens' of corporate social capital, we have an explicit way in which to study effective and ineffective organization. Because we are only at the outset of an encompassing theory of the


interrelationships of social structure and corporate social capital and liability, many fundamental questions are still unanswered. If CSC theory is to develop into a fruitful way of studying organizations, we are challenged to address several topics first. In this section, we will address four of these challenges. Many other issues are also of interest, but these four are among the most critical at this stage. These four challenges concern 1) the measurement of corporate social capital, 2) the conditions under which social structure conveys social capital or liability, 3) the effect of time on social structure and its outcomes, and 4), from a management-oriented perspective, the issue of how social capital can be created and maintained. We will briefly discuss each challenge below, posing questions rather than answering them.

The challenge of 'Measurement' Discussing social capital is one thing; measuring it is quite another. In empirical studies, social capital is often equated with some aspect of the social structure-for instance, the number of ties an actor has, or the centrality I of his or her network position. Alternatively, social capital is often measured by the existence of structural holes in a player's network structure.2 This is a helpful approach, because data on network ties is relatively easy to collect and the researcher has a large body of network measures and methodology at his disposal.

This approach, however, is not consistent with the view of social capital we have laid out in this volume, especially in our introductory chapter. We do agree that social capital and social structure are (strongly) associated. However, because we explicitly regard social capital as inherent in social structure, as an outcome and generator of social structure, it is inappropriate to equate the two. Consequently, an alternative approach is needed to measure social capital.

This is easier said than done. If social capital resides in social structure and can take on many different shapes, how would one measure it? Or, more appropriately, can social capital be measured at all? Could one justifiably claim that 'I have twice as much social capital as you'? It is not possible to provide a satisfactory answer to these questions at this point. Finding ways to measure social capital and social liability is one of the major challenges of esc research.

In some situations, measurement is relatively straightforward. For instance, some organizations reward their agents based on their agents' sales volume, which, in turn, is an indicator of the productive value of the agents' social ties. In this fashion, firms are in fact rewarding (and measuring) their agents' social capital.3

In most cases, it is difficult to measure social capital. In this volume, two papers explicitly address this measurement issue. Han and Breiger concern themselves with the question of in what aspects of social structure social capital is located. If social capital is inherent in social structure, the question is whether there is one specific structural characteristic that yields social capital, or whether we need to look for a mixture of characteristics. Rather than equating structure and capital, they search for the structural sources of social capital. Han and Breiger approach this question by studying models that produce estimates for the number of ties between pairs of actors, using three indicators of social capital: status, average number of ties sent and received, and the strength of the relationship. In this approach, models and measures are duals to each other. Han and Breiger show that their approach not only

CSC: An Agenda for the Future - 485

makes it possible to locate the sources of social capital in social structures, but that dimensions of this structure can be separated out as well, enabling the researcher to study their individual effects.

Doreian (this volume) argues that organizations that are well regarded by other organizations have higher social capital. Moreover, organizations that are well regarded by other well regarded organizations have even higher social capital. This argument allows Doreian to use input-output methods for generating measures of firm-level social capital. His approach allows one to assess from which part of its network a firm derives its social capital.

The studies by Han and Breiger and Doreian provide CSC theorists with ways of measuring aspects of social capital. But there are still many other measurement issues to be addressed in the near future. For instance, social structure can provide a firm with the social capital of increased learning ability, timely access to strategic information, access to physical resources, referrals from third parties, or the opportunity to obtain loans at a lower cost-different measures may be needed for different outcomes. Also, social relationships are often multiplex, leading to overlap of structures of advantage and disadvantage, which should be taken into account in future measures of social capital. In addition, measures are required that take into account that the same social structure can be beneficial for the attainment of one goal, but detrimental to the achievement of another goal. Further, social structure at the firm level and/or the individual level affects the social capital at both levels. Measuring this interplay of levels of analysis is perhaps the most difficult challenge.

The challenge of 'Contingency' One of the distinctive characteristics of social capital is its contingent nature. The same social structure provides some actors with social capital, while bringing social liability to others. The search for contingency factors can be guided by the following question: Which social structure is benejiciaVdetrimental for whom. for which goals. where. and when? Although the elements in this question are interrelated, we will briefly address each of them in turn, without attempting to fully separate them.

Which social structure The social network literature is replete with studies that discuss the structural characteristics that create advantages and disadvantages for corporate players. In the social networks literature, a number of structural characteristics have been associated with success and failure.

Undoubtedly the most commonly studied structural feature in the literature is the number of ties that reach or reach out from an actor. A large volume of ties has been argued to provide information and control benefits. It is also seen as an indicator of power and status. Although this view is certainly useful in a number of cases, it is built on the implicit assumption that ties and social capital can be equated. As argued in various places in this volume, that assumption often can not be maintained. However, it is important to study in which situations the volume of ties does have an effect on social capital, and when not (or to a lesser extent).

Initiated by the work of White, Boorman, and Breiger (1976) and Burt (1992) is the idea that the study of absence of ties provides vital information about networks


that goes beyond the classical study of ties present. In particular, Burt's concept of structural holes has inspired social capital researchers in recent years. Here also further research is necessary. For instance, do structural holes always provide social capital? Burt has provided many examples of situations in which structural holes can be associated with success in a competitive arena. However, Gabbay (1995, 1997) found that, although bridging actors may have a strategic edge, this does not mean that they necessarily do any 'better.' Gabbay's findings suggest that only when actors actively and consciously create structural holes-by adding ties and shaping their network-do they take advantage of structural holes. In the context of job mobility of R&D unit managers in Fortune 500 firms, Gabbay and Zuckerman (1998) found that, depending on contextual factors-the degree to which the unit is oriented toward individual rather than to collective incentives, the density of the overall pattern of interaction in the unit, and the extent to which brokerage strategies are legitimate-there is a varying effect of structural holes on managerial mobility.

Other structural characteristics have been studied, such as density, segregation, and level of multiplexity-and many potential others come to mind. At present, we do not yet have a full picture of which features of social networks really do provide benefits or barriers. It is likely that we should broaden our focus beyond the purely structural quality of networks and include characteristics of the nodes as well. For example, for a medical representative, there is a clear difference in opportunity provided by ties to general practitioners as compared to family ties. As another example, politicians surely need strong supportive ties with their grassroots support, but can not achieve their goals without productive ties with those in power.4

In addressing 'which social structure,' we have to search for the differential characteristics of social structure-including the attributes of the actors comprising the social structure-that affect the attainment of goals of corporate players. The findings reported above open the search for alternative characteristics.

For whom In CSC, an explicit distinction is made between structure and capital/liability at the level of the firm and at the level of the individual member of the firm. Particular structures may be beneficial for players at one level, but detrimental to players at another. Therefore, in addressing issues such as 'which structure' and 'for which goals,' one should be explicit in addressing 'for whom.'

Most research studying the effects of social structure on effective organization does not address the issue of these interconnected levels of analysis. Rather, most studies focus on only one level at a time. Perhaps this is all we can expect at this point. We should be able to walk before we run. But eventually we will need theories that explicitly address the interplay between these levels (and even refine these levels). Theories that do so may prove instrumental in furthering our understanding of many facets of organizations.

For which goals Corporate players have various goals. Firm goals can include high sales, low price of raw materials, timely access to market information, increasing its knowledge base, the identification of potential partners or employees, creating influence on

csc: An Agenda for the Future - 487

governmental decision-making, increasing productivity, identifying market opportUnIties, and so forth. Most of these goals require different types of relationships, both within the firm and between firms.

Goals of firm members can include good career opportunities, job satisfaction, high individual sales, opportunities to travel, high income, job security, status, and so forth. Again, most of these goals require different types of relations, both within the firm and with employees of other firms.

An example of this is provided in a study of R&D team performance (Kratzer, van Engelen, and Leenders, 1998). R&D teams are confronted with the need to resolve various issues. Kratzer et al. argue that, in the ideation phase of the R&D process, a segmented network structure is advantageous to the team's ability to solve administrative and coordination problems efficiently. However, segmentation negatively affects the team's problem-solving capacity. The same structural characteristic thus has a different effect on the accomplishment of two different goals.

Additionally, corporate players often attempt to advance multiple goals simultaneously. For the attainment of multiple goals, still other structures may be necessary.

One of the most fundamental questions in this area is the condition under which the goals of firm and their members coincide. Since the goals of the corporation and the goals of its members are often different, the pursuit of the attainment of individual goals by firm members may not further firm level goals, it may even impede their achievement.s Just as firm level social structure is not simply an aggregate of individual level social structure, and firm level social capital is not simply an aggregate of individual level social capital, so too are firm level goals not simply an aggregate of individual level goals. For example, Flap and Boxman (this volume: 215) bring to bear the issue of 'managers who act opportunistically and misuse corporate social structure to advance their own goals to the detriment of the goals of the company they work for'. We thus need to search for conditions under which individual level goals and firm level goals concur or diverge--rnore to the point, we need to search for conditions under which social structure at the firm and individual levels lead to the advancement of overlapping goals, or to the obtrsuction of goals at one of these levels.

This contingency simply indicates that research on corporate social capital should be explicit in the kind of goals sought after by the corporate players. One cannot expect a single theory of social capital to expiain structural effects on the furthering of every goal conceivable. We therefore need theories that are explicit in the goals addressed, and, preferably, can hint at their value for the attainment of alternative goals.6

Where The fourth contingency factor addressed here is 'where.' Most of the popular organization research written in the English language concerns corporate players in Western societies. However, cultural differences make it difficult to generalize findings. In engaging in business interactions, Arabs rely strongly on the trust and tacit understandings inherent in their relationships with their exchange partners.


Israelis, on the other hand, rely more on formal contracts and procedures.7 In Taiwan, where inheritance is the main organizing factor, the lion's share of businesses have fewer than twenty employees. Japan is well known for its highly networked keiretsu; Korea for its vertically structured chaebol. Consistent with historically developed social structures and practices, businesses organize differently in different cultures.

But even within the same country social structures may operate differently. Foss (1998) found that the same structural characteristics of networks of fish farming entrepreneurs in Norway had different effects in different areas of the country. For a large part of this century, networks in the eastern part of The Netherlands were relatively small, because the farming population relied on few, but strong, ties. The business population in the Western part of the country needed large networks, mainly to politicians and other businessmen.

It is conceivable that different organizational cultures make for different effects of the contingency factors. This is especially important because of the strong and propeling drive toward globalization. Ties between actors in different countries may carry a different baggage. For example, Harland (this volume: 416) notes that 'parties to UK-based relationships expected more distance than those in Spanish relationships studied where friendship between supplier and customer was expected.' With the presence of cultural differences with respect to the interpretation and expectation of social ties, one should look even deeper for the mechanisms that translate social structure into social capital and liability.

It will be most interesting to see the direction that will be taken with future theory that compares how various network structures affect the advancement of goals of corporate players in different organizational cultures.

When An anecdote tells about a consultant who did a good job in explaining to a group of managers the merit and importance of corporate culture. One of the managers present enthusiastically reacted with 'Sounds great, I would like that in my corporation, can I have it by Monday morning?'

This anecdote could have been about corporate social capital. Corporate social capital is not built overnight. In fact, it only exists by virtue of time. Trust needs time to grow. Relationships need time to mature. Payoff often follows the social structure after some time has passed. Benefits drawn are often not concentrated in one point in time, but accrue to a player over a longer time span.

There are two main reasons why a time dimension needs to be incorporated into the area of CSC research. First of all, social structure changes over time and so does its effects on corporate players. See, for instance, the chapter by Gargiulo and Benassi (this volume). Second, social structure both affects and is affected by the actions of corporate players. In order to determine causality, one needs a longitudinal approach. We will address the former issue here, and the latter in the section 'time and causality.'

As exchanges develop, become stronger and more long-standing, trust between the exchange partners is likely to increase. An employee who has been with a firm for thirty years may be less likely to search for a position in another firm than an


untenured colleague. Firms that have been involved in multiple collaborations in the past may be more likely to seek each other out for future collaboration. A shared history makes for increased levels of trust and creates higher levels of similarity in culture and narratives. Consequently, transaction costs are decreased. The downside to longstanding collaboration is that it may hamper the firm's ability to learn and its access to information on the market.

At different points in time, different structures come into play. In the initial stages of a firm-to-firm relationship, a third party may be needed in the role of hostage keeper, assuring both parties that vital information will not be leaked to outsiders. 8 At later stages, the presence of a third party may no longer be required. In their chapter in this volume, Higgins and Nohria discuss the effect of mentoring relationships on a protege's ability to create social capital over the course of his or her career. Their findings suggest that having a mentor early on in one's career can be obstructive to the protege's development of social capital. However, once the protege has developed a professional identity and reputation of his or her own, having a mentor is helpful in facilitating the formation of new ties. In accordance with our 'contingency question,9 above, this leads Higgins and Nohria to say that 'we need to consider not just knowing how these relationships work, but also who and when affect a protege's ability to progress in his or her career.'

An alternative example of the time contingency is given by Kratzer et al. (1998), who argue that different structures are necessary for R&D team performance at different stages in the innovation cycle. For example, controlling for other factors, they find that team leader centrality is beneficial for the team's performance at the development stage, but is negatively related to performance at the initial stages of the R&D process.

The issue of when is an important contingency factor that needs to be addressed in theories of CSc. Only then can we hope to begin to fully understand the interplay between social structure and the attainment of goals of corporate players.

The challenge of 'Time and Causality' The third issue on the agenda of research in the field of corporate social capital addresses the need for dynamic specifications of social structure and social capital: is social structure a generator of, or generated by social capital, or both? Below, we briefly address these questions in turn.

Social capital as an outcome of social structure Theories of social capital usually treat social structure as a generator of social capital. Actors are viewed as enmeshed in a web of relationships. Social capital then refers to the resources inherent in a dyadic relationship or in a set of relationships. This premise is theoretically limiting because social structure is constantly changing and actors can alter social structure to suit their needs.1O Firms engage in collaboration; buyers and sellers enter into exchange with the focal firm. Relationships are created and dissolved. Individuals move into and out of organizations, or move between teams or departments and climb the corporate ladder. 11


When the Dutch brewery Heineken set out to expand its business abroad, Pieter Feith, member of the board of directors, accidentally ran into John Fraser and David Neave, two representatives of the Singapore and Straits Aerated Water Company. With their help, Heineken established Malayan Breweries Limited and built the Singapore Tiger Brewery. Tiger soon became one of the best selling beers in Singapore. The joint venture was a rich source of corporate social capital for Heineken and resulted in large sales all over Asia. Years later, Heineken set out to promote its own brand in Singapore. Unfortunately, Fraser and Neave had created strong ties to many crucial regional business contacts in Singapore and were more interested in promoting Tiger beer than in promoting Heineken. This social structure resulted in poor availability of Heineken beer in Singapore. These unwanted changes in Heineken's social structure left the market open for European competitors (particularly for Carlsberg which soon sold 17 times as much beer in Singapore as did Heineken under its own brand name).12

If social structure is considered to convey social capital and social liability, the kind and amount of social capital and liability evolves along with the social structure. A static view of social capital merely presents a snapshot, a single frame from a movie. For a full picture of the history and evolution of corporate social capital, a dynamic point of view is required.

Social capital as a source of social structure Success is attractive. Successful accountants, lawyers, consultants, and sales representatives carry a large collection of social capital with them. When firms are able to hire them, their fruitful relationships also (to some extent) become part of the firm's network; in other words, the social ties of the individual now yield social capital at the firm level.

Companies central in the social structure have timely access to resources held by others.13 This then draws other firms to collaborate with the central player, in order to gain potential access to these resources (and status). The firm, in turn, becomes even more central in the social structure.14

Whereas success is attractive, failure surely is not. Firms that are strongly embedded in rigid business relationships are prone to suffer from decreasing learning ability. Such firms are unlikely candidates for generating new ties and engaging in new ventures. The social liability inherent in the focal firm's network both prevents it from entering into new relationships, and may even cause its current partners to disinvest in the mutual tie.

Social structure evolves with social capitaVIiability. There are only few studies that deal with this causal direction. For the understanding of how social structure is related to performance of firms and its members, studies in this area are imperative.

Co-evolution of social capital and social structure If social structure is considered to convey social capital and social liability, the kind and amount of social capital and liability co-evolves with the social structure.

The social capital (liability) of a focal player may draw other players to initiate (dissolve) relationships. By creating an additional tie in the social structure, a player hopes to gain access to resources held by the focal firm that may otherwise be


unavailable. The tie thus conveys social capital. The change in network structure follows from purposeful actions of players in their quest for social capitaI-in order to attain their individual goals.

The character of relationships also changes over time. First, relationships require time to be built and go through various stages. IS The social capital and social liability inherent in the relationship permutes over these stages. Second, the multiplex character of most social relationships draws attention to how the contents of relationships change over time. Two partner firms may move from codevelopment of a new product to joint marketing of this product, while collaborating in the development of a second product. These changes in content, often difficult to observe for an outside researcher, make for various kinds and amounts of social capital derived from the collaboration. In order to capture the dynamics of corporate social capital, a longitudinal point of view is instrumental.

From a methodological point of view, there is an additional reason to collect and study longitudinal data on social structure and social capital. If social structure and social capital co-evolve, one researcher can always empirically claim that social capital is a source of social structure, whereas another researcher can empirically claim, with the exact same data, that social capital was the generator of social structure. 16 Co-evolution of social structure and social capital make causal statements, based on static data, irrelevant and untenable. Only with the help of a longitudinal research design can one disentanglel7 the interrelationship of structure and capital.

Members of the firm engage in relationships with outside constituencies that may, over time, translate into social capital at the firm level. These ties may, over time, become institutionalized at the firm level. Other individuals within the firm may then draw from these relationships and extract social capital at the individual level. Change over time of social structure and social capital cuts across the analytical levels we distinguish in this volume -the firm and its members. Grasping this longitudinal crossover is perhaps the biggest challenge in CSC theory development and research.

The challenge of 'Social Capital Management' Once we have developed a notion of the contingencies affecting how social structure and social capital are related, and have put into place the theory and methodology for dynamic representations, the next part of the agenda refers to the practical application of the approach. Slightly rephrasing the words of Kanter and Eccles (1992: 527): it is from an action perspective that managers derive their interest in corporate social capital. The kind of knowledge that can be made available through studying the interplay of network structure and social capital and liability at the firm and individual level needs to be supplemented with knowledge about how to build and use networks in order to create (maximum) social capital.

'Corporate social capital management' refers to the purposeful alteration of social structure to fit players' goals. The corresponding questions are: how can social structure be changed to suit the attainment of goals, and what strategies do actors use to invest, reinvest, or disinvest in their social ties?


Although Burt (1992) empirically follows the 'social capital as an outcome of social structure' approach, structural holes theory suggests that actors can change their networks to make them more efficient and effective. This is manifested in the theory's explicit prescription of investing in non-redundant ties and disinvesting in relationships that are redundant. Burt implicitly calls for strategically alternating social networks to maximize efficiency.

To reiterate, management of corporate social capital can be discussed at the different levels of analysis central in this volume.

Structure at the firm level Over the past few decades, industry after industry has become globalized. The emerging global marketplace poses the challenge to firms of spreading their value activities worldwide, in pursuit of simultaneously realizing the benefits of low cost and differentiation. In this pursuit firms increasingly enter into strategic alliances, as it is often faster and more efficient to work through external alliances than to try to change the organizational mandates of subsidiaries. 18 This is but one example of the host of interorganizational ties into which firms enter. 19 Firms engage in interorganizational relationships in order to secure resources, gain access to new markets, lower costs, and stay abreast of market conditions.2O Careful selection is of utmost importance and will determine whether and how much social capital can be drawn from alliance relationships.21

Firms can also actively engage in inter-firm relationships with the goal of increasing the social capital of their employees. For instance, firms sometimes engage in international collaboration which, at least as a side-goal, gives the firm's employees the possibility to travel and be involved in international assignments. In the computer industry, in order to be able to provide their members with various kinds of employee benefits, many businesses nowadays engage in relationships with car manufacturers, insurance companies, mortgage companies, and so forth. Potential emplo~ees then partly base their choice of employers on the benefits offered to them. 2

Structure at the individual level Individual members of firms can also engineer the social structure so as to increase their social capital. For example, creating ties with mentors can be beneficial to one's career. Creation and maintainance of social ties may be instrumental to acquiring information about job opportunities.23 Gabbay (1995, 1997) showed that the individuals who are most successful in network marketing are exactly those who initially were at an extreme social structural disadvantage and engineered their social structure to suit their goals.

Individuals may also alter their social ties so as to further firm-level goals. For example, studying dependence relations, political alliances, and confidential discussion networks among decision makers in a cooperative Agri-business, Gargiulo (1993) shows that leaders build ties of interpersonal obligation with people directly affecting their performance in the organization. When policies' divergence or personal friction makes these ties untenable, leaders build strong cooperative


relationships with people who may constrain the performance of the party on whom they depend.24

Baker and Obstfeld (this volume) devote their chapter to the description of two strategies an entrepreneur can employ to access and mobilize social capital from social structure. In the disunion strategy, the social entrepreneur bridges between two disconnected alters. In the union strategy, on the other hand, ego closes the gap between two alters who are disconnected or in conflict. Baker and Obstfeld are convincing in their discussion of the conditions under which these strategies are beneficial to ego-their chapter is valuable for the study of social capital management. Baker and Obstfeld's chapter focuses on strategies ego can employ with respect to two alters, but we also need to look for the opportunities ego can create in larger systems. By both studying the opportunities ego can create and maintain in a subset of a network, and those within larger systems, we can start to understand how corporate social capital can be managed, both for firms and individuals.

The agenda on corporate social capital management takes the academic work on CSC to a practical level. The agenda embodies the search for how social capital can be created and maintained, and which factors playa role in doing so.

To the fundamental question we posed earlier: What social structure is beneficiaVobstructive for whom, for what goals, where and when?-we now add: 'how do we get and stay there?'

CONCLUSION

The 45 scholars who contributed to this volume have presented the reader with a thorough overview of the area of Corporate Social Capital. The wave of recent papers in various literatures employing a social capital framework highlights the accumulation of critical mass for the study and implications of social capital in general and corporate social capital in particular. CSC is only at its infancy stage. It will take additional research to bring it to adolescence.

The prominence of the authors of the chapters in this volume speaks for the merit CSC has in the study of organizations. We hope that this volume has inspired the reader, and that the study of Corporate Social Capital will prove fruitful in our understanding of organizations.

NOTES

I. The number of ties is of course also a measure of centrality. We mention them here separately because the number of ties is often used in research without reference to centrality. 2. Burt (1992). 3. At first sight, this may seem helpful, but still presents a lot of difficulty. For instance, besides the social capital the individual brings to the firm, reward systems often also depend on other characteristics. Some of these are idiosyncratic to the employee; many of them are determined by organizational culture and history, and the lack of flexibility in readjusting reward systems to new conditions. For instance, see Van Veen (1997). 4. Leenders (1991). 5. O'Aveni(199I). 6. A related issue, beyond the scope of this chapter, is the issue of how the attainment of these goals can be measured. This issue in itself is the focus of a large body of research.


7. Gabbay, Semyonov, and Mishal (1999). S. See Nooteboom's chapter in this volume. 9. What social structure is beneficiallobstructive for whom, for what goals, where, and when? 10. Amomg others, Gabbay and Sato (1996), Gabbay (1995, 1997), Leenders (1995ab, 1996). 11. Han (1996) shows that most contacts between firm members occur within divisions, rather than between them. Contacts across divisions are almost exclusively concentrated in the higher echelons. Individuals with higher formal positions tend to have denser personal networks. In addition, Han shows that hierarchical position affects the extent to which an individual initiates interactions, rather than receiving them. These findings indicate that an individual's social network changes as she advances in her career. In the 'social capital as an outcome of social structure' approach, Han's findings suggest that an individual's social capital changes with the individual's professional advancement. 12. Smit (1996). 13. Cf. Stuart (this volume). 14. Also see Freeman (this volume). 15. See Zajac and Olson (1993) and Omta and Van Rossum (this volume). 16. Leenders (1997). 17. Leenders (1995a, 1995b) calls this the 'untying of the knot: He extensively discusses models and approaches for the longitudinal study of networks. IS. Yoshino and Rangan (1995). 19. Also, see Knoke (this volume). 20. In the context of states as actors, Gabbay and Stein (1999) argue for the creation of Regional Social Capital in the context of creating technological interdependencies which will foster social relationships in the context of peace in the Middle East. They show how it is in the interest of the third party-the USAto foster regional social capital by forming technological infrastructurai dependencies. 21. Also seeOmta and Van Rossum (this volume) on this topic. 22. One of the major Dutch automation companies recently held its job interviews in showrooms of car dealers. Job candidates, who were informed of the outcome immediately after the interview, could pick out a car to their liking on the spot and drive home in it. In this fashion, the company tried to lure able new employees to fill its vacancies. 23. See, for instance, the chapter by Flap and Boxman. 24. Gargiulo (1993:1).

References

• Abel, R. L. 1989 American Lawyers. New York: Oxford University Press. Abernathy, W. 1978 The Productivity Dilemma. Baltimore: John Hopkins University Press. Abolafia, M. 1996 Making Markets: Opportunism and Restraint on Wall Street. Cambridge: Harvard University

Press. Abrahamson, E. 1991 'Managerial Fads and Fashions.' Academy of Management Review 163: 586-612. Acevado, M., E. M. Basualdo, and M. Khavisse 1990 Quien as Quien?: Los Duenos del Poder Eeconomico (Argentina 1973-1987). Buenos Aires:

Editora. Adams,J.S. 1976 The Structure and Dynamics of Behavior in Organizational Boundary Roles.' In M. D. Dunette

(ed.), Handbook of Industrial and Organizational Psychology, 1175-1199. Chicago: Rand McNally.

Adler, P. S. 1993 'The 'Learning Bureaucracy': New United Motor Manufacturing, Inc.' In L. L. Cummings and B.

M. Staw, Research in Organizational Behavior 15: 111-194. Agresti,A. 1990 Categorical Data Analysis. New York: Wiley. Ahuja,G. 1998 'Collaboration Networks, Structural Holes, and Innovation: A Longitudinal Study.' Best Paper

Proceedings, National Meetings of the Academy of Management, San Diego. Albert, M. B., D. Avery, F. Narin, and P. McAUister 1991 'Direct Validation of Citation Counts as Indicators of Industrially Important Patents.' Research

Policy 20: 251-259. Aldrich, H. E. 1979 Organizations and Environments. Englewood Cliffs: Prentice-Hall. Aldrich, H. E., and C. Zimmer 1986 'Entrepreneurship Through Social Networks.' In D. L. Sexton and R. W. Smilor (eds.), The Art and

Science of Entrepreneurship, 3-23. Cambridge: Ballinger. Alford,R. 1975 Health Care Politics. London: University of Chicago Press. Allison, G. T. 1971 Essence of Decision. Boston: Little Brown. Allport, F. H. 1933 Institutional behavior. Chapel Hill: University of North Carolina Press. Amason, A. C., and D. M. Schweiger 1994 'Resolving the Paradox of Conflict, Strategic Decision Making, and Organizational Performance.'

International Journal of Conflict Management 5: 239-253. Ammer,D.S. 1968 Materials Management. Homewood: Irwin.


Amsden, P. H. 1989 Asia's Next Giant: South Korea and Late Industrialization. New York: Oxford University Press. Anderson, P. 1995 'The Microcomputer lndustry.' In G. R. Carroll and M. T. Hannan (eds.), Organizations in

Industry: Strategy. Structure. and Selection. 37-58. New York: Oxford University Press. Andrews, K. R 1971 The Concept of Corporate Strategy. Homewood: Irwin. Anheier, H. K., J. Gerhards, and F. P. Romo 1995 'Forms of Capital and Social Structure in Cultural Fields: Examining Bourdieu's Social

Topography.' American Journal of Sociology 100: 859-903. Ansoff, H. I. 1965 Corporate Strategy. New York: McGraw-Hill. Aold M., B. Gustafsson and O. E. WilUamson (eds.) 1990 The Firm as a Nexus of Treaties. UK: Sage. Arnold, B. L., and F. M. Kay 1995 'Social Capital, Violations of Trust and the Vulnerability of Isolates: The Social Organization of

Law Practice and Professional Self-Regulation. ' International Journal of the Sociology of Law 23: 321-46.

Ashkenas, R, D. Ulrich, T. Jick, and S. Kerr 1995 The Boundaryless Organization: Breaking the Chains of Organizational Structure. San Francisco:

Jossey-Bass. Axelrod, R M. 1980 'Effective Choice in the Prisoner's Dilemma.' Journal of Conflict Resolution 24: 3-25. Axelsson, B., and G. Easton (eds.) 1992 Industrial Networks: A New View of Reality. London: Routledge. Baba, v. 1990 'Methodological Issues in Modeling Absence: A Comparison of Least Squares and Tobit

Analyses.' Journal of Applied Psychology 75: 428-432. Bailey, W..J., R Masson and R Raeside 1996 'Choosing Successful Technology Development Partners: A Best Practice Model.' Proceedings of

the 6th International Forum on Technology Management: 271-83. Baker, G., M. Gibbs, and B. Holmstrom 1994 The lnternal Economics of the Firm: Evidence from Personnel Data.' The Quarterly Journal of

Economics 109: 881- 919. Baker, W.E. 1984 The Social Structure of a National Securities Market.' American Journal of Sociology 89: 775-

811. 1990 'Market Networks and Corporate Behavior.' American Journal of Sociology 96: 589-625. 1992a The Network Organization in Theory and Practice.' In N. Nohria and R. G. Eccles (eds.),

Networks and Organizations: Structure, Form. and Action, 397-429. Boston: Harvard Business School Press.

1992b 'Competition, Cooperation, and Collusion: Three Faces of Economic Behavior.' Paper presented at the joint annual meetings of the Public Choice Society and Economic Science Association, New Orleans.

1994a Networking Smart: How To Build Relationships for Personal and Organizational Success. New York: McGraw-Hili.

1994b The Paradox of Empowerment' Chief Executive 93: 62-65. Baker, W. E., and R R Faulkner 1991 'Strategies for Managing suppliers of Professional Services.' California Management Review 33:

33-45. 1993 The Social-Organization of Conspiracy-megal Networks in the Heavy Electrical-Equipment

lndustry.' American Sociological Review 58: 837-860. Baker, W. E ., R R Faulkner, and G. A. F'lSher 1998 'Hazards of the Market: The Continuity and Dissolution of lnterorganizational Market

Relationships.' American Sociological Review 63: 147-177. Bandura,A. 1977 'Self-Efficacy: Towards a Unifying Theory of Behavioral Change.' Psychology Review 84: 191-

215.

References - 497

Bantel, K. A., and S. E. Jackson 1989 'Top Management and Innovations in Banking: Does the Composition of the Top Team make a

Difference?' Academy of Management Journal 10: 107-124. Barley, S., J. Freeman, and R. Hybels 1992 'Strategic Alliances in Commercial Biotechnology.' In N. Nohria, and R. G. Eccles (eds.),

Networks and Organiwtions: Structure, Form, and Action, 311-347. Boston: Harvard Business School Press,

Barnett, W. P. 1997 'The Dynamics of Competitive Intensity.' Administrative Science Quarterly 42(1): 128-160. Barney, J. B. 1991 'FllTII Resources and Sustained Competitive Advantage.' Journal of Management 17(1): 99-120 Barney, J. B., and M. H. Hansen 1994 'Trustworthiness As a Source of Competitive Advantage.' Strategic Management Joumall5: 175-

190. Baron,J. N. 1984 'Organizational Perspectives on Stratification. ' Annual Review of Sociology 10: 37-69. Baron, J. N., A. Davis-Blake, and W. T. Bielby 1986 'The Structure of Opportunity: How Promotion Ladders Vary within and Among Organizations.'

Administrative Science Quarterly 31: 248-273. Baron, J. N., F. Dobbin, and P. D. Jennings 1986 'War and Peace: The Evolution of Modem Personnel Administration in U.S. Industry.' American

Journal of Sociology 92: 350-383. Baron, J. N., P. D. Jennings, and F. Dobbin 1988 'Mission Control? The Development of Personnel Systems in U.S. Industry.' American

Sociological Review 53: 497-514. Barron, D. N. 1992 'The Analysis of Count Data: Overdispersion and Autocorrelation.' Sociological Methodology 22:

179-220. Barry,C.B. 1994 'New Directions in Research on Venture Capital Finance. ' Financial Management 23: 3-15. Bartlett, C. A., and S. Gboshal 1989 Managing Across Borders: The Transnational Solution. Boston: Harvard Business School Press. Bates, R.H. 1988 'Contra Contractarianism-Some Reflections on the New Institutionalism.' Politics and Society 16:

387-401. Batson, C. D. 1990 'How Social an Animal: The Human Capacity for Caring.' American Psychologist 45: 336-346. Bauer, M., and B. Bertin-Mourot 1991 Les 200' en France et en Allemagne. Deux modeles de ditection-selectionjonnation de dirigeants de

grands entreprises. Paris: CNRS and Hedrick and Struggles. Bawn, J. A. C., and C. Oliver 1991 'Institutional linkages and organizational mortality.' Administrative Science Quarterly 36: 187-

218. 1992 'Institutional Embeddedness and the Dynamics of Organizational Populations'. American

Sociological Review 57: 540-559. Baumol, W.J. 1967 Business Behavior: Value and Growth. New York: MacMillan. Bearman,P. 1997 'Generalized Exchange.' American Journal of Sociology 102: 1383-1415. Becker, G. S. 1964 Human Capital. New York: Columbia University Press. Belliveau, M. A., C. A. O'Reilly, m, and J. B. Wade 1996 'Social Capital at the Top: Effects of Social Similarity and Status on CEO Compensation.' Academy of

Management Journal 6: 1568-1593. Belons, R. S. 1989 The Contingent Economy: The Growth of the Temporary, Part-Time, and Subcontracted

Workforce. Washington: National Planning Association.


Benassi,M. 1993 Dalla Gerarchia alia Rete: Modelli ed Esperienze Organizzative. Milano: Stas. Ben-Porath, Y. 1980 'The F-Connection: Families, Friends, and Fmns and the Organization of Exchange.' Population

and Development Review 6: 1-30. Berger, J., N. G. Noorderhaven and B. Nooteboom 1995 'Detenninants of Supplier Dependence.' In J. Groenewegen, C. Pitelis, and S. E. Sjostrand (eds.), On

Economic Institutions. Aldershot: Edward Elgar. Berger, P., and Th. Luckman 1967 The Social Construction of Reality. New York: Doubleday. Berg1ov,E. 1994 'Ownership of Equity and C01J>Orate Governance-The Case of Sweden.' In T. Baums, R. M.

Buxbaum, and K. J. Hopt (eds.), Institutional Investors and Corporate Gove17Ulllce. Berlin: Walter de Gruyter.

Berk,R.A. 1983 'An Introduction to Sample Selection Bias in Sociological Data.' American Sociological Review

48: 386-398. Berry, L. L., and A. Parasuraman 1991 Marketing Services: Competing Through Quality. New York: Free Press. Berscheid, E., and E. H. Walster 1969 Interpersonal Attraction. Reading: Addison-Wesley. Bhagwati, J. 1982 'Directly, Unproductive Profit Seeking (DUP) Activities.' Journal of Political Economy 90: 988-

1002. Bian, Y. 1997 'Getting a Job Through a Network of Guanxi in China' In B. Wellman (ed.), Networks in the

Global Village . Boulder: Westview. Bianchi,M. 1994 'Hayek's Spontaneous Order: The 'Correct' versus the 'Corrigible' Society.' In 1. Bimer and R. Van

Zijp (eds.), Hayek, Coordination and Evolution, 232-251. London: Routledge. Biemans, W. G. 1992 Managing Innovation within Networks. London: Routledge. Bills,D.B. 1988 'Educational Credentials and Promotions: Does Schooling Do More than Get You in the Door?'

Sociology of Education 61: 52-60. Bioscan 1988-1996 Phoenix: Oryx Press. Birley,S. 1985 'The Role of Networks in the Entrepreneurial Process.' Journal of Business Venturing 1: 107- 117. Bishop, Y. M. M., S. E. Fienberg, and P. W. Holland 1975 Discrete Multivariate Analysis: Theory and Practice. Cambridge: Harvard University Press Bitner, M. J., J. D. Nyquist, and B. H. Booms 1985 'The Critical Incident as a Technique for Analysing the Service Encounter.' In T. M. Block, G. D.

Upah, and V. A. Zeitharnl (eds.), Services Marketing in a Changing Environment, American Marketing Association.

Bitner, M. J., B. M. Booms, and M. S. Tetreault 1990 'The Service Encounter: Diagnosing Favorable and Unfavorable Incidents.' Journal of Marketing

54: 71-84. Blackwell, N., and A. Santomero 1982 'Bank credit rationing and the customer relation.' Journal of Monetary Economics 9: 121-129. Blau, J. R., and R. D. Alba 1982 'Empowering Nets of Participation.' Administrative Science Quarterly 27: 363-379. Blau,P.M. 1955 The Dynamics of Bureaucracy. Chicago: University of Chicago Press. 1964 Exchange and Power in the Social Life. New York: Wiley. 1977 Inequality and Heterogeneity: A Primitive Theory of Social Structure. New York: Free Press.

References - 499

Bleakley, F. R. 1995 'Strange Bedfellows: Some Companies Let Suppliers Work on Site and Even Place Orders.' Wall

Street Journal (January 13). New York: Dow Jones & Company. Inc. Bleeke, J., and D. Ernst 1993 Collaborate to Compete: Using Strategic Alliances and Acquisitions in the Global Marketplace.

New York: Wiley. 1995 'Is Your Strategic Alliance Really a Sale?' Harvard Business Review 73: 97-105. Blois, K.J. 1972 'Vertical Quasi-Integration.' Journal of Industrial Economics 20: 253-272. Blossfelcl, H., and G. Rohwer 1995 Techniques of Event History Modeling. Mahwah: Lawrence Erlbaum Associates. Bonacich, E. 1973 'A Theory of Middleman Minorities.' American Sociological Review 38: 583-594. Bonacich, P. 1972 'Technique for Analyzing Overlapping Memberships.' In H. Costner (ed.), Sociological

Metlwdology 1972,176-185. San Francisco: Jossey-Bass. 1987 'Power and Centrality: A Family of Measures.' American Journal of Sociology 92: 1170-1182. Borgatti, S. P. 1997 'Structural Holes: Unpacking Burt's Redundancy Measures.' Connections 20: 35-38. Borgatti, S. P., M. G. Everett, and L. C. Freeman 1992 UCINET N Version 1.00. Columbia: Analytic Technologies. Boulding W., A. Kalra, R. Staelin, and V. A. Zeithaml 1993 'A Dynamic Process Model of Service Quality: From Expectations to Behavioral Intentions.'

Journal of Marketing Research 30: 7-27. Bourdieu, P. 1972 Esquisse d'une Theorie de la Pratique. Precedee de Trois Etudes d'Ethnologie Kabyle. Geneve:

Droz. 1980 'Le Capital Social: Notes Provisoires.' Actes de la Recherche en Sciences Sociales 3: 2-3. 1986 'The Forms of Capital.' In I. O. Richardson (ed.), Handbook of Theory and Research for the

Sociology of Education, 241-258. New York: Greenwood Press. 1994 Raisons Practiques. Sur la Theorie de [,Action. Paris: Editions du Seuil. Bourweu, P., and M. de Saint Martin 1982 'Le patronat.' Actes de la Recherche en Sciences Sociales 32: 3-82. Bourdieu, P., and L. J. D. Wacquant 1992 An Invitation to Reflexive Sociology. Chicago: University of Chicago Press. Boxman, E. A.W., P. M. De Graaf, and H. D. Flap 1991 'The hnpact of Social and Human Capital on the Income Attainment of Dutch Managers.' Social

Networks 13: 51-73. Bradach, J. L., and R. G. Eccles 1989 'Price, Authority, and Trust: From Ideal Types to Plural Forms.' Annual Review of Sociology 15:

97-118. Bradburn, N. M. 1969 The Structure of Psychological Well-Being. Chicago: Aldine. Brass, D.J. 1981 'Structural Relationships, Job Characteristics, and Worker Satisfaction and Performance.'

Administrative Science Quarterly 26: 331-348. 1984 'Being in the Right Place: A Structural Analysis of Individual Influence in an Organization.'

Administrative Science Quarterly 29: 518-539. 1985a 'Men's and Women's Networks: A Study of Interaction Patterns and Influence in an Organization.'

Academy of Management Journal 28: 327-343. 1985b 'Technology and the Structuring of Jobs: Employee Satisfaction, Performance, and Influence.'

Organiwtional Behavior and Human Decision Processes 35: 216-240. 1992 'Power in Organizations: A Social Network Perspective.' Research in Politics and Society 4: 295-

323. 1995a 'Creativity: It's all in Your social network.' In C. M. Ford and D. A. Gioia (eds.), Creativity Action

in Organiwtions, 94-99. London: Sage. 1995b 'A Social Network Perspective on Human Resources Management.' In O. Ferris (ed.), Research in

Personnel and Human Resources Management, 39-79. Greenwich: JAI Press.


Brass, D. J., and M. E. Burkhardt 1992 'Centrality and Power in Organizations.' In N. Nohria and R. G. Eccles (eds.J, Networks And

Organiwtions: Structure, Form, and Action. Boston: Harvard Business School Press. 1993 'Potential Power and Power Use: An Investigation of Structure and Behavior.' Academy of

Management Journal 36: 441-470. Brass, D. J., K. D. Butterfield, and B. C. Skaggs 1998 'Relationships And Unethical Behavior: A Social Network: Perspective.' Academy of Management

Review 23: 14-31. Breiger, R. L. 1990 'Social Control and Social Network:s: A Model from Georg Simmel.' In C. Calhoun, M. W. Meyer,

and W. R. Scott (eds.), Structures of Power and Constraint: Papers in Honor of Peter M. Blau, 453-476. New York:: Cambridge University Press.

1995 'Social Structure and the Phenomenology of Attainment.' Annual Review of Sociology 21: 115-136.

Breiger, R. L., and J. G. Ennis 1997 'Generalized Exchange in Social Network:s: Statistics and Structure.' VAnnie sociologique 47: 73-

88. Breiger, R. L., and J. M. Roberts, Jr. 1998 'Solidarity and Social Network:s.' In P. Doreian and T. 1. Fararo (eds.), The Problem of Solidarity:

Theories and Models. New York:: Gordon and Breach. Breton, A., and R. Wintrobe 1982 The Logic of Bureaucratic Conduct. An Economic Analysis of Competition, Exchange, and

Efficiency in Private and Public Organiwtions. New York:: Cambridge University Press. Bridges, W. P., and W. J. Villemez 1991 'Employment Relations and the Labor Mark:et: Integrating Institutional and Mark:et Perspectives.'

American Sociological Review 56: 748-764. Brockhaus, R. H. 1982 The Psychology of the Entrepreneur.' In C. A. Kent, D. L. Sexton, and K. H. Vesper (eds.),

Encyclopedia of Entrepreneurship, 39-56. Englewood Cliffs: Prentice·HaIl. Brockhaus, R. H., and P. S. Hurwitz 1986 'The Psychology of the Entrepreneur.' In D. L. Sexton and R. W. Smilor (eds.), The Art and Science

of Entrepreneurship. Cambridge: Ballinger. Brogowicz A. A., L. M. Delene, and D. M. Lyth 1990 'A Synthesised Service Model with Managerial hnplications.' International Journal of Service

Industry Management 1: 27-46. Brown, J. S., and J. P. Duguid 1991 'Organizationall.eaming and Communities of Practice. ' Organiwtion Science 2. (reprinted in M. D.

Cohen & L. S. Sproull (eds.) 1996, Organiwtional learning, 58-82. London: Sage.) 1997 Organizing knowledge. Paper presented at the First Berkeley Forum, UC Berk:eley. Brown, S. W., and T. A. Swartz 1989 'A Gap Analysis of Professional Service Quality.' Journal of Marketing 53: 92-98. Bruce M., F. Leverick, D. Littler, and D. Wilson 1995 'Success Factors for Collaborative Product Development: A Study of Information and

Communication Technology. ' R&D Management 25: 33-45. Briiderl, J., and P. Preisendiirfer 1997 'Network: support and the success of newly founded business.' Journal of Small Business Economics

9: 1-13. Bruno, A. V., and A. C. Cooper 1982 'Patterns of Development and Acquisitions for Silicon Valley Start-ups.' Technovation 1: 15-37. Brusco,S. 1982 The Emilian Model: Productive Decentralization and Social Integration.' Cambridge Journal of

Economics 6: 167-184. Bueno de Mesquita, B., and F. N. Stokman 1994 European Community Decision Making . New Haven: Yale University Press. Bulder, B., F. Leeuw, and H. F1ap 1996 'Networks and Evaluating Public-Sector Reforms.' Evaluation 2: 261-276. Burbidge, J. L. 1961 The New Approach to Production.' Production Engineer 40: 769-784.

References - 501

Burchell, B., and F. Wilkinson 1997 'Trust, Business Relationships and the Contractual Environment.' Cambridge Journal of

Economics 21: 217-237. Burgelman, R. A. 1983a 'Corporate Entrepreneurship and Strategic Management: Insights from a Process Study.'

Management Science 1349-1364. 1983b 'A Process Model of Internal Corporate Venturing in the Diversified Major Firm.' Administrative

Science Quanerly 28: 223-244. 1991 'lntraorganizational Ecology of Strategy Making and Organizational Adaptation: Theory and Field

Research.' Organization Science 2: 239-262. Burkhardt, M. E., and D. J. Brass 1990 'Changing Patterns or Patterns of Change: The Effect of a Change in Technology on Social

Network Structure and Power.' Administrative Science Quanerly 35: 104-127. Burt, R. S. 1982 Toward a Structural Theory of Action. New York: Academic Press. 1992 Structural Holes: The Social Structure of Competition. Cambridge: Harvard University Press. 1996 'Entrepreneurs, Trust and Third Parties: Paper presented at the conference Shared Cognition in

Organizations, Northwestern University, November. 1997 'The Contingent Value of Social Capital.' Administrative Science Quanerly 42: 339-365. Burt, R. S., S. M. Gabbay, G. Holt, and P. Moran 1994 'Contingent Organization as a Network Theory: The Culture-Performance Contingency Function:

Acta Sociologica 37: 345-370. Burt, R. S., and M. Knez 1995 'Kinds of Third-Party Effects on Trust.' Rationality and Society 7: 255-292. Burt, R. S., and D. Ronchi 1990 'Contested Control in a Large Manufacturing Plant.' In J. Weesie and H. D. Flap (eds.), Social

Networks through Time, 121-157. Utrecht: Isor. Burt, R. S., and I. Talmud 1993 'Market Niche'. Social Networks 15: 133-149. Business Week 1986 'And Now, The Post-Industrial Corporation: March 3: 64-71. Bygrave, W.D. 1988 'The Structure of Investment Networks of Venture Capital Firms.' Journal of Business Venturing

3: 137-157. Bygrave, W. D., and J. A. Timmons 1992 Venture Capital at the Crossroads. Cambridge: Harvard Business School Press. Cabo, P. G., W. van Rossum, V. Mangematin, D. Vinck, and P. Saviotti 1996 Academidlndustrial Research Collaboration in the Biotechnology Programme of the European

Community (1992-1994 J. Report for the European Community DO XII. Cadotte E. R., R. B. Woodruff, and R. L. Jenkins 1987 'Expectations and Norms in Models of Consumer Satisfaction: Journal of Marketing Research 24:

305-314. CampbellA. 1996 'The Agile Enterprise: Assessing the Technology Management Issues.' Proceedings of the 6th

Internationa"r Forum on Technology Management. 197-211. Campbell, K. E., P. V. Marsden, and J. S. Hurlbert 1986 'Social Resources and Socioeconomic Status.' Social Networks 8: 97-117. Cannon, W. B. 1932 The Wisdom of the Body. New York: Norton. Cappelli, P., and W. F. Cascio 1991 'Why Some Jobs Command Wage Premiums: A Test of Career Tournament and Internal Labor

Market Hypotheses.' Academy of Management Journal 34: 848-868. Cappelli, P., L. Bassi, H. Katz, D. Knoke, P. Osterman, and M. Useem 1997 Change at Work. New York: Oxford University Press. Carlisle J. A., and Parker R. C. 1989 Beyond Negotiation: Customer Supplier Relationships. UK: Wiley. Carroll, G. R., and A. C. Teo 1996 'On the Social Networks of Managers.' Academy of Management Journal 39: 421-440.


Caruso, R. E. 1992 Mentoring and the Business Environment: Asset or Liability? Vermont: Dartmouth. CASSIS 1987-1996 U.S. Patent and Trademark Office. Washington: U.S. Department of Commerce. CasteUs,M. 1996 The Rise of the Network Society. Oxford: Basil Blackwell. Chamberlain, N. W. 1968 Enterprise and Environment. The Firm in Time and Place. New York: McGraw-HilI. Chandler, A. 1977 The Visible Hand. Cambridge: Harvard University Press. Chatterji, D. 1996 'Accessing External Sources of Technology.' Research Technology Management 39: 48-56. Chesbrough, H. W., and D. J. Teece 1996 'When is Virtual Virtuous? Organizing for Innovation.' Harvard Business Review 74: 65-73. Chiesa, V., and R. Manzini 1996 'Profiting from the Virtual Organization of Technological Innovation: Suggestions from an

Empirical Study. ' Proceedings of the 6th International Forum on Technology Management, 211-27.

Cho, D. S., S. Nam, and L. Tung 1998 Korean Chaebols. New York: Oxford University Press. Christopher, M. G. 1992 Logistics and Supply Chain Management. London: Pitman Publishing. Chung,S. 1996 'Performance Effects of Cooperative Strategies Among Investment Banking Firms: A Loglinear

Analysis of Organizational Exchange Networks.' Social Networks 18: 121 -148. Chung, S., H. Singh, and K. Lee 1995 Comparative Capabilities and Reputational Capital as Drivers of Strategic Alliance Formation.

Nov. 1995. Snider Entrepreneurial Center, The Wharton School, University of Pennsylvania. Cicourel, A. 1973 Cognitive Sociology, Harmonsworth: Penguin Press. Clark.D. 1996 'Software Firm Fights to Remake Business After lll-Fated Merger.' Wall Street Journal, January

12: AI. Clark. K. B., and T. Fujimoto 1991 Product Development Performance: Strategy, Organization and Management in the World Auto

Industry. Boston: Harvard Business School Press. Clawson,J. 1980 'Mentoring in Managerial Careers.' In C. B. Derr (ed.), Work, Family and the Career, 144-165.

New York: Praeger. Clegg,S. R. 1990 Modem Organizations: Organization Studies in the Postmodem World. London: Sage. C1ogg, C. C., and E. S. Shihadeh 1994 Statistical Modelsfor Ordinal Variables. Thousand Oaks: Sage. Clutterbuck. D. 1985 Everyone Needs a Mentor. London: Institute of Personnel Management. Cm3807 1997 The New NHS: Modem, Dependable. London: HMSO. Coase,R.H. 1937 'The nature of the firm.' Economica 4: 386-405. Cohen, M. D., J. G. March, and J. P. Olsen 1972 'A Garbage Can Model of Organizational Choice.' Administrative Science Quarterly 17: 1-25. Cohen, Y., and J. Pfeffer 1986 'Organizational Hiring Standards.' Administrative Science Quarterly 31: 1-24. Coleman, J. S. 1986 'Norms as Social Capital.' In G. Radnitzky and P. Bernholz (eds.), Economic Imperialism: The

Economic Approach Applied Outside the Field of Economics, 133-155. New York: Paragon House.

References - 503

1988 'Social Capital in the Creation of Human Capital.' American Journal of Sociology 94 (special supplement): 95-120.

1990 F ouruiations of Social Theory. Cambridge: The Belknap Press of Harvard University Press. 1994 'A Rational Choice Perspective on Economic Sociology.' In N. Smelser and R. Swedberg (eds.).

The Handbook of Economic Sociology, 166-180. Princeton: Princeton University Press. Coleman, J. S., E. Katz, and H. Menzel 1966 Medical innovation. New York: Bobbs-Merrill. Collln,A. 1979 'Notes on Some Typologies of Managerial Development and the Role of Mentor in the Process of

Adaptation of the Individual to the Organization.' Personnel Review 8: 10-14. Contractor, F. J., and P. Lorange (eds.) 1988 Cooperative Strategies in Internationol Business: Joint Ventures and Technology Partnerships

Between Firms. Lexington: Lexington Books. Cook, K. S. 1977 'Exchange and Power in Networks of Interorganizational Relations.' Sociological Quarterly 43:

721-739. Cook, K. S., and R. M . Emerson 1978 'Power, Equity and Commitment in Exchange Networks.' American Sociological Review 43: 721-

739. Cook, K. S., R. M. Emerson, M. GUlmore, and T. Yamaglshl 1983 'The Distribution of Power in Exchange Networks: Theory and Experimental Results.' American

Journal of Sociology 89: 275-305. Corfman, K. P., and D. R. Lehmann 1987 'Models of Cooperative Group Decision-Making and Relative Influence: An Experimental

Investigation of Family Purchase Decisions.' Journal of Consumer Research 14: 1-13. Cox,A. 1997 Business Success. UK: Earlsgate Press. Cronin, J. J., and S. A. Taylor 1992 'Measuring Service Quality: A Reexamination and Extension.' Journal of Marketing 56: 55-68. Crosby, L. A., and N. Stephens 1987 'Effects of Relationship Marketing on Satisfaction. Retention, and Prices in the Life Insurance

Industry.' Journal of Marketing Research 24: 404-411. Crosby, L. A., K. R. Evans, and D. Cowles 1990 'Relationship Quality in Services Selling: An Interpersonal Influence Perspective.' Journal of

Marketing 54: 68-81. Crozier,M. 1960 The Bureaucratic Phenomena. Glencoe: Free Press. Crozier, M., and E. Friedberg 1977 L'Acteur et Ie Systeme. Paris: Seuil. Cyert, R. M., and J. G. March 1963 A Behavioral Theory of the Finn. Englewood Cliffs: Prentice-Hall. Daft, R. L. 1995 Organization Theory and Design . (Fifth Edition) St. Paul: West. Daloz,L.A. 1986 Effective Teaching and Mentoring. San Francisco: Jossey-Bass. Dasgupta, P., and P. David 1987 'lnfonnation Disclosure and the Economics of Science and Technology.' In G. R. Feiwel (ed.),

Arrow and the Ascent of Modem Economic Theory, 519-542. New York: New York University Press.

1994 'Toward a New Economics of Science.' Research Policy 23: 487-521. D'Avenl, R. A. 1990 'Top Managerial Prestige and Organizational Bankruptcy.' Organization Science, I : 121-142. Davidow, W. H., and M. S. Malone 1992 The Virtual Cooperation: Structuring and Revitalizing the Cooperation for the 21st Century. New

York: HarperBusiness. Davis,G. F. 1991 'Agents Without Principles? The Spread of the Poison Pill through the Intercorporate Network.'

Administrative Science Quarterly 36: 583-613.


Davis, G. F., K. A. Diekmann, and C. H. Tinsley 1994 The Decline and Fall of the Conglomerate Finn in the 1980s: The De-Institutionalization of an

Organizational Fonn.' American Sociological Review 59: 547-570. Davis, H. L., S. J. Hoch, and E. K. Easton Ragsdale 1986 'An Anchoring and Adjustment Model of Spousal Predictions.' Journal of Consumer Research 13:

25-37. Davis, G. F., and M. Mizruchi 1997 'The Money Center Cannot Hold: Conunercial Banks in the US System of Governance.' Working

paper. Davis, G. F., C. H. Tinsley, and K. A. Diekmann 1997 'Acquisitions by Large Industrials in the late 1980s: The Twilight of Unrelated Diversification?'

Working paper, Columbia University. Davis,J.A. 1966 'Structural Balance, Mechanical Solidarity, and Interpersonal Relations.' In J. Berger, M. Zelditch,

and B. Anderson (eds.), Sociological Theories in Progress I : 74- \OJ. Boston: Houghton Mifflin. Davis, K., and J. W. Newstrom 1985 Human Behavior at Work: OrganiZlltional Behavior (7th edition). New York: McGraw-Hill. Dawes,RM. 1980 'Social Dilemmas.' Annual Review of Psychology 31: 169-193. De Graaf, N. D., and IL D. Flap 1988 With a little Help from My Friends: Social Resources as an Explanation of Occupational Status and

Income in West Gennany, The Netherlands, and The United States.' Social Forces 67: 452-472. Dearborn, D. C., and H. A. Simon 1958 'Selective Perception: A Note on the Departmental Identifications of Executives.' Sociometry 21:

140-144. Dess, G. G., and R. L. Priem 1995 'Consensue-Perfonnance Relationship Research: Theoretical and Empirical Extensions.' Journal of

Management Studies 32. Deutsch,M. 1973 The Resolution of Conflict: Constructive and Destructive Processes. New Haven: Yale University

Press. Devine, T. J., and N. M. Kiefer 1991 Empirical Labour Economics: The Search Approach. New Yorlc: Oxford University Press. Diamond, D. W. 1984 'Financiallntennediation and Delegated Monitoring.' Review of Economic Studies 51: 393-414. Dickson, P. H. And M. K. Weaver 1997 'Environmental Determinants and Individual-Level Moderators of Alliance Use.' Academy of

Management Journal 40: 404-425. Diener, E., And R A. Emmons 1985 'The Independence of Positive and Negative Affect.' Journal of Personality and Social Psychology

47: 1758-1768. Dimaggio, P. J . 1994 'Culture and Economy.' In N. Smelser and R. Swedberg (eds.), The Handbook of Economic

Sociology, 27-57. Princeton: Princeton University Press. Dimaggio, P. J., and W. W. Powell 1983 'The Iron Cage Revisited: Institutional Isomorphism and Collective Rationality in Organizational

Fields.' American Sociological Review 48: 147-160. DiPrete, T. A. 1987 'Horizontal and Vertical Mobility in Organizations.' Administrative Science Quarterly 32: 422-

444. 1989 The Bureaucratic Labor Market: The Case of the Federal Civil Service. New Yorlc: Plenum Press. Ditcher, S. F. 1991 'The Organization of the '90s.' The McKinsey Quarterly I: 145-155. Dobbin, F., L. Edelman, J. W. Meyer, W. R Scott, and A. Swidler 1988 'The Expansion of Due Process in Organizations.' In L. Zucker (ed.), Institutional Patterns and

OrganiZlltions: Culture and Environment, 71-98. Cambridge: Ballinger. Dodgson,M. 1993 Technological Collaboration in Industry. London: Routledge.

Doherty,J. 1997 'The New Drexel.' Barron's (September 8): 17-21. Domhoff, G. W.

References - 505

1975 'Social Clubs, Policy-Planning Groups, and Corporations: A Network Study of Ruling-Class Cohesiveness.' Insurgent Sociologist 5: 173-84.

Dore,R. 1983 'Goodwill and the Spirit of Market Capitalism.' British Journal of Sociology 34: 459-482. Doreian,P. 1985 'A Measure of Standing of Journals in in Stratified Networks.' Scientometrics 8: 341-363. 1987 'A Revised Measure of Standing of Journals in Stratified Networks.' Scientometrics II: 71-80. Doreian, P., and K. L. Woodard 1992 'Fixed List versus Snowball Selection of Social Networks.' Social Science Research 21: 216-233. Dreher, G. F., and R. A. Ash 1990 'A Comparative Study of Mentoring Among Men and Women in Managerial, Professional, and

Technological Positions.' Journal of Applied Psychology 75: 539-546. Dubois,A. 1994 Organising Industrial Activities: An Analytical Framework. Dept of Ind. Marketing, Chalmers

University of Technology, Sweden. Dunleavy, P., and C. Hood 1994 'From Old Public Administration to the New Public Management.' Public Money and Management

July/Sept: 9-16. Durkheim, E. 1893 De la Division du Travail Social. Paris: Presses Universitaires de France. 189711951 Suicide: A Study in Sociology. Translated by J. A. Spaulding and O. Simpson. New York: Free

Press. Dyer,J. H. 1996a 'Specialized Supplier Networks as a Source of Competitive Advantage: Evidence from the Auto

Industry.' Strategic Management Journal 17: 271-291. 1996b 'How Chrysler Created an American Keiretsu.' Harvard Business Review 74: 42-56. 1997 'Improving Performance by Transforming Arm's-Length Relationships to Supplier Partnerships:

the Chrysler Case.' OrganilJltion Science (forthcoming). Easton, G., and L. Araujo 1993 'Language, Metaphors and Networks.' In D. Sharma (ed.), Advances in International Marketing:

Industrial Networks 5: 67-85. Greenwich: JAI Press. 1996 'Characterising Resources as an Organisational Competence: an Industrial Networks Approach.' In

R. Sanchez. A. Heene, and H. Thomas (eds.), Dynamics of Competence Based Competition: Theory and Practice in the New Strategic Management, 183-207. New York: Pergamon Press.

Easton, G., and M. Quayle 1990 'Single and Multiple Network Sourcing-Network Implications.' Proc. of 6th IMP Conference,

Milan. Ebeling, H., and L. Doorley 1983 'A Strategic Approach to Acquisitions.' Journal of Business Strategy 3: 44-55. Eccles, R. G. 1981 'Bureaucratic versus Craft Administration: The Relationship of Market Structure to the

Construction Firm.' Administrative Science Quarterly 26: 449-469. Eccles, R. G., and D. B. Crane 1988 Doing Deals: Investment Banks at Work. Boston: Harvard Business School Press. Eccles, R. G., and H. C. White 1988 'Price and Authority in Inter-Profit Center Transactions.' American Journal of Sociology 94

(special supplement): 17-51. Economist, The 1993 'Within the Wirldwind. A Survey of the Computer Industry.' The Economist February 27. Edstrom, A., and J. Galbraith 1977 'Transfer of Managers as a Coordination and Control Strategy in Multinational Organizations.'

Administrative Science Quarterly 22: 248-263. Edwards, R. 1979 Contested Terrain: The Transjonnation of the Workplace in the Twentieth Century. New York:

Basic Books.


Eisenberg, E. M., P. R. Monge, and K. I. Miller 1984 'Involvement in Communication Networks as a Predictor of Organizational Commitment.' Human

Communication Research 10: 179-201. Eisenhardt, K. M., and C. B. Schoonhoven 1996 'Resource-Based View of Strategic Alliance Fonnation: Strategic and Social Effects in

Entrepreneurial Firms.' Organiwtion Science 7: 136--150. Ekeh,P.P. 1974 Social Exchange Theory: The Two Traditions . Cambridge: Harvard University Press. EI-Ansary, A. I., and L. W. Stem 1972 'Power Management in the Distribution Channel.' Journal of Marketing Research 9: 47-52. ElIram,L. 1991 'Supply Chain Management: The Industrial Organisation Perspective.' International Journal of

Physical Distribution and Logistics Management 21 : 13-22. Emanuel, P. M. 1995 'Using Engineering Expertise to Develop a Successful World Wide Patent Portfolio.' In M. A.

Lechter et aI. (eds.). Successful Patenting for Scientists and Engineers. New York: IEEE Press. Emerson, R. M. 1962 'Power-Dependence relations.' American Sociological Review 27: 31-40. Emirbayer, M., and J. Goodwin 1994 'Network Analysis. Culture. and the Problem of Agency.' American Journal of Sociology 99:

1411-1454. Erens, F., R. Stoifelen, F. van de Ven, and L. Wildeman 1996 'Alliance and Networks: The Next Generation.' Proceedings of the 6th International Forum on

Technalogy Management 77- 91. Erickson, B. H. 1988 The Relational Basis of Attitudes.' In B. Wellman and S. D. Berkowitz (eds.). Social Structures: A

Network Approach. 99-121 . Cambridge: Cambridge University Press. Etgar,M. 1976 'Channel Domination and Countervailing Power in Distributive Channels.' Journal of Marketing

Research 13: 254-262. 1979 'Sources and Types of Intrachannel Conflict.' Journal of Retailing 55: 61-78. Etzkowitz, H. and A. Webster 1995 'Science as Intellectual Property.' In S. Jasanoff. O. Mankle. J. Peterson. and T. Pinch (eds.).

Handbook of Science and Technalogy Studies. Thousand Oaks: Sage. Evans,P. 1996 'Govemment Action. Social Capital and Development-Reviewing the Evidence on Synergy.'

World Development 24: 1119-1132. Evron, Y. 1992 The Military Industry in Israel. Tel Aviv, Ministry of Defense (in Hebrew). Faase,L. 1980 Tussenschakels op de Arbeidsmarkt.' [Links on the Labor Market, in Dutch]. Econamisch Statistische

Berichten 65: 1047-1049. Fagenson, E. A. 1988 The Power of a Mentor: Proteg~ and Nonproteg~' Perceptions of Their Own Power in

Organizations.' Group and Organiwtion Studies 13: 182-192. 1989 The Mentor Advantage: Perceived Career/Job Experiences of Proteg~ Versus Non-Proteg~.'

Journal ofOrganiwtional Behavior 10: 309-320. 1994 'Perceptions of Proteges' vs. Nonproteges' Relationships with Their Peers. Superiors. and

Departments.' Journal of Vocational Behavior 45: 55-78. Farmer, D. H., and R. Ploos von Amstel 1991 Effective Pipeline Management. UK: Gower. Farr, C. M., and W. A. Fischer 1992 'Managing International High Technology Cooperative Projects.' R&D Management 22: 55-67. Feld,S.L. 1981 'The Focused Organization of Social Ties.' American Journal of Sociology 86: 10 15-1 035. Fenton, E. 1996 The Internal Network Organisation: A Review. University of Warwick: Centre for Corporate

Strategy and Change.

References - 507

Ferlie, E. B., and A. M. Pettigrew 1996 'Managing Through Networks-Some Issues and Implications for the NHS.' British Journal of

Management 7: 81-99. Ferlie, E. B., L. Ashbumer, L. Fitzgerald, and A. M. Pettigrew 1996 The New Public Management in Action. Oxford: Oxford University Press. Fernandez, R. M., and N. Weinberg 1997 'Sifting and Sorting: Personal Contacts and Hiring in a Retail Bank.' American Sociological

Review 62: 883-902. Ferris, G. R., and T. A. Judge 1991 'PersonnellHuman Resources Management: A Political Influence Perspective.' Journal of

Management 17: 447-488. Festinger, L. 1954 'A Theory of Social Comparison Processes.' Human Relotions 7: 117-140. Festinger, L., S. Schachter, and K. Back 1950 Social Pressures in Informal Groups: A Study of Human Factors in Housing. Palo Alto: Stanford

University Press. Fevre,R. 1989 'Informal Practices, Aexible Firms, and Private Labour Maricets.' Sociology 23: 91-109. Fie!, J. 1991 'Network Reliance by Venture Capital Firms and Business Angels: An Empirical and Theoretical

Test.' In Frontiers of Entrepreneurial Research, 445-455. Wellesley: Babson College. Finkelstein, S., and D. C. Hambrick 1990 Top Management Team Tenure and Organizational Outcomes: The Moderating Role of

Managerial Discretion.' Administrative Science Quarterly 35: 484-503. Flscher,C. 1982 To Dwell Among Friends. Personal Networks in Town and City. Chicago: University of Chicago

Press. Flannery, T. P., D. A. Hofrichter, and P. E. Platten 1996 People, Performance, and Pay: Dynamic Compensation for Changing Organiwtions. New York:

Free Press. Flap,H.D. 1988 Conflict, Loyalty and Violence. The Effects of Social Networks on Behaviour. Bern: Peter Lang

Verlag. 1990 'Patronage: An Institution in its Own Right.' In M. Hechter, K. D. Opp and R. Wippler (cds.),

Social Institutions: Their Emergence, Maintenance and Effects. New York: A1dine de Gruyter. 1991 'Social Capital in the Reproduction of inequality: A Review.' Comparative Sociology of Family,

Health and Education 20: 6179-6202. Flap, H. D., and E. Boxman 1996 'Getting Started. The Influence of Social Capital on the Start of the Occupational Career.'

Communication ala Journie-dibat avec Harrison White, 3 Mai 1996, CNRS, Paris. Flap, H. D., B. Bulder, and B. VOlker 1998 'Intra-Organizational Networks and Perfonnance. A Review.' Computational and Mathemotical

Organiwtion Theory 4: 109-147. Flap, H. D., T. Snijders, and F. Van Wlnden 1996 'Creation and Returns of Social Capital. Social Networks in Education and Labour Maricets.'

Application for a research grant from the Dutch Science Foundation. Flap, H. D., and N. D. De Graat 1986 'Social Capital and Attained Occupational Status.' Nether/ands Journal of Sociology 22: 145-161. Fligstein. N. 1985 The Spread of the Multidivisional Form Among Large Firms, 1919-1979.' American Sociological

Review 50: 377-391. 1990 The Transformation of Corporate Control. Cambridge: Harvard University Press. 1991 The Structural Transformation of American Industry: An Institutional Account of the Causes of

Diversification in the Largest Firms, 1919-1979.' In W. W. Powell and P. J. DiMaggio (eds.), The New Institutionalism in Organiwtional Analysis, 311-336. Chicago: University of Chicago Press.

1996 'Maricets As Politics: A Political-Cultural Approach to Maricet Institutions.' American Sociological Review 61 : 656-673.


Florida, R., and M. Kenney 1988 'Venture Capital and High Technology Entrepreneurship.' Journal of Business Venturing 3: 301-

319. Flynn, R., G. Williams, and S. Pickard 1996 Markets and Networks: Contracting in Community Health Services. Buckingham: Open

University Press. Foley, M. W., and B. Edwards 1997 'Escape from Politics? Social Theory and the Social Capital Debate.' American Behavioral

Scientist 4(J: 550--561. Fombrun, C. J. 1996 Reputation: Realizing Valuefrom the Corporate Image. Boston: Harvard Business School Press. Ford, I. D. 1978 'Stability Factors in Industrial Marketing Channels.' Industrial Marketing Management 7: 410--

422. Forrest, J. E., and M. J. C. Martin 1992 'Strategic Alliances Between Large and Small Research Intensive Organizations: Experiences in

the Biotechnology Industry.' R&D Management 22: 41-53. Forrester, J. W. 1961 Industrial Dynamics. MIT Press. Forse,M. 1997 'Capital Social et Emploi.' L'Annee sociologique 47: 143- 181. Foss, N.J. 1993 'Theories of the Fmn: Contractual and Competence Perspectives.' Journal of Evolutionary &onomics

3: 12744. Foss, N. J., and C. Knudsen 1996 Towards a Competence Theory of the Firm. London: Routledge. Frank,O. 1991 'Statistical Analysis of Change in Networks.' Statistica Neerlandica 3: 283-293. Frank, 0 ., and D. Strauss 1986 'Markov Graphs.' Journal of the American Statistical Association 81 : 832-842. Frazier, G. L., R. E. Spekman, and C. R. O'Neal 1988 'lust-in-Time Exchange Relationships in Industrial Markets.' Journal of Marketing 52: 52-67. Freeman, J. H. 1986 'Entrepreneurs as Organizational Products: Semiconductor Fmns and Venture Capital Fmns.' In G.

Libecap (ed.), Advances in the Study of Entrepreneurship, Innovation and &onomic Growth. Greenwich: lAI Press.

Freeman, J. H., G. R. Carrou, and M. T. Hannan 1983 'The Liability of Newness: Age Dependence in Organizational Death Rates.' American Sociological

Review 48: 692-710. Freeman, L. C. 1977 'A Set of Measures of Centrality Based on Betweenness.' Sociometry 4(J: 35-41 . 1979 'Centrality in Social Networks I. Conceptual Clarification.' Social Networks I : 215-239. Freeman, R. B., and J . L. Medoff 1984 What Do Unions Do? New York: Basic Books. French, J. R. P.,Jr., and B. Raven 1968 'The Social Bases of Power.' In D. Cartwright and A. Zander (eds.), Group Dynamics (3rd ed.),

259-269. New York: Harper and Row. Frenzen, J. K., and K. Nakamoto 1993 'Structure, Cooperation, and the Flow of Market Information.' Journal of Consumer Research 20:

360-375. Fried, V. H., and R. D. Hisrich 1995 'The Venture Capitalist: A Relationship Investor.' California Management Review 37: 101-113. Friedberg, E. 1993 Le Pouvoir et La Regie. Paris: Seuil. Friedkin, N. E. 1998 A Structural Theory of Social Influence . New York: Cambridge University Press.

References - 509

Friedman, R. A., and J . Podolny 1992 'Differentiation of Boundary Spanning Roles: Labor Negotiations and hnplications for Role

Conflict.' Administrative Science Quarterly 37: 28-47. Fruin,M. 1992 The Japanese Enterprise System. Oxford: Clarendon Press. Fukuyama, F. 1995a 'Social Capital and the Global Economy.' Foreign Affairs 5: 89-103. 1995b Trust: The Social Virtues and the Creation of Prosperity. New York: The Free Press. Gabbay, S. M. 1997 Social Capital in the Creation of Financial Capital. Ph.D. dissertation, Columbia University .. 1997 Social Capital in the Creation of Financial Capital: The Case of Network Marketing. Champaign:

Stipes Publishing. Gabbay, S. M, and R. Th. A. J. Leenders 1999'Corporate Social Capital (CSC) and Global Market Competition.' Paper presented at the

International Sociological Association Meetings, Tel Aviv, Israel. Gabbay, S. M., M. Semyonov, and S. Mishal 1999 'Entrepreneurial Networks in an Era of Peace.' The Chaim Center for Middle East Studies and

Diplomacy. Ben Gurion University. Gabbay, S. M, and Y. Sato 1996 'Revisiting Social Capital through hnages of Social Structure.' Japanese Mathematical Sociology

19: 8-21. Gabbay, S. M, and A. J. Stein 1999 'Embedding Social Structure in Technological Infrastructure: Constructing Regional Social Capital

for a Sustainable Peace in the Middle East.' In J. Wright (ed.), The Political Economy of Middle East Peace. Boston: Routledge.

Gabbay, S. M., and E. Zuckerman 1998 'Social Capital and Opportunity in Corporate R&D: The Contingent Effect of Contact Density on

Mobility Expectations.' Social Science Research 27: 189-217. Galanter, M., and T. Palay 1991 Tournament of Lawyers: The Transformation of the Big Law Firm. Chicago: The University of

Chicago Press. Galasldewicz, J. 1979 Exchange networks and community politics. Beverly Hills: Sage. 1985 Social OrganiZlltion of an Urban Grants Economy. Orlando: Academic Press. Gallini, N. T. 1992 'Patent Policy and Costly Imitation.' RAND Journal of Economics 23: 52-63. Gambardella, A. 1992 'Competitive Advantages from In-House Scientific Research. The US Pharmaceutical Industry in

the 1980s.' Research Policy 21 : 391-407. Gambetta, D. 1988a 'Can We Trust Trust?' In D. Gambetta (ed.), Trust: Making and Breaking Cooperative Relations,

213-238. Oxford: Blackwell. 1988b 'Mafia: The Price of Distrust.' In D. Gambetta (ed.), Trust: Making and Breaking Cooperative

Relations, 158-175. Oxford: Blackwell. Gambetta, D. (ed.) 1988c Trust: Making and Breaking Cooperative Relations. New York: Blackwell. Gargiulo, M. 1993 Two-Step Leverage: Managing Constraint in Organizational Politics.' Administrative Science

Quarterly 39: 1-19. Gartner, W. B. 1985 'A Conceptual Framework for Describing the Phenomenon of New Venture Creation.' Academy of

Management Review 10: 696-706. 1989 "Who Is an Entrepreneur?' Is the Wrong Question.' Entrepreneurship Theory and Practice 13: 47-

68. 1990 'What Are We Talking About When We Talk About Entrepreneurship?' Journal of Business

Venturing 5: 15-28.


Gasse, Y. 1982 'Elaborations on the Psychology of the Entrepreneur.' In C. A. Kent, D. L. Sexton, and K. H.

Vesper (eds.), Encyclopedia of Entrepreneurship, 57-71. Englewood Cliffs: Prentice-Hall. Gattoms, J. 1978 'Channel of Distribution Conceptualisations: a State of the Art Review.' European Journal of

Marketing 12: 470-512. Gelauff, G. M. M., and C. den Broeder 1996 'Governance of Stakeholder Relationships, The Gennan and Dutch Experience.' Research

Memorandum 127. The Hague: Central Planning Bureau. Gerlach, M. L. 1987 'Business Alliances and the Strategy of the Japanese Fmn.' In G. Carroll and C. Vogel (eds.),

OrganilJJtional Approaches to Strategy. Berkeley: University of Califiornia Press. 1992a Alliance Capitalism: The Social OrganilJJtion of Japanese Business. Berkeley: University of

California Press. 1992b 'Twillight of the Keiretzu: A Critical Assessment'. Journal of Japanese Studies 18: 79-118. Ghoshal S., and P. Moran 1996 'Bad for Practice: A Critique of Transaction Cost Theory.' Academy of Management Review 21:

13-47. Ghoshal, S., and N, Nohria 1989 'Internal Differentiation within Multinational Corporations.' Strategic Management Journal 10:

323-337. Ghoshal, S., H. Korine, and G. Szulanski 1994 'Interunit Conununication in Multinational Corporations.' Management Science 40: 96-110. Giddens, A. 1976 New Rules of Sociological Method. London: Hutchinson. 1979 Central Problems in Social Theory. London: Macmillan. Gilad, N., S. Kalsh, and J. Ronen 1989 'Infonnation, Search, and Entrepreneurship: A Pilot Study.' The Journal of Behavioral Economics

18: 217-231. Gilroy, B. M, 1993 Networking in Multinational Enterprises: The Importance of Strategic Alliances. Columbia:

University of South Carolina Press. Gilson, R. J" and R. H. Mnookin 1985 'Sharing Among Human Capitalists: An Economic Inquiry into the Corporate Law Fmn and How

Partners Split Profits.' Stanford lAw Review 37: 313-392. Glmeno, J., and C, Woo 1996 'Economic Multiplexity: The Structural Embeddedness of Cooperation in Multiple Relations of

Interdependence.' In J. A. C. Baum and J. E. Dutton (eds.), Advances in Strategic Management: The Embeddedness of Strategy, 323-361. Greenwich: JAI Press.

Gilman, L. J. 1979 Principles of Managerial Finance. New York: Harper and Row. Glain,S, 1997 'How Beijing Officials Outnegotiated AT&T on Marine Cable Plans.' Wall Street Journal (July

16). New York: Dow Jones & Company, Inc. Go~asseres, B, 1996 The Alliance Revolution: The New Shape of Business Rivalry. Cambridge: Harvard University

Press. Goodman L. A. 1984 The Analysis of Cross-Classified Data Having Ordered Categories. Cambridge: Harvard

University Press. 1985 'The Analysis of Cross-Classified Data Having Ordered andlor Unordered Categories: Association

Models, Correlation Models, and Asymmetry Models for Contingency Tables with or without Missing Entries.' Annals of Statistics 13: 10-69.

Gouldner, A, W. 1960 'The Nonn of Reciprocity.' American Sociological Review 25: 161-178.

References - 511

Grabber,G. 1993 'The Weakness of Strong Ties: the Lock-in of Regional Development in the Rube Area.' In G.

Grabher (ed.), The Embedded Firm: On the Socioeconomics of Industrial Networks, 255-277. New York: Routledge.

Granovetter, M. S. 1973 'The Strength of Weak Ties.' American Journal of Sociology 78: 13®-1380. 1982 The Strength of Weak Ties: A Network Theory Revisited.' In P. V. Marsden and N. Lin (eds.),

Social Structure and Network Analysis, 105-130. Beverly Hills: Sage. 1985 'Economic Action and Social Structure: The Problem of Embeddedness.' American Journal of

Sociology 91: 481-510. 1988 'The Sociological and Economic Approaches to Labour Market Analysis, a Social Structural View.' In

G. Farkas and P. Fngland (eds.), Industries, Firms and Jobs: Sociological and Economic Approaches, 187-217. New York: Plenum Press.

1993 The Nature of Economic Relationships.' In R. Swedberg (ed.), Explorations in Economic Sociology, ~1. New York: Russell Sage.

1994 'Business Groups.' In N. Smelser and R. Swedberg (eds.). The Handbook of Economic Sociology. 453-475. Princeton: Princeton University Press.

1995 Getting A Job: A Study of Contacts and Careers. Chicago: University of Chicago Press (first edition published in 1974).

Greeley,A. 1997 'Coleman Revisited: Religious Structures as a Source of Social Capital.' American Behavioral

Scientist 40: 587-594. Gregory, D. A., C. W. Saber, and J. D. Grossman 1994 Introduction to Intellectual Property Law. Washington: The Bureau of National Affairs. Greif,A. 1994 'Cultural Beliefs and the Organization of Society-A Historical and Theoretical Reflection on

Collectivist and Individualist Societies.' Journal of Political Economy 102: 912-950. Grieco,M. 1987 Keeping it in the Family. Oxford: Oxford University Press. Grillcbes, Z. 1990 'Patent Statistics as Economic Indicators: A Survey.' Journal of Economic literature 28: 1661-

1707. Grimm, C. M., and K. G. Smltb 1991 'Management and Organizational Change: A Note on the Railroad Industry.' Strategic

AlanagementJournaI12:557-562. Grindley, P. C., and D. J. Teece 1997 'Managing Intellectual Capital: Licensing and Cross-Licensing in Semiconductors and

Electronics.' California Alanagement Review 39: 8-4 I. Groenewegen, J. 1997 'Institutions of Capitalism: American. European and Japanese Systems Compared.' Journal of

Economic Issues 31 : 333-347. Gronroos, C. 1990 Service Alanagement and Alarlceting. Lexington: Lexington Books. Guillen, M. F. 1997 'Business Groups in Economic Development.' Best papers Proceedings. 170-174. Academy of

Management. Annual Meeting. Boston. Gulatl,R. 1995a 'Social Structure and Alliance Formation Pattern: A Longitudinal Analysis.' Administrative

Science Quanerly40: 619-652. 1995b 'Does Familiarity Breed Trust? The bnplications of Repeated Ties for Contractual Choices in

Alliances.' Academy of Alanagement Journal 38: 85-112. Gummesson, E. 1987 'Lip Service: A Neglected Area in Services Marketing.' Journal of Services Alarlceting 1: 19-23. Guo,G. 1993 'Event-History Analysis for Left-Truncated Data.' Sociological Alethodology 23: 217-243. Gupta, A. K., and V. Govlndrajan 1991 'Knowledge Flows and the Structure of Control within Multinational Corporations.' Academy of

Alanagement Review 16: 768-792.


Hackman, J. R., and R. Wageman 1995 'Total Quality Management: Empirical. Conceptual, and Practical Issues.' Administrative Science

Quanerly 40: 309-342. Hagedoorn, J. 1993a 'Strategic Technology Alliances and Modes of Cooperation in High-Technology Industries.' In G.

Grabher (ed.), The Embedded Firm: On the Socioeconomics of Industrial Networks, 116-137. New York: Routledge.

1993b 'Understanding the Rational of Strategic Technology Partnering: International Modes of Cooperation and Sectoral Differences.' Strategic Management 10umal14: 371-85.

Hagedoorn, J., and J. Schakenraad 1994 'The Effect of Strategic Technology Alliances on Company Performance.' Strategic Management

loumal 15: 291-309. Hakansson, H. (ed.) 1982 International Marketing and Purchasing of Industrial Goods: an Interaction Approach.

Chichester: Wiley. 1987 Industrial Technological Development: A Network Approach. London: Croom Helm Hakansson, H., and J. Johanson 1993 The Network as a Governance Structure: Interfirm Cooperation Beyond Markets and Hierarchies.'

In F. Grabber (ed.), The Embedded Firm. London: Routledge. Hakansson, H., and I. Snehota (eds.) 1995 Developing Relationships in Business Networks . London: Routledge. Ham,C. 1981 Policy Making in the National Health Service. London: Macmillan. Hambrick, D. C. 1981 'Environment, Strategy, and Power Within Top Management Teams.' Administrative Science

Quanerly 26: 253-275. Hambrick, D. C., and G. D. Fukutomi 1991 'The Seasons of a CEO's Tenure.' Academy o/Management Review 16: 719-742. Hambrick, D. C., and P. A. Mason 1984 'Upper Echelons: The Organizations as a Reflection of its Top Managers.' Academy 0/

Management Review 9: 193-206. Hamel, G., Y. L. Doz, and C. K. Prahalad 1989 'Collaborate with Your Competitors-And Win.' Harvard Business Review 67: 133--139. Hampson, K., and T. Kwok 1996 'Analysing Strategic Alliances in Building Construction.' In M. Sobolewski and M. Fox (eds.),

Advances in Concurrent Engineering, 172-180. Toronto: Technomic Publishing. Han,S.-K. 1993 'Churning Firms in Stable Markets.' Social Science Research 21: 406-418. 1994 'Mimetic Isomorphism and its Effect on the Audit Services Market.' Social Forces 73: 637-663. 1996 'Structuring Relations in On-the-Job Networks.' Social Networks 18: 47...fJ7. Han, S.-K., and R. L. Breiger 1996 'Switching and Stacking: Network Multiplexity in the Supreme Court.' Paper presented at the 16th

Sunbelt Intemational Social Network Conference, Charleston, SC. Hannan, M. T., and G. R. Carron 1992 Dynamics of OrganiZlltional Populations: Density, Competition and Legitimation. New Yorle Oxford

University Press. HalllUlIl, M. T., and J. Freeman 1984 'Structural Inertia and Organizational Change.' American Sociological Review 49: 149...fJ4. 1989 OrganiZlltional Ecology. Cambridge: Harvard University Press. Hannerz, U. 1969 Soulside: Inquiries into Ghetto Culture and Community. New York: Columbia University Press. Hannifan, L,J. 1920 The Community Center. Boston: Silver Burdette. Hansen, M. N. 1996 'Earnings in Elite Groups: the Impact of Social Class Origin.' Acta Sociologica 39: 385-408. Hansen, M. T. 1997 'Integrating Knowledge through the Intra-Company Network: Implications for Product

Development Time.' Working Paper, Harvard Business School.

Hardin,R. 1993 'The Street-Level Epistemology of Trust.' Politics and Society 21: 505-529. 1996 'Trustworthiness.' Ethics 107: 26-42. Harland, C. M.

References - 513

1995 'The Dynamics of Customer Dissatisfaction in Supply Chains.' Production Planning and Control. Supply Chain Management 6 (special issue): 209-217.

1996a 'Supply Chain Management: Relationships, Chains and Networks.' British Journal of Management 7: 63-80.

1996b 'Supply Network Strategies: the Case of Health Supplies.' European Journal of Purchasing and Supply Management 2: 183-192.

1997a 'Empirical Research in Characterising Supply Networks.' Proc. of lst International Conference on Responsive ManUfacturing, 27-37. September, Univ. of Nottingham.

1997b 'Supply Chain Operational Performance Roles.' Integrated Manufacturing Systems 8: 6-14. Harland, C. M., and J. R. Wensley 1996 'Supply Chain Management as a Response to the Strategy Leaf Kickers.' Paper Presented at BAM

Conference, September, Aston University. Harrigan, K. R. 1983 Strategies for Venical Integration. Lexington: Lexington Books. Harris, R. C., R. C. Insinga, J. Morone, and M. J. Werle 1996 'The Virtual R&D Laboratory.' Research Technology Management 39: 32-36. Haspeslagb, P. C. 1982 'Portfolio Planning: Uses and Limits.' Harvard Business Review 60: 58-73. Haspeslagh, P. C., and D. B. Jemison 1991 Managing Acquisitions: Creating Value through Corporate Renewal. New York: Free Press. Haunsehild, P. R. 1993 'lnterorganizational hnitation: The Impact of Interlocks on Corporate Acquisition Activity.'

Administrative Science Quanerly 38: 564-592. 1994 'How Much Is That Company Worth?: Interorganizational Relationships, Uncertainty, and

Acquisition Premiums.' Administrative Science Quarterly 39: 391-411. Hauser, R. M. 1980 'On 'Stratification in a Dual Economy' .' American Sociological Review 45: 702-12. Hayek, F. A. 1978 New Studies in Philosophy, Politics, Economics and the History of Ideas. Chicago: The University

of Chicago Press. Hayes, R. H., and W. J. Abernathy 1980 'Managing Our Way to Economic Decline.' Harvard Business Review 58: 67-77. Hayes, R. H., and S. C. Wheelwrigbt 1984 Restoring our Competitive Edge: Competing Through Manufacturing . New York: Wiley. Hayes, S. L., m 1971 'Investment Banking Power Structure in Flux.' Harvard Business Review 49: 136-152. 1979 'The Transformation of Investment Banking.' Harvard Business Review 57: 153-170. Haywood-Farmer, J., and J. Nollet 1991 Services Plus: Effective Service Management. Boucherville, Quebec: Morin. Hecbter,M. 1987 Principles of Group Solidarity. Berkeley: University of California Press. Heinz, J. P., E. O. Laumann, C. L. Cappell, T. C. Halliday, and M. Scbaalman 1982 Chicago Lawyers: The Social Structure of the Bar. New York: Russell Sage Foundation and

American Bar Foundation. Heinz, J. P., E. O. Laumann, R. L. Nelson and P. S. Schnorr 1997 'The Constituencies of Elite Urban Lawyers.' Law and Society Review 31: 441-72. Helper,S. 1990 'Comparative Supplier Relations in the US and Japanese Auto Industries: An Exit-Voice

Approach.' Business Economic History 19: 153-162. 1991 'Strategy and Irreversibility In Supplier Relations--The Case of the United-States Automobile

Industry.' Business History Review 65: 781-824. Herrigel, G. 1996 Industrial Constructions: The Sources of German Industrial Power. New York: Cambridge

University Press.


Heskett, J. L., T . O. Jones, G. W. Loveman, W. Earl Sasser, Jr., and L. A. Scblesinger 1994 'Putting the Service-Profit Chain to Work.' Harvard Business Review 72: 164-174. Heydebrand, W. V. 1989 'New Organizational Fonns.' Work and Occupations 16: 323-57. Hibbard, J ., and D. Iacobucci 1997 'Toward an Encompassing Theory of Relationships in Business and Consumer Marketing: A

Systematic Empirical Generalization Analysis and Future Directions.' Working Paper. Hkks,J. 1942 The Social Framework: An Introduction to Economics. New York: Clarendon Press Higgins, M. C. 1997a 'When is Helping Helpful? Problems of Individual Learning in Organizations.' Harvard Business

School Working Paper #97-044. Boston: Harvard Business School Division of Research. 1997b 'Early Career Change: Toward a Portfolio Theory of Relations.' Harvard Business School Working

Paper #98-016. Boston: Harvard Business School Division of Research. Higgins, M. C., and D. A. Thomas 1997 'Mentoring. Mobility. and Organizational Attachment in the Careers of Lawyers: A Longitudinal

Study.' Harvard Business School Working Paper #97-080. Boston: Harvard Business School Division of Research.

Hill, L., and N. Kamprath 1991 'Beyond the Myth of the Perfect Mentor: Building a Network of Developmental Relationships.'

Harvard Business School Teaching Note #9-491-096. Boston: Harvard Business School Division of Research.

Hill, T. 1985 Manufacturing Strategy. London: Macmillan. Hirsch, P. M. 1972 'Processing Fads and Fashions: An Organization-Set Analysis of Cultural Industry Systems.'

American Journal of Sociology 77: 63~59. Hofstede, G. H. 1980 Culture's Consequences. Beverly Hills: Sage. 1994 Cultures and Organisations. London: Harper and Collins. Homans, G. C. 1950 The Human Group. New York: Harcourt Brace. 1961 Social Behavior: Its Elementary Forms. New York: Harcourt, Brace. and World. Hood,C. 1991 'A Public Management for All Seasons?' Public Administration 69: 3-19. Hope,K. 1982 'Vertical and Non-Vertical Class Mobility in Three Countries.' American Sociological Review 47:

99-113. Hornby,D. 1992 in 'Partnership Sourcing: Partnership Sourcing Ud. CBI HoshI, T., A. Kashyap, and D. S. ScharCstein 1990 'Bank Monitoring and Investment: Evidence from the Changing Structure of Japanese Corporate

Banking Relationships .• In R. G. Hubbard (ed.). Asymmetric Information. Corporate Finance. and Investment. 105-125. Chicago: University of Chicago Press.

HoshI, T., A. Kashyap, and D. S. ScbarCstein 1991 'Corporate Structure. Liquidity and Investment: Evidence from Japanese Industrial Groups.'

Quarterly Journal of Economics 106: 33-60. Houlihan, J . 1984 'Supply Chain Management.' Proc. of 19th International Technical Conference, BPICS. 101-110. Hout, M., O. D. Duncan, and M. E. Sobel 1987 'Association and Heterogeneity: Structural Models of Similarities and Differences.' In C. C. Clogg

(ed.), Sociological Methodology 1987. 145-184. Washington: American Sociological Association. Hsiao,C. 1986 Analysis of Panel Data. New York: Cambridge University Press. Hubbell, C. H. 1965 'An Input-Output Approach to Clique Detection.' Sociometry 28: 277-299. Humphrey, W. 1987 Managingfor Innovation: Leading Technical People. New Jersey: Prentice Hall.

References - 515

Hutt, M. D., P. H. Reingen, and J. R. Ronchetto, Jr. 1988 'Tracing Emergent Processes in Marketing Strategy Formation.' Journal of Marketing 52: 4-19. Hyatt,J. 1989 'The Last Shift: INC., February: 74-80. Iacobucci, D. 1997 'A Game and Norm Theoretic Metaphor for the Customer-Service Provider Dyadic Interaction:

Conceptual Explanations for Poor Customer Service and Managerial Implications for Its Improvement: In J. Sherry (ed.), Servicescapes: The Built Environment of the Sales Encounter. Glenview: NTC Publishing Group (forthcoming).

1998 'Services Marketing and Management: What Do We Know and Where Shall We Go?' In T. A. Swartz, D. E. Bowen, and S. W. Brown (eds.), Advances in Services Marketing and Management: Research and Practice 7. Greenwich: JAI Press (forthcoming).

Iacobucci, D., and A. Ostrom 1996 'Commercial and Interpersonal Relationships: Using the Structure of Interpersonal Relationships to

Understand Individual-to-Individual, Individual-to-Firm, and Fmn-to-Fmn Relationships in Commerce: International Journal of Research in Marketing 13: 53-72.

Iacobucci, D., and N. Hopkins 1992 'Modeling Dyadic Interactions and Networks in Marketing.' Journal of Marketing Research 29: 5-

17. Iacobucci, D., and S. Wasserman 1987 'Dyadic Social Interactions.' Psychological Bulletin 102: 293-306. 1988 'A General Framework for the Statistical Analysis of Sequential Dyadic Interaction Data:

Psychological Bulletin 103: 379-390. Ibarra, H. 1992 'Homophily and Differential Returns: Sex Differences in Network Structure and Access in an

Advertising Fmn: Administrative Science Quarterly 37: 422-447. 1993a Untangling the Web of Interconnections: An Exploration of Competing Explanations for Gender

Differences in Managerial Networks. Cambridge: Harvard University Graduate School of Business Administration mimeo.

1993b 'Network Centra1ity, Power and Innovation Involvement: Determinants of Technical and Administrative Roles.' Academy of Management Journal 36: 471-501.

1993c 'Personal Networks of Women and Minorities in Management: A Conceptual Framework: Academy of Management Review 18: 56-87.

1997 'Paving an Alternative Route: Gender Differences in Managerial Networks: Social Psychology Quarterly 60: 91-102.

Ibarra, H., and S. B. Andrews 1993 'Power, Social Influence and Sense Making: Effects of Network Centrality and Proximity on

Employee Perceptions.' Administrative Science Quarterly 38: 277-303. Imal, K.I., I. Nonaka, and H. Takeuchi 1985 'Managing the New Product Development Process: How Japanese Companies Learn and Unlearn.'

In K. Clark, R. Hayes, and C. Lorenz (eds.), The Uneasy Alliance: Managing the ProductivityTechnology Dilemma. Cambridge: Harvard Business School Press.

Inkpen,A. 1995 The Management of International Joint Ventures: An Organiltltional Learning Perspective.

London: Routledge. Itami, H. 1987 Mobilizing Invisible Assets. Cambridge: Harvard University Press. braeH, D. N., and I. Talmud 1997 'Getting Aboard: Mode of Recruitment and Gender Composition: The Case of Women Directors in

Israel' International Review of Women and Leadership 3: 26-45. braeH, D. N., and I. Talmud 1998 The Gender of Social Capital and Recruitment to the Corporate Board. Paper addressed to a

Showcase Symposium on Women in the Corporate Board, Academy of Management Conference, San Diego, CA.

Jablin, F. M., and K. J. Krone 1987 'Organizational Assimilation.' In C. Berger and S. H. Chaffee (eds.), Handbook of Communication

Science, 711-746. Newbury Park: Sage.


Jackson, M. 0., and A. Wolinsky 1996 'A Strategic Model of Social and Economic Networks.' Journal of Economic Theory 71: 44-74. Jacobs,J. 1965 The Death and Life of Great American Cities. London: Penguin. 1970 The Economy of Cities. New York: Random House. Jacoby,S. 1985 Employing Bureaucracy: Managers. Unions. and the Transformation of Work in American

Industry. 1900-1945. New York: Columbia University Press. Jaffe, A. B. 1986 Technological Opportunity and Spillovers of R&D: Evidence from r1ClIJS' Patents, Profits, and

Market Value.' American Economic Review 76: 984-1002. Jaffe, A. B., M. Trajtenberg, and R. Henderson 1993 'Geographic Localization of Knowledge Spillovers. as Evidenced by Patent Citations.' Quarterly

Journal of Economics: 577-598. Jarillo, J. C. 1988 'On Strategic Networks.' Strategic Management Journal 9: 31-41 . 1993 Strategic Networks: Creating the Borderless Organisation. Oxford: Butterworth Heinemann. Jarillo, J, C., and H. H. Stevenson 1991 'Co-operative Strategies: The Payoffs and the Pitfalls.' Long Range Planning 24. Jenkins, R., A. Bryman, J. Ford, T. KeD, and A. Beardsworth 1983 'Information in the Labor Market: The Impact of Recession.' Sociology 17: 260-267. Jensen, M. C. 1984 Takeovers: Folklore and science.' Harvard Business Review 62: 109-121. Jensen, M. C., and W. H. Meckling 1976 Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure.' Journal of

Financial Economics 3: 305-360. Johnston, R., and P. R. Lawrence 1988 'Beyond Vertical Integration: the Rise of Value Adding Partnerships.' Harvard Business Review

66: 94-101. Johnston, R., S. Chambers, C. M. Harland, A. Harrison, and N. Slack 1997 Cases in Operations Management (2nd. edition). London: Pitman. Jonash, R. S. 1996 'Strategic Technology Leveraging: Making Outsourcing Work for You.' Research Technology

Management 39: 19-25. Jones, C., W., Hesterly, and S. Borgattl 1997 'A General Theory of Network Governance: Exchange Conditions and Social Mechanisms.'

Academy of Management Review 22: 911-945. Judge,G. 1985 The Theory and Practice of Econometrics (2nd edition). New York: Wiley. Kahneman, D., and A. Tversky 1982 'The Psychology of Preferences.' Scientific American 246: 161-173. Kahneman, D., and A. Tversky 1984 'Choices, Values, and Frames.' American Psychologists 39: 341-350. Kalleberg, A. L., D. Knoke, and P. V. Marsden 1995 'lnterorganizational Networks and the Changing Employment Contract.' Connections 18: 32-49. Kalleberg, A. L., D. Knoke, P. V. Marsden, and J. L. Spaeth 1996 Organizations in America: Analyzjng Their Structures and Human Resource Practices. Newbury

Park: Sage. Kalleberg, A. L., and J. Lincoln 1988 'The Structure of Earnings Inequality in the United States and Japan.' American Journal of

Sociology 94 (special supplement): 121-153. Kanter, R. M. 1977 Men and Women of the Corporation. New York: Basic Books. 1983a The Change Masters: Innovation and Entrepreneurship in the American Corporation . New York:

Simon & Schuster. 1983b'Variations in Managerial Career Structures in High-Technology Firms: The Impact of

Organizational Characteristics on Internal Labor Market Patterns.' In P. Osterman (ed.), Internal Labor Markets. 109-131. Cambridge: MIT Press.

References - 517

1989 When Giants Learn to Dance. London: Unwin. 1991 The Future of Bureaucracy and Hierarchy in Organizational Theory: A Report from the Field.' In

P. Bourdieu and 1. S. Coleman (eds.), Social Theory for a Changing Society, 63-87. Boulder: Westview Press.

Kanter, R. M., and R. G. Eccles 1992 'Making Network Research Relevant to Practice.' In N. Nohria and R. G. Eccles (eds.), Networks and

Organiwtions: Structure, Form, and Action. Boston: Harvard Business School Press. Kanter, R. M., and P. S. Myers 1991 'lnterorganizational Bonds and Intraorganizational Behavior: How Alliances and Partnerships

Change the Organizations Forming Them.' In A. Etzioni and P. R. Lawrence (eds.), Socioeconomics: Toward a New Synthesis, 329-344. Armonk: M.E. Sharpe.

Kanter, R. M., B. Stein, and T. Jick (eds.) 1992 The Challenge of Organiwtional Change: How Companies Experience It and Leaders Guide It.

New York: Free Press. Kaplan,J. 1995 Stanup! A Silicon Valley Adventure. Boston: Houghton Mifflin. Katz,L. 1953 'A New Status Index derived from Sociometric Analysis.' Psychometrika 18: 249-56. Katz,R. 1982 The Effects of Group Longevity on Project Communication and Performance.' Administrative

Science Quanerly 27: 81-104. Katzenbach, J. R., and D. K. Smith 1993 The Wisdom of Teams: Creating the High-Performance Organiwtion. Boston: Harvard Business

School Press. Keele,R. 1986 'Mentoring or Networking? Strong and Weak Ties in Career Development.' In L. L. Moore (ed.),

Not as Far as You Think: The Realities of Working Women, 53-68. Lexington: Lexington Books. Kegan, R. 1982 The Evolving Self: Problem and Process in Human Development. Cambridge: Harvard University

Press. Kenworthy, L. 1997 'Civic Engagement, Social Capital, and Economic Cooperation.' American Behavioral Scientist 40:

645-656. Keohane, R. O. 1986 'Reciprocity in International-Relations.' International Organiwtion 40: 1-27. Kickert, W. J. M., E. H. Klijn, and J. F. M. Koppenjan 1997 Managing Complex Networks. London: Sage. Kilduff, M. 1990 'The Interpersonal Structure of Decision Making: A social Comparison Approach to

Organizational Choice.' Organiwtional Behavior and Human Decision Processes 47: 27~288. Kilduff, M., and D. Krackbardt 1993 'Schemas at Work: Making Sense of Organizational Relationships.' Unpublished manuscript. 1994 'Bringing the Individual back in: A Structural Analysis of the Internal Market for Reputation in

Organizations.' Academy of Management Journal 37: 87-108. Kim, E. U. 1997 Big Business, Strong State: Collusion and Conflict in South Korean Development, 1960-1990.

Albany: SUNY Press. Kim, Y. H. 1989 Kwangwon Kyongje Tukhye Kyongje [Stateled Economy, Preferential Treatment Economy, in

Korean]. Seoul: Ch'ung Am. Kimberly, J., and M. Evanisko 1981 'Organizational Innovation: The Influence of the Individual, Organizational, and Contextual

Factors on Hospital Adoption of Technological and Administrative Innovations.' Academy of Management Journal 24: 689-713.

Kinch,N. 1992 'Entering a Tightly Structured Network-Strategic Visions or Network Realities.' In M. Forgren

and 1. Johanson (eds.), Managing Networks in International Business, 194-214. Philadelphia: Gordon & Breach.


Kinston, W. 1992 'Is the United States Right About 'First-to-Invent'?' European IntellectlUJl Property Review 14:

223-226. Kieimann, E. 1992 Israeli Military Export in the Global Market, Tel Aviv: Am Oved (in Hebrew). Klein WooltbuIs, R. 1996 'Entrepreneurial Activity through Inter-Organisational Relationships.' Faculty of Technology and

Management, University of Twente, paper RENT X. Research in Entrepreneurship and Small Business, Brussels, 20--23 November.

Knee, D., and D. Walters 1985 Strategy in Retailing: Theory and Applications. London: Philip Allen. Knigbt, F. H. 1921 Risk. Uncenainty. and Profit. Hart, Schaftner. and Marx. Knoke,D. 1990 Organizing for Collective Action: The Political Economies of Associations. New York: Aldine de

Gruyter. Knoke, D., and R. S. Burt 1983 'Prominence.' In R. S. Burt and M. 1. Minor (eds.). Applied Network Analysis, 195-222. Beverly

Hills: Sage. Knoke, D., and J. H. Kuklinski 1982 Network Analysis. Beverly Hills: Sage. Knoke, D., and E. O. Laumann 1982 'The Social Organization of National Policy Domains: An Exploration of Some Structural

Hypotheses.' In P. V. Marsden and N. Un (eds.). Social Structure and Network Analysis, 255-270. Beverly Hills: Sage.

Knoke, D., F. U. PappI, J. Broadbent, and Y. Tsujinaka 1996 Comparing Policy Networks: Labor Politics in the U.S .. Germany and Japan. New York:

Cambridge University Press. Kocban, T. A., H. C. Katz, and R. B. McKersie 1994 The Transformation of American Industrial Relations. Ithaca: ILR Press. Koelemeijer, K. 1995 'The Retail Service Encounter: Identifying Critical Service Experiences.' In P. Kunst and J.

Lemmink (eds.), Managing Service Quality. 2~3. London: Chapman. KoelewiJn, J. 1997 Een Nationale Waan [A National Delusion, in Dutch]. Amsterdam: De Bezige Bij. Kogut,B. 1984 'Normative Observations on the International Value Added Chain and Strategic Groups.' Journal

of International Business Studies Fall: 151-167. 1985 'Designing Global Strategies: Comparative and Competitive Value Added Chains.' Sloan

Management Review Summer: 15-28. 1988 'A Study of the Ufe Cycle of Joint Ventures.' In F. J. Contractor and P. Lorange (eds.),

Cooperative Strategies in International Business: Joint Ventures and Technology Partnerships Between Firms. 169-185. Lexington: Lexington Books.

Kogut, B., and U. Zander 1996 'What Firms Do? Coordination, Identity and Learning.' Organization Science 7: 502-518. 1997 'Did Socialism Fail to Innovate? A Natural Experiment of the Two Zeiss Companies.' Working

paper. The Wharton School, University of Pennsylvania. Kogut, B., W. Shan, and G. Walker 1992 'The Make-or-Cooperate Decision in the Context of an Industry Network.' In N. Nohria and R. G.

Eccles (eds.). Networks and Organizations: Structure. Form. and Action, 348-365. Boston: Harvard Business School Press.

Kollock,P. 1993 'An Eye for an Eye Leaves Everyone Blind: Cooperation and Accounting Systems.' American

Sociological Review 58: 768-786. Koput, K. W., W. W. Powell, and L. Smith-Doerr 1998 'Interorganizational Relations and Elite Sponsorship: Mobilizing Resources in Biotechnology.'

Paper presented at the annual meetings of the American Sociological Association, San Fransisco.

References - 519

Koput, K. W., L. Smith-Doerr, and W. W. Powell 1997 'Strategies of Learning and Industry Structure: The Evolution of Networks in Biotechnology.' In J.

P. Walsh and A. S. Huff (eds.), Organizational Learning and Strategic Management. Advances in Strategic Management Research, Vol. 14: 229-254. Greenwich: JAI Press.

Korenman, S., and D. Neumark 1991 'Does Mamage Really make Men more Productive?' Journal of Human Resources 26: 282-307. Kotter, J. P. 1982 The General Managers. New York: Free Press. Kotter, J. P., and J. L. Haskett 1992 Corporate Culture and Perfonnance. New York: Free Press. Krackhardt, D. 1990 'Assessing the Political Landscape: Structure, Cognition and Power in Organizations.'

Administrative Science Quarterly 35: 342-369. 1992 The Strength of Strong Ties: The hnportance of Philos in Organisations.' In N. Nohria and R. G.

Eccles (eds.), Networks and Organisations: Structure, Form and Action, 216-240. Boston.: Harvard Business School Press.

1995 'Entrepreneurial Opportunities in an Entrepreneurial Fmn: A Structural Approach.' Entrepreneurship: Theory and Practice 19: 53-69.

Krackhardt, D., and D. J. Brass 1994 'Intraorganizational Networks: The Micro Side.' In S. Wassennan and J. Galaskiewicz (eds.),

Advances in Social Network Analysis, 207-229. Thousand Oaks: Sage. Krackhardt, D., and J. R. Hanson 1993 'Infonnal Networks: The Company Behind the Chart.' Harvard Business Review 71 : 104-111. Krackhardt, D., and L. W. Porter 1985 'When Friends Leave: A Structural Analysis of the Relationship Between Turnover and Stayers'

Attitudes.' Administrative Science Quarterly 30: 242-261. 1986 The Snowball Effect: Turnover Embedded in Communication Networks.' Journal of Applied

Psychology 71: 5(}-'55. Krackhardt, D., and R. N. Stem 1988 'Infonnal Networks and Organizational Crisis: An Experimental Simulation.' Social Psychology

Quarterly 51: 123-140. Kram,K.E. 1985 Mentoring at Work: Developmental Relationships in Organizational life. Glenview: Scott,

Foresman and Company. Kramer, R., M. Brewer, and B. Hanna 1996 'ColJective Trust and Collective Action: The Decision to Trust as a Social Decision.' In R. Kramer

and T. Tylor (eds.), Trust in Organizations: Frontiers of Theory and Research, 357-389. London, Sage.

Kratzer, J ., J. M. L. van Engelen, R. Th. A. J. Leenders 1998 'Structuring the communication infrastructure of innovation teams for success.' Journal of Product

Innovation Management. Conference Proceedings Series 9: 43·56. Kruska1, J. B. 1964 'Multidimensional scaling by optimizing goodness of fit to a nonmetric hypothesis.' Psychometrika

29: 1-27. Laat, P. B. de 19% 'Dangerous Liaisons: Sharing Knowledge within R&D Alliances. ' Paper EMOT workshop (European

Management and Organization in Transition, of the European Science Foundation), September, Modena; Faculty of Philosophy, University of Groningen, the Netherlands.

Lablanca, G., and D. J. Brass 1997 'Exploring the Social Ledger: The Role of Negative Affective Relationships in Social Networks.'

Proceedings of the Academy of Management Meetings, Boston. Lablanca, G., D. J. Brass, and B. Gray 1998 'Social Networks and Perceptions of Intergroup Conflict: The Role of Negative Relationships and

Third Parties.' Academy of Management Journal 41 : 55-67. Lamming,R. 1989 The Causes and Effects of Structural Change in the European Automotive Components Industry.

lMVP. MIT. Cambridge. Mass. 1993 Beyond Partnership: Strategiesfor Innovation and Lean Supply. Englewood Cliffs: Prentice Hall.


Langendijk, H. P. 1990 Wisseling van Accountantskantoor: een Fata Morgana in de Toekomst? MaandbUuJ voor

Accountancy en Bedrijfseconomie 64: 264-270. Lanjouw,J. o. 1993 'Patent Protection: of what value and for how long?' NBER Working Paper #4475. Lanjouw, J. 0., A. Pakes, and J. Putnam 1996 'How to Count Patents and Value Intellectual Property: Uses of Patent Renewal and Application

Data.' NBER Working Paper #5741. Larson,A. 1992 'Network Dyads in Entrepreneurial Settings: A Study of the Governance of Exchange

Relationships.' Administrative Science Quanerly 37: 76-104. Lascelles, D. M., and B. G. Dale 1990 'Examining the Barriers to Supplier Development.' International Journal of Quality and Resource

Management. Laumann, E. O. 1973 Bonds of Pluralism: The Fonn and Substance of Urban Social Networks. New York: Wiley. Laumann, E. 0., and D. Knoke 1987 The Organizational State: Social Choice in National Policy Domains. Madison: University of

Wisconsin Press. Laumann, E. 0., and F. U. Pappi 1976 Networks of Collective Action: A Perspective on Community Influence . New York: Academic

Press. Laumann, E. 0., P. V. Marsden, and D. Prensky 1983 'The Boundary Specification Problem in Network Analysis.' In R. S. Burt and M. 1. Minor (eds.),

Applied Network Analysis: A Methodological Introduction, 18--34. Beverly Hills: Sage. Lawler, E. E., m 1992 The Ultimate Advantage: Creating the High-Involvement Organization. San Francisco: Jossey-

Bass. Lawrence, P. R., and J. W. Lorsch 1986 Organization and Environment. Boston: Harvard Business School Press. Lawrence, R. Z. 1991 'Efficient or Exclusionist? The Import Behavior of Japanese Corporate Groups.' Brookings Papers

on Economic Activity 1: 311-314. Lazega,E. 1992a 'Analyse de R~seaux d'une Organisation Collegiale: les Avocats d'Affaires.' Revue Franraise de

Sociologie 33: 559-589. 1992b Bureaucratie, Collegialite et Professionnalisme: Structure et Fonctionnement I"fonnels des

Grandes Finnes d'Avocats d'Affaires en Nouvelle Angleterre. Report to the Swiss National Foundation for Scientific Research, Geneva, Switzerland.

1993 'Bureaucratie et Collegialite dans les Firmes Americaines d'Avocats d'Affaires.' Droit et Societe, 23/24: 15-40.

1994 'Les Conflits d'lnterets dans les Cabinets Americains d'Avocats d'Affaires: Concurrence et Autoregulation.' Sociologie du Travail 35: 315-336.

1995a 'Concurrence, Cooperation et Flux de Conseils dans un Cabinet Americain d'Avocats d'Affaires: Les Echanges d'Idees entre Collegues.' Revue Suisse de Sociologie 21: 61-84.

1995b 'Protecting the Common Good Among Equals: A Lateral Control Regime of Partners in a Law Finn.' LAS MAS Preprint N° 96/1.

1998a Riseaux Sociaux et Structures Relationnelles. Paris: Presses Universitaires de France. 1998b The Collegial Phenomenon: Combining Neo-Weberian and Structural Approaches to Collective

Action Among Peers'. Unpublished manuscript. 1999 'Teaming Up and Out? Cooperation and Solidarity in a Collegial Organization.' In W.Raub and

1.Weesie (eds.), The Management of Durable Relations. Amsterdam: Thesis Publishers (forthcoming).

Lazega, E., and D. Krackhardt 1998 'Spreading and Shifting Costs of Lateral Control in a Law Partnership: Structural Analysis at the

Individual Level.' Structural Analysis 1. Forthcoming. Lazega, E., and M. O. Lebeaux 1995 'Capital Social et Contrainte Laterale.' Revue Franraise de Sociologie 36: 759-777.

References - 521

Lazega, E., and M. Van Duijn 1997 'Position in Formal Structure, Personal Characteristics and Choices of Advisors in a Law Firm: A

Logistic Regression Model for Dyadic Network Data.' Social Networks 19: 375-397. Lazega, E., and P. E. Pattison 1998 'Generalized and Multiplex Exchange in Organizational Cooperation.' Forthcoming. Lazega, E., and S. Vari 1992 'Acteurs. Cibles et Leviers: Analyse Factorielle des Relations de Controle Indirect dans une Finne

Americaine d'Avocats d'Affaires.' Bulletin de Methodologie Sociologique 37: 41-51. Lazerson, M. H. 1988 'Organizational Growth of Small Firms.' American Sociological Review 53: 330-342. 1995 'A New Phoenix? Modem Putting-Out In the Modena Knitwear Industry.' Administrative Scienc(!

Quarterly 40: 34-59. Lee, L., and D. Dobler 1965 Purchasing and Materials Management. New York: McGraw-Hili. Leenders, R. Th. A. J. 1991 'Prediction of Committees in Collective Decision Making Processes.' NWO Grant Proposal

(subsidized). Department of Sociology. University of Groningen. 1995a 'Models for Network Dynamics: A Markovian Framework.' Journal of Mathematical Sociology

20: 1-21. 1995b Structure and Influence. Amsterdam: Thesis Publishers. 1996 'Evolution of Friendship and Best Friendship Choices.' Journal of Mathematical Sociology 21:

133-148. 1997 'Longitudinal Behavior of Network Structure and Actor Attributes: Modeling Interdependence of

Contagion and Selection.' In P. Doreian and F.N. Stokman (eds.). Evolution of Social Networks. Langhorne: Gordon and Breach.

1999 'Modeling social influence through network autocorrelation: on constructing the weight matrix.' Social Networks Forthcoming.

Le Grand, C., R. Szulkin, and M. Tihlin 1994 'Organizational Structures and Job Rewards in Sweden: Acta Sociologica 37: 231-251. Leifer, E. M. 1988 'Interaction Preludes to Role Setting: Exploratory Local Action.' American Sociological Review

53: 865-878. 1990 'Market and Authority in League Sports: American Journal of Sociology 96: 655-683. Leonard-Barton, D. 1995 WeI/springs of Knowledge. Building and Sustaining the Sources of Innovation. Boston: Harvard

Business School Press. Leontlef, W. W. 1951 The Structure of the American Economy, 1919-1939 (2nd. edition). New York: Oxford University

Press. Lerner,J. 1995 'Patenting in the Shadow of Competitors: Journal of Law and Economics 38: 463-496. Levaclc, R. 1987 Economic Policy Making. Brighton: Wheatsheaf. Levl,M. 1996 'Social and Unsocial Capital-a Review-Essay of Putnam. Robert Making Democracy Work:

Politics and Society 24: 45-55. Levine, D. I. 1990 'Participation, Productivity, and the Firm's Environment.' California Management Review 32: 86-

100. Levine,J. H. 1972 'The Sphere of Influence.' American Sociological Review 37: 14-27. Levinson, D. J., C. N. Darrow, E. B. Klein, M. A. Levinson, and B. McKee 1978 Seasons ofa Man's Ufe. New York: Knopf. Uvi-Strauss, C. 1949 Les Formes Etementaires de la Parente. Paris: PIon. Lewis,J.D. 1990 Partnerships for Profit: Structuring and Managing Strategic Alliances. New York: Free Press.


Lewis, J. D., and A. Weigert 1985 'Trust as a Social Reality.' Social Forces 63: 967-985. Likert, R. 1961 The Human OrganiZiltion. New York: Wiley. Limerick, D., and B. Cunnington 1993 Managing the New OrganiZiltion: A Blueprint for Networks and Strategic Alliances. San

Francisco: Jossey-Bass. Lln,N. 1982 'Social Resources and Instrumental Action.' In P. V. Marsden and N. Un (eds.), Social Structure

and Network Analysis. Beverly Hills: Sage. 1990 'Social Resources and Social Mobility: A Structural Theory of Status Attainment.' In R. L. Breiger

(ed.), Social Mobility and Social Structure. New York: Cambridge University Press. 1995a 'Les Resources Sociales: une lMorie du Capital Social.' Revue Fran~aise de Sociologie 36: 685-

704. 1995b 'Institutions and Networks: The Infrastructure of Society. ' Paper presented to International Social

Network Conference, London. Lin, N., W. M. Ensel, and J. C. Vaughn 1981 'Social Resources and Strength of Ties: Structural Factors in Occupational Status Attainment.'

American Sociological Review 46: 393-405. Lin, N., and M. Durnil 1986 'Access to Occupations through Social Ties.' Social Networks 8: 365-385. Lincoln, J. R., Ahmadjian, and E. Mason 1998 'Organizational Learning and Purchase-Supply Relations in Japan: Hitachi, Matsushita and Toyota

Compared.' California Management Review (forthcoming). Lincoln, J . R., M. L. Gerlach, and P. Takahashi 1992 'Keiretsu Networks in the Japanese Economy: A Dyad Analysis of Intercorporate Ties.' American

Sociological Review 57: 561-85. Lindenberg, S. M. 1990 'Homo Socio-Economicus: the Emergence of a General Model of Man in the Social Sciences.'

Journal of Institutional and Theoretical &onomics 146: 727-748. 1993 'Club Hierarchy, Social Metering and Context Instruction: Governance Structures in Response to

Varying Self-Command Capital.' In S. M. Undenberg and H. Schreuder (eds.), Interdisciplinary Perspectives on OrganiZiltion Studies. Oxford: Pergamon Press.

1996 'Multiple-Tie Networks, Structural Dependence, and Path-Dependency: Another Look at Hybrid Forms of Governance.' Journal of Institutional and Theoretical Economics 152: 188-196.

Llpnack, J., and J. Stamps 1994 The Age of the Network. Essex Junction: Omneo. Lippman, S. A., and R. P. Romelt 1982 'Uncertain lmitability: An Analysis of Interfirm Differences in Efficiency under Competition.' Bell

Journal of Economics 13: 418-438. Long,J.s. 1997 Regression Models for Categorical and Limited Dependent Variables. Thousand Oaks: Sage. Lorange, P., and J. Roos 1992 Strategic Alliances: Fonnation, Implementation, and Evolution . Cambridge: Blackwell Business. Lorange, P., J. Roos, and P. G. Bronn 1992 'Building Successful Alliances.' Long Range Planning 25: 10-17. Lorenz, E. H. 1993 'Flexible Production Systems and the Social Construction of Trust.' Politics and Society, 21: 307-

324. Lorenzonl, G., and C. Baden-FuUer 1995 'Creating a Strategic Center to Manage a Web of Partners.' California Management Review 37:

146-163. Lorenzoni, G., and O. Ornati 1988 'Constellations of Firms and New Ventures.' Journal of Business Venturing 3: 41-57. Lorenzonl, G., and A. Llparinl 1997 'Leveraging Internal and External Competencies in Boundary Shifting Strategies. A Longitudinal

Approach.' Working Paper, Department of Management, Economics faculty, University of Bologna.

References - 523

Louls,M. R. 1990 'Acculturation in the Workplace: Newcomers as Lay Ethnographers.' In I. L. Goldstein (ed.),

Frontiers of Industrial and OrganilJltionai Psyclwlogy, 85-129. San Francisco: Jossey-Bass. Loury,G. 1977 'A Dynamic Theory of Racial Income Differences.' In P. A. Wallace and A. LeMund (eds.)

Women, Minorities, and Employment Discrimination. Lexington: Lexington Books. 1987 'Why Should We Care about Group Inequality?' Social Philosophy and Policy 5: 249-271. Low, M. B., and I. C. MacMillan 1988 'Entrepreneurship: Past Research and Future Challenges.' Journal of Management 14: 139-161. Luhmann,N. 1979 Trust and Power. Chichester: Wiley. Lusch,R. F. 1976 'Sources of Power: Their Impact on Intrachannel Conflict.' Journal of Marketing Research 13:

382-390. Lusch, R. F., and J. R. Brown 1982 'A Modified Model of Power in the Marketing Channel.' Journal of Marketing Research 19: 312-

323. Macaulay, S. 1963 Non-Contractual Relations in Business: A Preliminary Study. American Sociological Review 28:

55-66. Macbeth, D. K., and N. Ferguson 1994 Partnership Sourcing: An Integrated Supply Chain Approach. London: Pitman. MacKie-Mason, J. K. 1990 'Do Firms Care who Provides Their Financing?' In R. G. Hubbard (ed.), Asymmetric Information,

Corporate Finance, and Investment, 63-103. Chicago: University of Chicago Press. MacLachlan, A. 1995 'Trusting Outsiders to do Your Research: How does Industry Learn to do it?' Research Technology

Management 38: 48-53. Macneil, I. R. 1986 'Exchange Revisited, Individual Utility and Social Solidarity.' Ethics 96: 567-593. Macy, M. W., and A. Flache 1995 'Beyond Rationality in Models of Choice.' Annual Review of Sociology 21 : 73-91. Maddala, G. S. 1983 Limited-Dependent and Qualitative Variables in Econometrics. Cambridge: Cambridge University

Press. Malster, D. H. 1993 Managing the Professional Service Firm. New York: Free Press. Mansfield, E. 1984 'R&D and Innovation: Some Empirical Findings.' In Z. Griliches (ed.), R&D Patents and

Productivity. Chicago: University of Chicago Press. March,J.G. 1988 Decisions and OrganilJltions. New York: Blackwell. Markovsky, B., and M. V. Chaffee 1995 'Social Identification and Solidarity: A Revised Formulation.' In B. Markovsky, 1. O'Brien, and K.

Heimer (eds.), Advances in Group Processes 12,249-270. Greenwich: JAI Press. Markovsky, B., and E. J. Lawler 1994 'A New Theory of Group Solidarity.' In B. Markovsky, J. O'Brien, and K. Heimer (eds.), Advances

in Group Processes II, 113-137. Greenwich: JAI Press. Markowsky, B., D. Willer, and T. Patton 1988 'Power Relations in Exchange Networks.' American Sociological Review 53: 220-236. Marris, R. 1964 The &onomic Theory of Managerial Capitalism. London: MacMillan. Marsden, P. V. 1982 'Brokerage Behavior in Restricted Exchange Networks.' In P. V. Marsden and N. Lin (eds.), Social

Structure and Network Analysis. Beverly Hills: Sage. 1983 'Restricted Access in Networks and Models of Power.' American Journal of Sociology 88: 686-

717. 1987 'Core Discussion Networks of Americans.' American Sociological Review 52: 122-131.


1995 'The Staffing Process. Recruitment and Selection Methods.' In A. L. Kalleberg, D. Knoke, P. V. Marsden and J. L Spaeth, OrganiYltions in America. Analyzjng Their Structures and Human Resources Practices, 133-156. Thousand Oaks: Sage.

Marsden, P. V., and K. E. Campbell 1990 'Recruitment and Selection Processes: The Organizational Side of Job Searches.' In R. L. Breiger

(ed.), Social Mobility and Social Structure, 59-79. New York: Cambridge University Press. Marsden, P. V., and J. S. Hurlbert 1988 'Social Resources and Mobility Outcomes: A Replication and Extension.' Social Forces 66: 1038-

1059. Marsh, D., and R. A. W. Rhodes 1992 Policy Networks in British Government. Oxford: Clarendon Press. Marshal,A. 1890 Principles of &onomics (First Edition). London: Oxford University Press. 1923 Industry and Trade . London: Macrnillam. Martinelli, A. 1994 'Entrepreneurship and Management.' In N. Smelser and R. Swedberg (eds.), The Handbook of

Economic Sociology, 476-503. Princeton: Princeton University Press. Mathworks, Inc., The 1994 MATLAB version 4.2c. Natick: The Mathworks, Inc. Mayer, R. C., J. H. Davis, and F. D. Schoorman 1995 'An Integrative Model of Organizational Trust.' Academy of Management Review 20: 709-734. McAllister, D. J. 1995 'Affect· and Cognitive-Based Trust as Foundations for Interpersonal Cooperation in

Organizations.' Academy of Management Journal 38: 24-59. McClelland, D. C. 1967 The Achieving Society. New York: Free Press. McKenna, C. J. 1985 Uncertainty and the Labour Market: Recent Developments in the Job-Search Theory. Brighton:

Wheatsheaf Books. McPherson, J. M., P. A. Popielarz, and S. Drobnic 1992 'Social Networks and Organizational Dynamics.' American Sociological Review 57: 153-170. Mehra, A., M. Kilduff, and D. J. Brass 1998 'At the Margins: A Distinctiveness Approach to the Social Identity and Social Networks of

Underrepresented Groups.' Academy of Management Journal 41: 441-452. Menon, G., B. Bickart, S. Sudman, and J. Blair 1995 'How Well Do You Know Your Partner? Strategies for Formulating Proxy-Reports and Their

Effects on Convergence to Self-Reports.' Journal of Marketing Research 32: 75-84. Mergers and Acquisitions 1986-1993 The Journal of Corporate Venture. Washington: Mergers and Acquisitions. Merges, R. P., and R. R. Nelson 1990 'On the Complex Economics of Patent Scope.' Columbia Law Review 90. Merton, R. K. 1964 'The Mathew Effect in Science.' Science 159: 56-63. Meurer, M. J. 1989 'The Settlement of Patent litigation.' RAND Journal of Economics 20: 77-91. Meyer, J. W., and B. Rowan 1991 'Institutionalized Organizations: Formal Structure as Myth and Ceremony.' In W. W. Powell and P.

J. DiMaggio (eds.), The New Institutionalism in OrganiYltional Analysis, 41-62. Chicago: University of Chicago Press.

Meyer,M. W. 1994 'Measuring Performance in Economic Organizations.' In N. Smelser and R. Swedberg (eds.),

Handbook of &onomic Sociology. Princeton: Russell Sage. Meyerson, E. M. 1992 The Impact of Ownership Structure and Executive Team Composition on Firm Peiformance.

Stockholm: Almqvist & Wickselllnternational. 1994 'Human Capital, Social Capital and Compensation: the Relative Contribution of Social Contacts to

Managers' Incomes.' Acta Socioiogica 37: 383-399.

References - 525

Miles, R. E., and C. C. Snow 1986 'Organizations: New Concepts for New Fonns.' California Managment Review 28: 62-73. 1992 'Causes of Failures in Network Organizations.' California Management Review 34: 53-72. 1995 'The New Network Finn: A Spherical Structure Built on a Human Investment Philosophy.'

Organizational Dynamics 23: 5-18. Milgrom, P., and J. Roberts 1992 Economics, Organization and Management. Englewood Cliffs: Prentice Hall. 1996 'Bargaining Costs, Influence Costs. and the Organization.' In L. Puttennan and R. S. Kroszner

(eds.). The Economic Nature of the Firm: A Reader (2nd. edition). Chicago: University of Chicago Press.

Miller, J. A. 1976 Exploring Some Alternative Measure of Consumer Satisfaction in Marketing: 1776-1976 and

Beyond. Chicago. Miller, D. 1991 'Stale in the Saddle: CEO Tenure and the Match between Organization and Environment.'

Management Science 37: 34-52. Miller, G. J. 1992 Managerial Dilemmas: The Political Economy of Hierarchy. Cambridge: Cambridge University

Press. MilIer,P. 1994 'Accounting and Objectivity: The Invention of Calculating Selves and Calculable Spaces.' In A.

Megill (ed.). Rethinking Objectivity. Durham: Duke University Press. Millson, M. R., S. P. Raj, and D. Wilemon 1996 'Strategic Partnering for Developing New Products.' Research Technology Management 39: 41-

49. Miner, A. S., and S. E. Estler 1985 'Accrual Mobility: Job Mobility in Higher Education through Responsibility Accrual.' Journal of

Higher Education 56: 121-143. Mintz, B., and M. Scbwartz 1985 The Power Structure of American Business. Chicago: University of Chicago Press. Mintzberg, H. 1973 The Nature of Managerial Work. New York: Harper and Row. 1983 Structure in Fives: Designing Effective Organizatians. Englewood Cliffs: Prentice-Hall. Misblra,A. 1996 'Organizational Response to Crisis: The Centrality of Trust.' I. R. Kramer and T. Tylor (eds.).

Trust in Organizations: Frontiers of Theory and Research. 261-287. London: Sage. Misurni,K. 1989 'Communication Structure. Trust, and the Free Rider Problem.' Journal of Mathematical Sociology

14: 273-282. Mitsui, I 1990 'EC-Japan Subcontracting Opportunities.' Proc. of Palais des Congres Conference. Jun 21-22. EC

DGXXIII. Mizrucbl, M. S. 1992 The Structure of Corporate Political Action. Cambridge: Harvard University Press. 1994 'Social Network Analysis: Recent Achievements and Current Controversies.' Acta Sociologica 37:

329-343. Mizrucbi, M. S., and L. B. Stearns 1994 'A Longitudinal Study of Borrowing by Large American Corporations.' Administrative Science

Quarterly 39: 118-140. Mizrucbi, M. S., P. Mariolis, M. Scbwartz, and B. Mintz 1986 'Techniques for Disaggregating Centrality Scores in Social Networks.' In N. B. Tuma (ed.).

Sociological Methodology 1986.26-48. Washington: American Sociological Association. Moerbeek, H., H. Flap, and W. U1tee 1997 'How Dutch Men and Women find a Job and What it Does to Their Career.' Paper presented at the

conference on Social Stratification in Modern Welfare States. Research Committee 28. Stockholm. May 29th-June 2nd.


Monge, P. R., and E. M. Eisenberg 1987 'Emergent Communication Networks.' In F. M. Jablin. L. L. Putman. K. H. Roberts. and L. W.

Porter (eds.). Handbook of Organiwtional Communication: An Interdisciplinary Perspective. 304-342. Newbury Park: Sage.

Monge, P.R., M. Cozzens, and N. Contractor 1991 'Communication and Motivational Predictors of the Dynamics of Organizational Innovation.'

Organiwtion Science 2: 1-25. Moore,G. 1992 'Gender and lnfonnal Networks in State Government.' Social Science Quanerly 73: 46-61 . Moran, P., and S. Ghoshal 1996 'Value Creation by Firms.' Academy of Management Best Papers Proceedings: 41-45. Morek, R., A. Shleifer, and R. W. Vishny 1990 '00 Managerial Objectives drive bad Acquisitions?' Journal of Finance 45: 31-48. Morgan, I. 1987 The Purchasing Revolution.' McKinsey Quanerly Spring: 49-55. Morgan, G., and S. Maddock 1997 Markets, Managers and the NHS: An Institutionalist Approach. University of Manchester:

Manchester Business School. Mowery, D. C. (ed.) 1988 International Collaborative Ventures in US ManUfacturing. Cambridge: Ballinger. Mueller, R. K. 1986 Corporate Networking. New York: Free Press. Nahapiet, J ., and S. Ghoshal 1998 'Social Capital. Intellectual Capital, and the Organizational Advantage.' Academy of Management

Review 23: 242-266. Nakatani, I. 1984 'The Economic Role of Financial Corporate Grouping.' In M. Aoki (ed.). The Economic Analysis

of Japanese Firm. 227-258. New York: North-Holland. Neckennan, K. M., and R. M. Fernandez 1997 'Keeping a Job: Network Hiring and Turnover in a Retail Bank.' Paper presented at the American

Sociological Association. Toronto. Ontario. Aug. 13. 1997. Nelson, R. E. 1989 The Strength of Strong Ties: Social Networks and Intergroup Conflict in Organizations.' Academy

of Management Journal 32: 377-401. Nelson, R. R., and S. G. Winter 1982 An Evolutionary Theory of Economic Change. Cambridge: Harvard University Press. Nelson, R. L . 1988 Partners With Power: The Social Transformation of the Large Law Firm. Berkeley: University of

California Press. Neu,D. 1991 Trust, Contracting and the Prospectus Process.' Accounting, Organiwtions and Society 16: 243-

256. Newcomb,T. 1961 The Acquaintance Process. New York: Holt, Rinehart, and Winston. Newton,K. 1997 'Social Capital and Democracy.' American Behavioral Scientist 40: 575-586. Nicholls, J. G. 1978 The Development of the Concepts of Effort and Ability. Perception of Academic Attainment, and

the Understanding that Difficult Tasks Require More Ability.' Child Development 49: 800-814. Nicholson, N. 1993 'Purgatory or Place of Safety? The Managerial Plateau and Organizational Agegrading.' Human

Relations 46: 1369-1389. Niemiiller, B., W. H. Van Schuur, and F. N. Stokman 1980 'Stochastic Cumulative Scaling: Mollen Scale. Mollen Test.' STAP User's Manual, Volume 4.

Technisch Centrum FSW. University of Amsterdam. Nishiguchi, T. 1994 Strategic Industrial Sourcing: The Japanese Advantage. New York: Oxford University Press.

References - 527

Nishiyama, T. 1982 'The Structure of Managerial Control: Who Owns and Controls Japanese Businesses.' Japanese

Economic Studies Fall: 37-77. Nobeoka, K., and J. H. Dyer 1998 The Influence of Customer Scope on Supplier Learning and Peiformance in the Japanese

Automobile Industry. Working Paper, The Wharton School, University of Pennsylvania, Philadelphia.

Noe, R.A. 1988a 'An Investigation of the Determinants of Successful Assigned Mentoring Relationships.' Personnel

Psychology 41: 457-479. 1988b 'Women and Mentoring: A Review and Research Agenda.' Academy of Management Review 13:

65-78. Nohria,N. 1992a 'Is a Network Perspective a Useful Way of Studying Organizations?' In N. Nohria and R. G. Eccles

(eds.), Networks and Organizations: Structure, Form, and Action. Boston: Harvard Business School Press.

1992b 'Information and Search in the Creation of New Business Ventures: The Case of the 128 Venture Group.' In N. Nohria and R. G. Eccles (eds.), Networks and Organizations: Structure, Form, and Action, 240-261 . Boston: Harvard Business School Press.

Nohria, N., and R. G. Eccles 1993 'Face-to-Face: Making Network Organizations Work.' In N. Nohria and R. G. Eccles (eds.),

Networks and Organizations: Structure, Form, and Action, 288-308. Boston: Harvard Business School Press.

Nohria, N., and R. G. Eccles (eds.) 1992 Networks and Organizations: Structure, Form, and Action. Boston: Harvard Business School

Press. Nohria, N., and S. Ghoshal 1997 The Differentiated Network: A New Model for Organizing Multinational Corporations. San

Francisco: Jossey Bass. Nohrla, N., and J. D. Berkley 1995 'From Structure to Structuring: A Pragmatic Perspective on Organizational Design.' Harvard

Business School Working Paper #96-053. Nonaka, I., and H. Takeuchi 1995 The Knowledge Creating Company. Oxford: Oxford University Press. Nooteboom, B. 1992 'Towards a Dynamic Theory of Transactions.' Journal of Evolutionary Economics 2: 281-299. 1996 'Trust, Opportunism and Governance: A Process and Control Model.' Organization Studies 17:

985-1010. 1997 'Voice and Exit Based Forms of Corporate Control: Anglo-American, European and Japanese.' Paper

for the EAEPE conference, Athens, 6-9 November. 1998a 'Cost, Quality and Learning Based Governance of Transactions.' In M. Colombo (ed.), The Changing

Boundaries of the Finn. London: Routledge. 1998b 'Trust as a Governance Device: Theory and Empirical Evidence.' Paper for a symposium on cultural

factors in economic growth. Marienrode, Germany, 3-5 April. 1999 'The Dynamic Efficiency of Networks.' In A. Grandori (ed.), Inter·Firm Networks: Organizational

and Industrial Competitiveness. London: Routledge (forthcoming). Nooteboom, B., J. Berger, and N. G. Noorderhaven 1997 'Effects of Trust and Governance on Relational Risk.' Academy of Management Journal 40: 308-338. Norburn, D., and S. Birley 1988 'The Top Management Team and Corporate Performance.' Strategic Management Journal 9: 225-

237. North,D.C. 1990 Institutions, Institutional Change and Economic Peiformance. Cambridge: Cambridge University

Press. North, D. C., and R. Thomas 1973 The Rise of the New World: A New Economic History. Cambridge: Cambridge University Press. Northrup, H. R., R. M. Cowin, L. G. Vanden Plas, and W. E. Fulmer


1978 The Objective Selection of Supervisors. Philadelphia: Industrial Research Unit, The Wharton School, University of Pennsylvania.

Obstfeld, D. 1997 'Different Fonns of Social Capital: Sorting Out Two Alternative Conceptions of Social

Entrepreneurship.' Paper presented at the Interdisciplinary Consortium for Organizotional Studies (ICOS). University of Michigan.

Office of Technology Assessment and Forecast, U.s. Department of Conunerce, Patent, and Trademark Office 1976 Technology Assessment and Forecast. U.S. GPO. Washington. Obmae,K. 1989 The Global Logic of Strategic Alliances.' Harvard Business Review 67. Oliver, R. K., and M. D. Webber 1982 Outlook. Booz Allen and Hamilton Inc Olson,M. 1965 The Logic of Collective Action. Cambridge: Harvard University Press. Ornta,S. W.F. 1995 Critical Success Factors in Biomedical Research and Pharmaceutical Innavation. Dordrecht:

Kluwer Academic Publishers. O'Neil, R., and R. E. Quinn 1983 'Applications of the Competing Values Frameworlc:.' Human Resource Management 32: 1-7. Opotow,S. 1990 'Moral Exclusion and Injustice: An Introduction.' Journal of Social Issues 46: 1-20. O'ReDly,B. 1994 The New Deal: What Companies and Employees Owe One Another.' Fortune 13: 44-52. Osborn, R. N., and C. C. Baughn 1990 'Fonns of Interorganizational Governance for Multinational Alliances.' Academy of Management

Journal 33: 503-519. Ostrom,E. 1994 'Constituting Social Capital and Collective Action.' Journal of Theoretical Politics 6: 527-562. Packer, K., and A. Webster 1996 'Patenting Culture in Science: Reinventing the Wheel of Scientific Credibility.' Science,

Technology, and Human Values 21: 427-53. Pakes, A., and M. Schankerman 1984 The Rate of Obsolescence of Patents, Research Gestation Lags, and the Private Rate of Return to

Research Resources.' In Z. Griliches (ed.), R&D, Patents, and Productivity. Chicago: University of Chicago Press.

Pakes, A., and Z. Griliches 1980 'Patents and R&D at the Firm Level: A First Look.' Economics Letters 5: 377-81. Palmer, D. A., P. D. Jennings, and X. Zhou 1993 'Late Adoption of the Multidivisional Form by Large U.S. Corporations: Institutional, Political,

and Economic Accounts.' Administrative Science Quarterly 38: 1~131. Papa,M.J. 1990 'Communication Networlc: Patterns and Employee Performance with a New Technology.'

Communication Research 17: 344-368. Parasuraman, A., V. A. Zeithaml, and L. L. Berry 1985 'A Conceptual Model of Service Quality and its Implications for Future Research.' Journal of

Marketing 49: 41-50. Parcel, T. L., R. L. Kaufman, and L. Jolly 1991 'Going Up the Ladder: Multiplicity Sampling to Create Unked Macro-to-Micro Organizational

Samples.' In Peter V. Marsden (ed.), Sociological Methodology 1991, 43-79. Oxford: Basil Blackwell.

Parker, R. E. 1994 Flesh Peddlers and Warm Bodies: The Temporary Help Industry and Its Workers. New

Brunswick: Rutgers University Press. Parsons, T. 1951 The Social System. Glencoe: Free Press. 1963 'On the Concept of Influence.' Public Opinion Quarterly 27: 37-62.

Pennings, J. M. 1980 Interlocking Directorates. San Francisco: Jossey Bass. Pennings, J. M., and F. Barianto

References - 529

1992 The Diffusion of Technological Innovation in the Commercial Banking Industry.' Strategic Management Journal 13: 29-46.

Pennings, J. M., K. Lee, and A. Van Witteloostuyn 1998 'Human and Capital, Social Capital, and Firm Dissolution.' Academy of Management Journal 41:

425-440. Penrose,E. 1959 The Theory of the Growth of the Firm. Oxford: Basil Blackwell. Perrow,C. 1992 'Small Firm Networks.' In N. Nohria and R. G. Eccles (eds.), Networks and Organiwtions:

Structure, Form, and Action, 445-470. Cambridge: Harvard Business School Press. Petersen, M. A., and R. G. Rajan 1994 The Benefits of Lending Relationships: Evidence from SmaIl Business Data.' Journal of Finance

49: 3-37. 1995 'The Effect of Credit Market Competition on Lending Relationships.' Quarterly Journal of

Economics 110: 407-443. Pettigrew, A. 1974 The Politics of Organiwtion Decision Making. London: Tavistock Pettigrew, A. M., E. B. Ferlle, and L. McKee 1992 Shaping Strategic Change. London: Sage. Pfeffer,J. 1977 'Toward an Examination of Stratification in Organizations.' Administrative Science Quarterly 22:

553-567. 1981 Power in Organiwtions. Cambridge: Ballinger. 1983 Organizational Demography.' In L. L. Cummings, and B. Staw (eds.), Research in Organiwtional

Behavior 5: 299-357. Greenwich: JAI Press. 1987 'A Resource Dependence Perspective on Intercorporate Relations.' In M. Mizruchi and M.

Schwartz (eds.), Intercorporate Relations: The Structural Analysis of Business, 25-55. New York: Cambridge University Press.

1989 'A Political Perspective on Careers: Interests, Networks. and Environments.' In M. B. Arthur, D. T. Hall, and B. S. Lawrence (eds.), Handbook of Career Theory, 380-396. New York: Cambridge University Press.

1992 Managing with Power: Politics and Influence in Organiwtions. Boston, MA: Harvard Business School Press.

Pfeffer, J., and Y. Cohen 1984 'Detenninants of Internal Labor Markets in Organizations.' Administrative Science Quarterly 29:

550-572. Pfeffer, J., and P. Nowak 1976 'Joint Ventures and Interorganizational Interdependence.' Administrative Science Quarterly 21:

398-418. Pfeffer, J., and G. R. Salancik 1978 The External Control ofOrganiwtions: A Resource Dependence Perspective. New York: Harper

and Row. Pfeffer, J., and J. Ross 1982 The Effects of Marriage and a Working Wife on Occupational and Wage Attainment.' Administrative

Science Quarterly 27: 66-80. Phillips-Jones, L. 1982 Mentors and Proteges. New York: Arbor House. Piaget,J. 1932 The Moral Judgement of the Child. London: Kegan, Paul, Trench, and Teubner. Pl1des, R. B. 1996 'The Destruction of Social Capital Through Law.' University of Pennsylvania lAw Review 144:

2055-2077. Pinchot,G. 1985 Intrapreneuring: Why You Don't Have to Leave The Corporation to Become an Entrepreneur.

New York: Harper and Row.


Pinfield, L. T. 1995 The Operation of Internal Labor Markets: Staffing Practices and Vacancy Chains. New York:

Plenum Press. Platteau, J. P. 1994a 'Behind the Market Stage Where Real Societies Exist I : The Role of Public and Private Order

Institutions.' Journal of Development Studies 30: 533-577. 1994b 'Behind the Market Stage Where Real Societies Exist 2: The Role of Moral Norms.' Journal of

Development Studies 30: 753-817. Podolny, J. M. 1993 'A Status-Based Model of Market Competition.' American Journal of Sociology 98: 829- 872. 1994 'Market Uncertainty and the Social Character of Economic Exchange.' Administrative Science

Quarterly 39: 458-483. Podolny, J. M., and J. N. Baron 1997 'Resources and Relationships: Social Networks and Mobility in the Workplace.' American

Sociological Review 62: 673-{i93. Podolny, J. M., and A. Feldman 1997 'Is it Better to Have Status or Know What You Are Doing: Position and Capability in Venture

Capital Markets.' Graduate School of Business, Stanford University: Working Paper. Podolny, J. M., and K. L. Page 1998 'Network Forms of Organization.' Annual Review of Sociology (forthcoming). Podolny, J. M., and D. J. Phillips 1996 The Dynamics of Organizational Status.' Industrial and Corporate Change 5: 453-472. Podolny, J. M., and T. E. Stuart 1995 'A Role-Based Ecology of Technological Change.' American Journal of Sociology 100: 1224-

1260. Podolny, J. M., T. E. Stuart, and M. T. Hannan 1996 'Networks, Knowledge, and Niches: Competition in the Worldwide Semiconductor Industry,

1984-1991.' American Journal of Sociology 102: 659-{j89. Podolny, J. M., and J. Baron 1997 'Resources and Relationships: Social Networks and Mobility in the Workplace.' American

Sociological Review 62: 673-{i93. Polanyi,K. 1957 The Great Transformation. Boston: Beacon Press. 1958 Personal Knowledge, London, Routledge & Kegan Paul. Pollitt, C. 1990 The New Managerialism and the Public Services: The Anglo American Experience . Oxford: Basil

Blackwell. Porter, M. E. 1980 Competitive Strategy: Techniques for Analyzing Industries and Competitors. New York: Free

Press. 1985 Competitive Advantage: Creating and Sustaining Superior Performance. New York: Free Press. 1987 'Managing Value-From Competitive Advantage to Corporate Strategy.' Harvard Business

Review 65: 43-58. 1991 'Toward a Dynamic Theory of Strategy.' Strategic Management Journal 12: 95-117. Porter, M. E., and K. Fuller 1986 'Coalitions and Corporate Strategy.' In M. Porter (ed.), Competition in Global Industries. Boston:

Harvard Business School Press. Portes, A., and J. Sensenbrenner 1993 'Embeddedness and Immigration: Notes on the Social Determinants of Economic Action.'

American Journal of Sociology 98: 1320-1350. Posner, E. A. 1996 The Regulation of Groups-The Influence of Legal and Nonlegal Sanctions on Collective Action.'

University of Chicago Law Review 63: 133-197. Powell, W. W. 1982 Getting into Print. The Decision-Maldng Process in Scholarly Publishing. Chicago: University of

Chicago Press. 1987 'Hybrid Organizational Arrangements: New Form or Transitional Development?' California

Management Review 30: 67-87.

References - 531

1990 'Neither Market nor Hierarchy: Network Fonns of Organization.' In L. L. Cummings and B. M. Slaw (eds.), Research in Organiwtional Behavior 12: 295-336. Greenwich: JAI Press.

Powell, W. W., and J. Owen-Smith 1998 'Commercialism in Universities: life Sciences Research and its Linkage with Industry.' Journal of

Policy Analysis and Management 17. Powell, W. W., and P. Brantley 1992 'Competitive Cooperation in Biotechnology: Learning Through Networks?' In N. Nohria and R. G.

Eccles (eds.) Networks and Organiwtions: Structure, Form, and Action, 366-394. Boston: Harvard Business School Press.

Powell, W. W., and L. Smith-Doerr 1994 'Networks and Economic life.' In N. Smelser and R. Swedberg (eds.), The Handbook of Economic

Sociology, 368-402. Princeton: Princeton University Press. Powell, W. W., K. W. Koput, and L. Smith-Doerr 1996 'Interorganizational Collaboration and the Locus of Innovation: Networks of Learning in

Biotechnology.' Administrative Science Quarterly 41: 116-45. Prahalad, C. K., and G. Hamel 1990 "The Core Competences of the Corporation.' Harvard Business Review 68: 79-90. Prelsendijrfer, P., and T. Voss 1988 'Arbeitsmarkt und Soziale Netzwerke: Die Bedeutung Sozialer Konta1cte beim Zugang zu Arbeitsplat

zen.' Sowle Welt 39: 104-119. Price, L. L., E. J. Arnould, and P. Tierney 1995 'Ooing to Extremes: Managing Service Encounters and Assessing Provider Performance.' Journal

of Marketing 59: 83-97. Provan, K. G., and H. B. Milward 1995 'A Preliminary Theory of Interorganizational Network Effectiveness: A Comparative Study of

Four Community Mental Health Systems.' Administrative Science Quarterly 40: 1-33. Pruitt, D. G., and M. J. Klnunel 1977 'Twenty Years of Experimental Gaming: Critique, Synthesis, and Suggestions for the Future.'

Annual Review of P:rychology 28: 363-392. Putnam, R. D. 1993a Making Democracy Work: Civic Traditions in Modem Italy. Princeton: Princeton University

Press. 1993b 'The Prosperous Community: Social Capital and Public life.' American Prospect 13: 35-42. 1995a 'Bowling Alone: America's Declining Social Capital.' Journal of DefTWcracy 6: 65-78. 1995b 'Bowling Alone, Revisited.' The Responsive Community Spring: 18-33. Puto, C., W. Patton, and R. King 1985 'Risk Handling Strategies in Industrial Vendor Selection Decisions.' Journal of Marketing 49: 89-

98. Rablnow,P. 1996 Making PCR: A Story of Biotechnology. Chicago: University of Chicago Press. Raftery, A. E. 1986 'Choosing Models for Cross-Classifications.' American Sociological Review 51 : 145-46. 1995 'Bayesian Model Selection in Social Research.' In P. V. Marsden (ed.), Sociological Methodology

1995,111-163. Washington: American Sociological Association. Raglus, B. R., and D. B. McFarUn 1990 'Perceptions of Mentor Roles in Cross-Gender Mentoring Relationships.' Journal of Vocational

Behavior 37: 321-339. Raglus, B. R., and T. A. Scandura 1994 'Gender Differences in Expected Outcomes of Mentoring Relationships.' Academy of Management

Journal 37: 957-971. Ral, A., S. Borah, and A. Ramaprasad 1996 'Critical Success Factors for Strategic Alliances in the Information Technology Industry: An

Empirical Study.' Decision Sciences 27: 141-155. Ranson, S., and J. Stewart 1994 Managementfor the Public Domain: Enabling the Learning Society. London: Macmillan. Rapoport, A., and M. Guyer 1966 'A Taxonomy of 2x2 Games.' In General Systems: Yearbook of the Society for General Systems

Research 11: 203-214.


Raub, W., and J. Weesie 1990 'Reputation and Efficiency in Social Interactions: An Example of Network Effects.' American

Jou17IlJ1 of Sociology 96: 626-654. Ravenscraft, D. J. 1987 'The 19805 Merger Wave: An Industrial Organization Perspective.' In L. E. Browne and E. S.

Rosengren (eds.), The Merger Boom, 17-37. Boston: Federal Reserve Bank. Rees,A. 1966 'Information Networks in Labor Markets.' American Economic Review 56: 559-566. Reichheld, F. F., and W. E. Sasser, Jr. 1990 'Zero Defections: Quality Comes to Services.' Harvard Business Review 68: 105-111. Requena,F. 1991 'Social Resources and Occupational Status Attainment in Spain: a Cross-National Comparison with

the United States and the Netherlands.' International Jou17IlJ1 of Comparative Sociology 32: 233-242. Reskin, B. F., and H. I. Hartmann 1986 Women's Work, Men's Work: Sex Segregation on the Job. Washington: National Academy Press. Reve, T. 1990 'The Firm as a Nexus of Internal and External Contracts.' In M. Aoki, B. Gustafson, and O.

Williamson (eds.), The Firm as a Nexus of Treaties, 133-161. London: Sage. Rhodes, R. A. W. 1997 From Marketisation to Diplomacy: It's the Mix That Matters. University of Newcastle:

Department of Politics. RIchardson, G. B. 1972 'The Organisation of Industry.' Economic Jou17IlJ182 RIley, S., and D. Wrench 1985 'Mentoring Among Female Lawyers.' Jou17IlJ1 of Applied Social Psychology 15: 374--386. RIng, P. S., and A. H. Van de Ven 1992 'Structuring Cooperative Relationships Between Organisations.' Strategic Manogement JouT7UJl

13: 483-498. 1994 'Developmental Processes of Cooperative Interorganizational Relationships.' Academy of

Manogement Jou17IlJ119: 90-118. Roberts, K. H., and C. A. O'Reilly, m 1979 'Some Correlates of Communication Roles in Organizations.' Academy of Manogement Jou17IlJ1

22:42-57. Roche,G. 1979 'Much Ado About Mentors.' Harvard Business Review 57: 14--28. Rogers, E. M. 1971 Communication of Innovations. New York: Free Press. Romo, F. P., and M. Schwartz 1995 The Structural Embeddedness of Business Decisions: The Migration of Manufacturing Plants in

New York State, 1960-1985.' American SOCiological Review 60: 847-907. Roncek, D. W. 1992 'Learning More from Tobit Coefficients: Extending a Comparative Analysis of Political Protest.'

American Sociological Review 57: 503-507. Rosenbaum, J. 1979a 'Organizational Career Mobility: Promotion Chances in a Corporation during Periods of Growth

and Contraction.' American Jou17IlJ1 of Sociology 85: 21-48. 1979b 'Tournament Mobility: Career Patterns in a Corporation.' Administrative Science Quarterly 24:

22~241.

Rosenburg, L. J., and L. W. Stem 1971 'Conflict Measurement in the Distribution Channel.' Jou17IlJ1 of Marketing Research 8: 437-442. Roussel, P. A., K. A. Saad, and T. J. Erickson 1991 Third Generation R&D. Boston: Harvard Business School Press. Rowley, S., and Rowley, R. 1916 and 1960 Rowley on Partnership. New York: Bobbs-Merrill. Russell, J. A. 1979 'Affective Space is Bipolar.' Journal of Personality and Social Psychology 37: 345-356.

References - 533

Saaty, T.L. 1990 The Analytic Hierarchy Process. Planning. Priority Setting. Resource Allocation (2nd. edition).

Pittsburgh: RWS Publications. 1990 Multicriteria Decision Making. Pittsburgh: RWS Publications. Sabel,C.F. 1995 'Design, Deliberation and Democracy: On the New Pragmatism of Firms and Public Institutions.'

Paper presented at the liberal Institutions. Economic Constitutional Rights and the Role of Organiwtions. European University Institute, Florence.

Sabel, C. F., G. Herrigel, R. Kazis, and R. Deeg 1987 'How to Keep Mature Industries Innovative.' Technology Review 90: 26-35. Sabel, C. F., and J. Zeitlin 1997 'Stories. Strategies, Structures: Rethinking Historical Alternatives to Mass Production.' In C. F.

Sabel and J. Zeitlin (eds.), World of Possibilities. Flexibility and Mass Production in Western Industrialisation, 1-33. Cambridge: Cambridge University Press.

Sailer, L. D. 1978 'Structural Equivalence: Meaning and Definition, Computation and Application.' Social Networks

1: 73-90. Sako,M. 1991 Price. Quality and Trust: lnteljinn Relations in Britain and Japan. Cambridge: Cambridge

University Press. Salancik, G. R. 1986 'An Index of Subgroup Influence in Dependency Networks.' Administrative Science Quarterly 31 :

194-211. 1995 'Wanted: A Good Network Theory of Organization.' Administrative Science Quarterly 40: 345-

349. Samuel, Y. 1998 'Sociotechnical Realigning to Environmental Change: Defense Industries in Israel.' Unpublished

manuscript, University of Haifa, Israel. Sandefur, R. L., and E. O. Laumann 1997 'The Changing Value and Character of Social Capital in an Expanding Social System: A

Preliminary Test.' American Bar Foundation Working Paper #9606. Chicago: American Bar Foundation.

1998a 'Organization and Inequality: Changing Patterns of Income Inequality in the Chicago Bar, 1975 and 1995.' Paper prepared for presentation at the Annual Meetings of the American Sociological Association. San Francisco, August.

1998b 'A Paradigm for Social Capital.' Rationality and Society (forthcoming). Sasser, W. E., R. P. Olsen, and D. D. Wyckoff 1978 Management of Service Operations: Text, Cases and Readings. Boston: Allyn and Bacon. Saxenian, A. L. 1990 'Regional Networks and the Resurgence of Silicon Valley.' California Management Review 33:

89-112. 1991 'Response to Richard Florida and Martin Kenney 'Silicon Valley and Route 128 Won't Save Us."

California Management Review 33: 136-142. 1994 Regional Advantage-Culture and Competition in Silicon Valley and Route 128. Cambridge:

Harvard University Press. Scandura, T. A. 1992 'Mentorship and Career Mobility: An Empirical Investigation.' Journal of Organiwtional Behavior

13: 169-174. Scandura, T. A., and B. R. Ragins 1993 'The Effects of Sex and Gender Role Orientation on Mentorship in Male-Dominated Occupations.'

Journal of Vocational Behavior 43: 251-265. Scandura, T. A., and C. A. Schriesheim 1994 'Leader-Member Exchange and Supervisor Career Mentoring as Complementary Constructs in

Leadership Research.' Academy of Management Journal 37: 1588-1602. Scharpf, F. W. 1993 'Coordination in Hierarchies and Networks.' In F. W. Scharpf (ed.). Games in Hierarchies and

Networks: Analytical and Empirical Approaches to the Study of Governance Institutions, 125-165. Frankfurt: Campus Verlag !Westview Press.


Schecter, M. D. 1995 'How to Read a Patent.' In M. A. Lechter, E.C. Clifford, and R.B. Famiglio (eds.), Successful

Patenting for Scientists and Engineers. New Yorlc ffiEE Press. Schein, E. 1968 'Organizational Socialization and the Profession of Management.' In B. Staw (ed.), Psychological

Foundations ofOrganiwtional Behavior, 210--233. Santa Monica: Goodyear Publishing. Scher,M.J. 1996 'The Relational Access Paradigm and Japanese Interfinn Netwooo; Why Outsiders Seldom Wm:

Outcomes and Policy Implications.' Paper presented at the EMOT workshop (European Management and Organization in Transition, of the European Science Foundation) Turin, November.

Schere,J. 1982 'Tolerance of Ambiguity as a Discriminating Variable Between Entrepreneurs and Managers.'

Proceedings 404--408. Academy of Management, New Yon. Schermerhorn, J. R. 1977 'Information Sharing as an Interorganizational Activity.' Academy of Management Journal 20:

148-153. Schleifer, A., and R. W. Vlshny 1990 The Takeover Wave of the 1980s.' Science 17: 745--749. Schlesinger, L. A., and J. L. Heskett 1991 'Breaking the Cycle of Failure in Services.' Sloan Management Review Spring: 17-28. Schmidt, V. A. 1996 From State to Market?: The Transformation of French Business and Government. New Yon:

Cambridge University Press. Schmookler, J. 1966 Invention and Economic Growth. Cambridge: Harvard University Press. Schonberger, R. J. 1986 World Class Manufacturing: The Lessons of Simplicity Applied. New Yon: Free Press. Schumpeter, J. A. 1934 The Theory of Economic Development. Cambridge: Harvard University Press. Scott, W. R. 1992 Organiwtions: Rational, Natural. and Open Systems. Englewood Cliffs: Prentice Hall. 1995 Institutions and Organiwtions. Thousand Oaks: Sage. Scott, W. R., and J. W. Meyer 1991 The Organization of Sectors: Propositions and Early Evidence.' In W. W. Powell and P. J.

DiMaggio (eds.), The New Institutionalism in Organiwtional Analysis, 108-140. Chicago: University of Chicago Press.

Seabright, M. A., D. A. Levinthal, and M. Flchman 1992 'Role of Individual Attachments in the Dissolution of Interorganizational Relationships.' Academy

of Management Journal 35: 122-160. Seron,C. 1996 The Business of Practicing Law: The Work lives of Solo and Small Firm Attorneys. Philadelphia:

Temple University Press. Sexton, D. L., and N. Bowman 1985 The Entrepreneur: A Capable Executive and More.' Journal of Business Venturing 1: 129-140. Shah,P. 1996 'Network Destruction: The Structural Implications of Downsizing.' Paper presented at Academy of

Management Meetings, Cincinnati. Shapiro, D. L., B. H. Sheppard, and L. Cheraskln 1992 'In Theory: Business on a Handshake.' Negotiation Journal 8: 365--377. Shapiro,E. 1997 'Judge Rules Against Viacom in Dispute with Seagram Over Cable-TV Conflict.' Wall Street

Journal May 16: B6. Shaw,M.E. 1964 'Communication Netwooo'. In L. Benowitz (ed.), Advances in experimental social psychology 1:

111-147. New York: Academic Press. Sheppard, B. H., and M. Tuchlnsky 1996 'Micro-OB and the Netwon Organization.' In R. M. Kramer and T. R. Tyler (eds.), Trust in

Organiwtions. Thousand Oaks: Sage.

References - 535

Sherer, P. D. 1995 'Leveraging Human Assets in Law Finns: Human Capital Structures and Organizational

Capabilities.' Industrial and Labor Relations Review 48: 671-691. Sherman, J. D., H. L. Smith, and E. R. Mansfield 1986 'The Impact of Emergent Network Structure on Organizational Socialization.' Journal of Applied

Behavioral Science 22: 53-63. Sheth,J.N. 1973 'A Model of Industrial Buyer Behavior.' Journal of Marketing 37: 50-56. Shl, Y., and M. J. Gregory 1994 'International Manufacturing Strategy: Structuring World·Wide Manufacturing Network.' Working

Papers in Manufacturing, No II, Cambridge University. Shrum, W., and R. Wuthnow 1988 'Reputational Status of Organizations in Technical Systems.' American Journal of Sociology 93:

882-912. Slmmel,G. 1950 The Sociology of Georg Simmel. Translated by Kurt H. Wolff. Glencoe: Free Press. 1955 Conflict and the Web of Group-Affiliations. Translated by K. H. Wolff and R. Bendix. New York:

Free Press. 1959 Sociology of Religion. Translated by C. Rosenthal. New York: Philosophical Library. Simmons, D. B. 1985 'The Nature of the Organizational Grapevine.' Supervisory Management Nov.: 39-42. Simon, H. A. 1957 Models of Man: Social and Rational: Mathematical Essays on Rational Human Behavior in Social

Settings. New York: Wiley. 1991 'Organizations and Markets.' Journal of Economic Perspectives 5: 25-44. Skinner, W. 1969 'Manufacturing-Missing Link in Corporate Strategy.' Harvard Business Review 47: 136-145. Skowronski, J. J., and D. E. Carlston 1989 'Negativity and Extremity Biases in Impression Formation: A Review of Explanations.'

Psychological Bulletin 105: 131-142. Slack, N. 1991 The Manufacturing Advantage. London: Mercury Business Books. Slaughter, S., and L. L. Leslie 1997 Academic Capitalism: Politics. Policies. and the Entrepreneurial University. Baltimore: Johns

Hopkins University Press. Smart,A. 1993 'Gifts, Bribes, and Guanxi-A Reconsideration of Bourdieu's Social Capital.' Cultural

Anthropology 8: 388-408. Smelser, N., and R. Swedberg (eds.) 1994 Handbook of Economic Sociology. Princeton: Princeton University Press. Smigel, E. O. 1969 The Wall Street Lawyer. Professional Organization Man? Bloomington: Indiana University Press. Smit,B. 1996 Heineken. un Leven in de Brouwerij [Heineken, A life in the Brewery, in Dutch]. Nijmegen: Sun. Smith,H.C. 1996 Sensitivity to people. New York: McGraw-Hill. Smith, K. G., S. J. Carroll, and S. J. Ashford 1995 'lntra- and Interorganizational Cooperation: Toward a Research Agenda.' Academy of Management

Journal 38: 7-23. Smith-Doerr, L. 1997 'Framing Scientific Careers: Biotechnology as Legitimate Science.' Paper presented at the annual

meetings of the Society for Social Studies of Science, Tucson, AZ. Smitka, M. J. 1991 Competitive Ties: Subcontracting in the Japanese Automotive Industry. New York: Columbia

University Press. Snijders, T. A. B. 1996 'Stochastic Actor-Oriented Dynamic Analysis.' Journal of Mathematical SOCiology 21: 149-172.


Snow, C. C., and S. A. SneU 1993 'Staffing as Strategy.' In N. Schmitt and W. C. Bonnan (eds.), Personnel Selection in

Organizations, 448-478. San Francisco: Jossey-Bass. Snow, C. C., R. E. Miles, and H. J. Coleman, Jr. 1992 'Managing 21st Century Network Organizations.' Organizational Dynamics 20: 5-20. Sobel, M. E., M. Hout, and O. D. Duncan 1985 'Exchange, Structure, and Symmetry in Occupational Mobility.' American Journal of Sociology

91: 359-372. Solomon, M. R., C. F. Surprenant, J. A. Czepiel, and E. G. Gutman 1985 'A Role Theory Perspective on Dyadic Interactions: The Service Encounter.' Journal of Marketing

49: 99-111. Song,J. H. 1982 'Diversification Strategies and the Experience of Top Executives of Large FI!IIlS.' Strategic

Management Journal 3: 377-380. Sorenson, A. 1996 'The Structural Basis of Social Inequality.' American Journal of Sociology 101: 1333-65. Spaeth, J. L., and D. P. O'Rourke 1994 'Design and Implementation of a National Sample of Work Establishments.' American Behavioral

Scientist 37: 872-890. Spector,M. 1973 'Secrecy in Job Seeking Among Government Attorneys: Two Contingencies in the 1beory of

Subcultures.' Urban Life and Culture 2: 211-229. Starr, J. A., and I. C. MacMiUan 1990 'Resource Cooptation Via Social Contracting: Resource Acquisition Strategies for New Ventures.'

Strategic Management Journal 11 : 79-92. Steers, R. M., Y. K. Shin, and G. Ungson 1989 The Chaebol: Korea's New Industrial Might. New York: Harper and Row. Stevens, G, C, 1989 'Integrating the Supply Chain.' International Journal of Physical Distribution and Materials

Management 19: 3-8. Stewart, J. B. 1991 Den of Thieves. New York: Simon & Schuster. Stewman,S. 1986 'Demographic Models of Internal Labor Markets.' Administrative Science Quarterly 31: 212-247. Stewman, S., and S. L. Konda 1983 'Careers and Organizational Labor Markets: Demographic Models of Organizational Behavior.'

American Journal of Sociology 88: 637~85. Stigler, G. J. 1961 'The Economics of Information.' Journal of Political Economy 69: 213-225. 1962 'Information in the Labour Market.' Journal of Political Economy 70: 94-105. Stinchcombe, A, L. 1965 'Social Structure and Organizations.' In James G. March (ed.), Handbook of Organizations, 142-193.

Chicago: Rand McNally. 1985 'Contracts as Hierarchical Documents.' In A. Stinchcombe and C. Heimer, Organization Theory

and Project Management, 121-171. Oslo: Norwegian University Press. 1986 'Nonns of Exchange.' In A. L. Stinchcombe, Stratification and Organization, Selected Papers,

231-267. New York: Cambridge University Press. 1990 Information and Organizations. Berkeley: University of California Press. Stock, J. R., and D. M. Lambert 1987 Strategic Logistics Management (2nd. edition). Homewood: Irwin. Stoelhorst, J. W. 1997 'In Search of a Dynamic Theory of the FInD.' Doctoral dissertation, Twente University, The

Netherlands. Stokman, F. N., J . van der Knoop, and F. W. Wasseur 1988 'Interlocks in the Netherlands: Stability and Careers in the Period 1960-1980.' Social Networks 10:

183-208. Stokman, F. N., R. Ziegler, and J. Scott 1985 Networks of Corporate Power. Cambridge: Polity Press.

References - 537

Stolzenberg, R. M. 1980 'Measurement and Decomposition of Causal Effects in Nonlinear and Nonadditive Models.'

Sociological Methodology 11: 459-488. Strauss, D., and M. Ikeda 1990 'Pseudolikelihood Estimation for Social Networks.' Journal of the American Statistical Association

85: 204-212. Streed" W. 1997 'Beneficial Constraints: On The Economic limits of Rational Voluntarism.' In 1. R. Hollingsworth

and R. Boyer (eds.), Contemporary Capitalism. The Embeddedness of Institutions, 197-219. Cambridge: Cambridge University Press.

Stuart, T. E. 1998a 'Interorganizational Alliances and the Perfonnance of Firms: A Study of Growth and Innovation

Rates in a High-Technology Industry.' Working paper, University of Chicago. 1998b 'Producer Network Positions and Propensities to Collaborate: An Investigation of Strategic

Alliance Formation in a High-Technology Industry.' Administrative Science Quarterly (forthcoming).

Stuart, T. E., H. Hoang, and R. Hybels 1999 'Interorganizational Endorsements and the Performance of Entrepreneurial Ventures.'

Administrative Science Quarterly (forthcoming). Suchman, M. C. 1994 'On Advice of Counsel: Law Firms and Venture Capital Funds as Information Intermediaries in the

Structuration of Silicon Valley.' Unpublished doctoral dissertation, Department of Sociology, Stanford University.

Sundqvlst, S. I. 1990 Agarna och Markten I Sveriges Borsbolag 1986. [Owners and Power in Sweden 1986, in

Swedish] . Stockholm: Dagens Nyheters ForJag. Surprenant, C. F., and M. R. Solomon 1987 'Predictability and Personalization in the Service Encounter.' Journal of Marketing 51 : 86--96. Suzuki, S., and R. W. Wright 1985 'Financial Structure and Bankruptcy Risk in Japanese Companies.' Journal of 1ntemational

Business Studies 16: 97-110. Swan, J. E., and F. Trawick 1980 'Satisfaction Related to Predictive vs. Desired Expectations.' Association of Consumer Research Swartz, T. A., and S. W. Brown 1989 'Consumer and Provider Expectations and Experiences in Evaluating Professional Service QUality.'

Journal of the Acackmy of Marketing Science 17: 189-195. Swedberg, R. 1994 'Markets as Social Structures.' In N. J. Smelser and R. Swedberg (eds.), The Handbook of

Economic Sociology, 255-282. Princeton: Princeton University Press. Tagiuri,R. 1958 'Social Preference and its Perception.' In R. Tagiuri and L. Petrullo (eds.), Person Perception and

Interpersonal Behavior. Stanford: Stanford University Press. Talmud,I. 1992 'Industry Market Power, Industry Political Power, and State Support: The Case of Israeli Industry.'

In G. Moore and J.A. Whitt (eds.), Research in Politics and Society 4, 35-62. Greenwich: JAI Press.

1994 'Relations and Profits: Imperfect Competition and Its Outcome.' Social Science Research 23: 109-135.

Talmud, I., and D. N. Izraell 1998 'The Relationship Between Gender and Performance Issues of Concern To Directors: Correlates

Or Institution?' Journal of Organizational Behavior 19 (forthcoming). Talmud, I., and G. S. Mesch 1997 'Market Embeddedness and Corporate Instability: The Ecology of Inter-Industrial Networks' Social

Science Research 26: 41~1. Talmud, I., and Y. Yanovitzky 1998 'The Contradictory Demand Paradox: Social Embeddednesss and the Political Economy of the

Firm' Israeli Sociology 1: 1-43 (forthcoming, in Hebrew).


Tam,S. 1990 'Centrifugal and Centripetal Growth Processes: Contrasting Ideal Types for Conceptualizing the

Developmental Patterns of Chinese and Japanese Finns.' In S. R. Clegg and S.G. Redding (eds.), Capitalism in Contrasting Cultures, 153-183. New York: De Gruyter.

Tarrow,S. 1996'Making Social-Science Work Across Space and Time-A Critical Reflection on Putnam, Robert

'Making Democracy Work'.' American Political Science Review 90: 389-397. Taylor, J. B. 1994 'Discovery of New Medicines.' In J. P. Griffin, 1. O'Grady, and F.O. Wells (eds.), The Textbook of

Pharmaceutical Medicine (2nd. edition). Antrim: Greystone Books. Taylor, S. E. 1991 'Asymmetrical Effects of Positive and Negative Events: The Mobilization-Minimization

Hypothesis.' Psychological Bulletin 110: 67-85. Taylor, M., and S. Singleton 1993 'The Communal Resource-Transaction Costs and the Solution of Collective Action Problems.'

Politics and Society 21 : 195-214. Teece,D.J. 1986 'Profiting from Technological Innovation: Implications for Integration, Collaboration, licensing and

Public Policy.' Research Policy 15: 285-305. 1988 'Technological Change and the Nature of the FJCID.' In G. Dosi, C. Freeman, R. Nelson, G. Silverberg,

and L Soete (eds.), Technical Change and Economic Theory, 256-281. London: Pinter. Tevlin,J. 1997 'Non-Compete Clauses are a Growing Trend.' Minneapolis Star Tribune May 23: 01. Thackray, J. 1986 'America's Vertical Cutback.' Management Today, June. Thaler R., and H. Sherrin 1981 'An Economic Theory of Self-Control.' Journal of Political Economy 89: 392-406. Thibaut, J. W., and H. H. Kelley 1959 The Social Psychology of Groups. New York: Wiley. Thomas, D. A. 1993 'Racial Dynamics in Cross-Race Developmental Relationships.' Administrative Science Quarterly

38: 169-194. Thomas, D. A., and M. C. Higgins 1996 'Mentoring and the Boundaryless Career: Lessons from the Minority Experience.' In M. Arthur and

D. Rousseau (eds.), Boundaryless Career: A New Employment Principle for a New Organimtional Era. New York: Oxford University Press.

Thompson, G. 1982 'The Finn as a Dispersed Social Agency.' Economy and Society 11: 233-250. Thompson, G., J. Frances, R. Levacic, and J. Mitchell 1991 Markets, Hierarchies and Networks. London: Sage. Thompson, J. D. 1967 Organimtions in Action: Social Science Bases of Administration. New York: McGraw-Hill. Thorelli, H. B. 1990 'Networks: Between Markets and Hierarchies.' In D. Ford (ed.), Understanding Business Markets.

London: Academic Press. Timmons, J. A. 1995 New Venture Creation: Entrepreneurship for the 21" Century. Burr Ridge: Irwin. Tolbert, P. S., and L. G. Zucker 1983 'Institutional Sources of Change in the Formal Structure of Organizations: The Diffusion of Civil

Service Refonn, 1880-1935.' Administrative Science Quarterly 28: 22-39. Towill,D. 1991 'Supply Chain Dynamics.' Computer Integrated Manufacturing 4: 197-208. Trajtenberg, M. 1990 'A Penny for Your Quotes: Patent Citations and the Value of Information.' Bell Journal of

Economics 21 : 172-187. Travers, J., and S. Milgram 1969 'An Experimental Study of the 'Small World' Problem.' Sociometry 32: 425-443.

Triandls, H. C. 1995 Individualism and Collectivism. Boulder: Westview. Trice, H. M., and D. A. Morand

References - 539

1989 'Rites of Passage in Work Careers.' In M. B. Arthur, D. T. Hall, and B. S. Lawrence (eds.), Handbook of Career Theory, 397-416. New York: Cambridge University Press.

Tsui, A. S., and J. L. Fahr 1997 'Where Guanxi Matter. Relational Demography and Guanxi in the Chinese Context.' Work and

Occupation 24: 56-79. Tsui, A. E., and C. A. O'ReiUy, III 1989 'Beyond Simple Demographic Effects: The Importance of Relational Demography in Superior

Subordinate Dyads.' Academy of Management Journal 32: 402-423. Tsui, A. E., T. D. Egan, and C. A. O'Reilly, m 1992 'Being Different: Relational Demography and Organizational Attachment.' Administrative Science

Quanerly 37: 549-579. Tuma, N. B., and M. T. Hannan 1984 Social Dynamics: Models and Methods. New Yorlc: Academic Press. Turban, D. B., and T. W. Dougherty 1994 'Role of Protege Personality in Receipt of Mentoring and Career Success.' Academy of

Management Journal 37: 688-702. Turnbull, P., and J. P. Valla (eds.) 1986 Strategies for International Industrial Marketing. London: Croom-Helm. Tushman, M. L. 1977 'Special Boundary Roles in the Innovation Process.' Administrative Science Quanerly 22: 587-

605. 1978 'Task Characteristics and Technical Communication in R&D.' Academy of Management Journal

21: 624-645. Tushman, M. L., and P. Anderson 1986 'Technological Discontinuities and Organizational Environments.' Administrative Science

Quanerly 31: 439-465. Upton, D. M., and A. McAfee 1996 'The Real Virtual Factory.' Harvard Business Review 74: 123-133. Useem,M. 1979 'The Social Organization of the American Business Elite and Participation of Corporate Directors

in the Governance of American Institutions.' American Sociological Review 44: 553-572. 1982 'Classwide Rationality in the Politics of Managers and Directors of Large Corporations in the

United States and Great Britain.' Administrative Science Quanerly 27: 199-226. 1996 Investor Capitalism: How Money Managers Are Changing the Face of Corporate America. New

York: Basic Books. Useem, M., and J. Karabel 1986 'Pathways to Top Corporate Management.' American Sociological Review 51: 184-200. Uzzl,B. 1996a 'The Sources and Consequences of Embeddedness for the Economic Performance of

Organizations: The Networlc Effect.' American Sociological Review 61: 674-698. 1996b 'Coase Encounters of a Sociological Kind: A Sociological Perspective on Strategy.' In J. Dutton

and J. A. C. Baum (eds.), The Embeddedness of Strategy. New York: JAI Press. 1997a 'Social Structure and Competition in Interfirm Networks: The Paradox of Embeddedness.'

Administrative Science Quarterly 42: 3~7. 1997b 'A Structural Embeddedness Approach to Deindustriatization and Organizational Decline.' The

International Journal of Sociology and Social Policy 17: 111-155. Uzzi, B., and Z. Barsness 1998 'Contingent Employment in British Establishments: Organizational Determinants of the Use of

Fixed-Term Hires and Part-Time Workers.' Social Forces 76: 967-1005. Uzzi, B., and J. Gillespie 1998 'Embeddedness and Market Dynamics: The Case of Banks, Corporate Borrowers and the Cost and

Availability of Financial Capital.' Working paper. Northwestern University. Van de Mark, D. 1996 'Ovitz Out at Disney.' Cable News Network December 12.


Van de Ven, A. H. 1993 'The Development of an Infrastructure for Entrepreneurship.' Journal of Business Venturing 8:

211-230. Van de Ven, A. H., D. C. Emmit, and R. Koenig 1975 'Frameworks for Inter-Organisational Analysis.' In A. R. Negandhi (ed.), Inter-Organisational

Theory. Kent State: Kent State University Press. Van Engelen, J. M. L., P. Terlouw, D. J. Kiewiet, and S. W. F. Ornta 1999 'Towards a Polarity Strategy in Managing Product Development Teams'. Unpublished Manuscript,

School of Management and Organization, University of Groningen. Van Lange, P. A. M., W. B. G. Liebrand, D. M. Messick, and H. A. M. Wilke 1992 'Social Dilemmas: The State of the Art.' In W. Liebrand, D. M. Messick, and H. Wilke (eds.),

Social Dilemmas: Theoretical Issues and Research Findings. Oxford: Pergamon Press. Van Maanen, J. 1975 'Police Socialization: A Longitudinal Examination of Job Attitudes in an Urban Police

Department.' Administrative Science Quarterly 20: 207-228. 1983 'Golden Passports: Managerial Socialization and Graduate Education.' The Review of Higher

Education 6: 435-455. Van Maanen, J., and E. Scbein 1979 'Toward a Theory of Organizational Socialization.' In B. M. Staw (ed.), Research on

Organizational Behavior 1: 209-264. Greenwich: JAI Press. Viilker, B., and H. Flap 1998 'Getting Ahead in the GDR. Social Capital in the Status-Attainment under Communism.' Acta

Sociologica 41 (forthcoming). Voss, C. A., C. G. Armistead, R. Johnston, and B. Morris 1985 Operations Management in Service Industries and the Public Sector. Chichester: Wiley. Wagner, G. W., J. Pfeffer, and C. A. O'ReUly, m 1984 'Organizational Demography and Turnover in Top-Management Groups.' Administrative Science

Quarterly 29: 74-92. Waldinger, R. 1995 'The Other Side of Embeddedness-A Case-Study of the Interplay of Economy and Ethnicity.'

Ethnic and Racial Studies 18: 555-580. Walker, G., B. Kogut, and W. Shan 1997 'Social Capital, Structural Holes and the Formation of an Industry Network.' Organization Science

8: 109-125. Walker, M. E., and S. Wasserman

1987 TRIADS. A Computer Program for Triodic Analyses. Urbana: University of Dlinois. Walsh, J. S. Kuker, N. Maloney, and S. M. Gabbay 1999 'Connecting Minds: CMC and Scientific Work.' Journal of the American Society for Information

Science. Forthcoming. Walsh,J.P. 1988 'Selectivity and Selective Perception: An Investigation of Managers' Belief Structures and

Information Processing.' Academy of Management Journal 31 : 873-896. Walsh, J. P., and J. K. Seward 1990 'On the Efficiency of Internal and External Corporate Control Mechanisms.' Academy of

Management Review 15: 421-458. Walton,M. 1997 Car: A Drama of the American Workplace. New York: Norton. Wasserman, S., and D. Iacobucci 1988 'Sequential Social Network Data.' Psychometrika 53: 261-282. Wasserman, S., and K. Faust 1994 Social Network Analysis: Methods and Applications. New York: Cambridge University Press. Waters,M. 1989 'Collegiality, Bureaucratization, and Professionalization: A Weberian Analysis.' American Journal

of Sociology 94: 945-72. Waterson, M. 1990 'The Economics of Product Patents.' American Economic Review 80: 860-869. Watson, D., and A. Tellegen 1985 'Toward a Consensual Structure of Mood.' Psychological Bulletin 98: 219-235.

References - 541

Webb,E. 1996 'Trust and Crisis.' In R. Kramer and T. Tylor (eds.), Trust in Organizotions: Frontiers of Theory

and Research, 288-301. London: Sage. Weber,M. 1978 Economy and Society: An Outline of Interpretive Sociology. Translated by E. Fischoff et al., edited

by G. Rothard and C. Wittich. Berkeley: University of California Press. (first edition published in 1922).

Wegener,B. 1991 'Job Mobility and Social Ties: Social Resources, Prior Job, and Status Attainment.' American

Sociological Review 56: 60-71. Weick, K. E, 1979 The Social Psychology of Organizing. Reading: Addison Wesley. 1995 Sensemaking in Organizotions. Thousand Oaks: Sage. Weldenbaum, M., and S. Hugbes 1996 The Bamboo Network. New York: The Free Press. Wells,M. 1995 'Hollywood Super Agent Michael Ovitz Named Disney President.' Detroit News August 15. Wernerfelt, B. 1984 'A Resource Based View of the Finn.' Strategic Management Journal 5: 171-180. Weston, J., K. Cbung and S. Hoag 1990 Mergers, Restructuring. and Corporate Control. Englewood Cliffs: Prentice-Hall. Wetzel, W. E. 1987 'The Infonnal Venture Capital Market: Aspects of Scale and Market Efficiency.' Journal of

Business Venturing 2: 299-313. White,H.C. 1961 'Management Conflict and Sociometric Structure.' American Journal of Sociology 67: 185-199. 1981 'Where Do Markets Come From?' American Journal of Sociology 87: 517-547. 1988 'Varieties of Markets.' In B. Wellman and S. D. Berkowitz (eds.). Social Structure: A Network

Approach, 226-260. New York: Cambridge University Press. 1992 Identity and Control: A Structural Theory of Social Action. Princeton, NJ: Princeton University

Press. White, H. C., S. A. Boorman, and R. L. Brelger 1976 'Social Structure from Multiple Networks I: Blockmodels of Roles and Positions.' American

Journal of Sociology 81 : 730-780. Whitely, W., T . W. Dougherty, and G. F. Dreher 1991 'Relationship of Career Mentoring and Socioeconomic Origin to Managers' and Professionals'

Early Career Progress.' Academy of Management Journal 34: 331-351. Wiersema, M. F., and K. A. Bantel 1992 'Top Management Team Demography and Corporate Strategic Change.' Academy of Management

Journal 35: 91-121. Wllklnson, I, F. 1981 'Power, Conflict and Satisfaction in Distribution Channels-An Empirical Study.' International

Journal of Physical Distribution and Materials Management II: 20-30. Williamson, O. E. 1975 Markets and Hierarchies: Analysis and Antitrust Implications. New York: Free Press. 1981 'The Economics of Organization: The Transaction Cost Approach.' American Journal of Sociology

87:548-577. 1985 The Economic Institutions of Capitalism: Firms, Markets. Relational Contracting. New York: Free

Press. 1993 'Ca1culativeness, Trust. and Economic Organization.' Journal of Law and Economics 36: 453-486. 1994 'Transaction Cost Economics.' In N. J. Smelser and R. Swedberg (eds.), The Handbook of Economic

Sociology, 77-107. Princeton: Princeton University Press. 1996 The Mechanisms of Governance. New Yode Oxford University Press. Wills, T.A. 1991 'Social Support and Interpersonal Relationships.' In M. S. Clark (ed.), Prosocial Behavior. Review

of Personality and Social Psychology 12: 265-289. Newbury Park: Sage.


Windolf, P., and S. Wood 1988 Recruitment and Selection in the Labour Market. A Comparative Study of Britain and West Germany.

Avebury: Aldershot. Winship, C., and R. Mare 1984 'Regression Models with Ordinal Variables.' American Sociological Review 49: 512-525. Wistow,G. 1992 'The Health Service Policy Community.' In D. Marsh and R. A. W. Rhodes (eds.). Policy Networks

in British Government. Oxford: Clarendon Press. WoUT,M.F. 1994 'Building Trust in Alliances.' Research Technology Management 37: 12-15. Womack, J. P., D. T. Jones, and D. Roos 1991 The Machine That Changed the World: The Story of Lean Production. New York: Rawson

Associates. Woodard, K. L. 1995 'Introduction and Acceptance of Inter-Organizational Agreements: The Experience of Seventy

Five Administrators in One County.' Administration in Social Work 19: 51-83. Woodard, K. L., and P. Doreian 1994 'Utilizing and Understanding Community Service Provision Networks: A Report of Three Case

Studies Having 583 Participants.' Journal of Social Service Research 18: 1--41. Woolcock, M. 1997 'Social Capital and Economic Development: Toward a theoretical synthesis and policy framework.'

Theory and Society 27: 151-208. Worldbank 1997 'Social capital: The Missing linkT Expanding the Measure of Wealth: Indicators of Environmentally

Sustainable Development. Environmentally Sustainable Development Studies and Monograph Series 17. Washington D.C.: The World Bank.

Wright, E. 0., and D. Cho 1992 ibe Relative Permeability of Class Boundaries to Cross-Class Friendships: a Comparative Study of

the United States, Canada, Sweden. and Norway.' American Sociological Review 57: 85-102. Wrong,D.H. 1961 The Oversocialized Concept of Man in Modem Sociology.' American Sociological Review 26:

183-193. Yoffie,D.B. 1996 'Competing in the Age of Digital Convergence.' California Management Review 38: 31-53. Yoshino, M. Y., and T. B. Lirson 1986 The Invisible Link: Japan's Sogo Shosha and the Organization of Trade. Cambridge: MIT Press. Yoshino, M. Y., and U. S. Rangan 1995 Strategic Alliances: An Entrepreneurial Approach to Globalization. Boston: Harvard Business

School Press. Zagare, F. C. 1984 Game Theory: Concepts and Applications. Newbury Park: Sage. Zaheer, A., and N. Venkatraman 1995 'Relational Governance as an Interorganizational Strategy: An Empirical Test of the Role of Trust

in Economic Exchange.' Strategic Management Journal 16: 373-392. Zajac, E. J., and C. P. Olson 1993 'From Transaction Costs to Transactional Value Analysis: Implications for the Study of

Interorganizational Strategies'. Journal of Management Studies. 30: 131-145. Zeggelink, E. P. H. 1994 'Dynamics of Structure: An Individual Oriented Approach.' Social Networks 16: 295-333. Zenger, T. R., and B. S. Lawrence 1989 'Organizational Demography: The Differential Effects of Age and Tenure Distributions on

Technical Communications.' Academy of Management Journal 32: 353-376. Zey,M. 1984 The Mentor Connection. Homewood: Dow Jones-Irwin. Zucker, L. G. 1986 'Production of Trust: Institutional Sources of Economic Structure 1840-1920. 'In B. M. Staw & L.

L. Cummings (eds.). Research in Organizational Behavior. 8: 53-III. Greenwich: JAI Press.

References - 543

Zuckerman, E. W. 1999 'The Categorical hnperative: Securities Analysts and the Legitimacy Discount.' American Journal

of Sociology (forthcoming). Zukin, S., and P. DiMaggio 1990 'Introduction.' In S. Zukin and P. DiMaggio (eds.), Structures of Capital. The Social Organization

of the Economy, 1-36. New York: Cambridge University Press. Zuscovitch, E. 1994 'Sustainable Differentiation; Economic Dynamism and Social Nonns.' Paper presented at the J. A.

Schumpeter conference, Miinster, August.

Index

• A absence of ties 92. 103 acceptance 163 access 164.389.392.479

restrictions 377 to infonnation 299. 356. 453

accessibility 80. 81 accumulation processes 461 acquisition strategies 272 acquisitions 266-283. 462. 468. 474. 478 action

dilemmas 77 set 29

activity model 410 actor

centrality 26 connectedness 25 dissimilarity 4. 328. 329 prestige 377 similarity 327. 329. 334 • s location 444

actors 80. 123. 167. 285. 300. 302. 303. 324. 327.379. 400.409.410.449.451.458.461

boundaries 22 positions 18

adaptability 308. 318 adhocracies 462 administrative costs 185 advantage viii adversarial ties 120 adverse selection 448 advertisements 205. 207 advice 248. 263 advising 24. 245 age 402 aggregation 46 alliance

capital 376-389 formation 30. 387. 432 networks 29 partners 433 relationships 491

alliances 24. 165 outcomes 36

allocation 461 of resources 446

altercentric uncertainty 431. 433. 436-439. 443 Analytic Hierarchy Process (AHP) method 370.

372.373.375 anticipated cooperation 305 Apple 22. 67. 95. 423 application 394 appropriable social structure 219 Araujo. Luis 59. 69. 71. 75. 76. 78. 300. 410. 505 arm's-Iength ties 446. 452. 453. 458 Arrow's paradox 349 Asian societies 54 assessments 134. 136 asset 74. 358. 447. 462

of social capital 318 specifity 30

assignment of patents 391 associate group 47 associates 247. 250 attitude similarity 328 audit firms 59. 60 authority 24

structures 85 system 243

automobile industry 100. 102 automorphic equivalence 27 avoidance of ties 95. 97

B background effects 281 Baker. Wayne E. 37. 80. 93. 95. 96. 97. 98. 99.

100. 104. 111. 119. 149. 180. 300. 338. 447. 449.450.452.453.454.458.493.496

bank-firm relationship 450. 454. 459 banking relationship 451 base technologies 368 baseline model 249 basic advice network 263 Bayesian infonnation criterion (BIC) 128 behavioral theories 424 Benassi. Mario 4. 58. 80. 115. 117. 292. 294.

306.325.488.498


benefit-rich networks 426 benefits 219. 220 Benetton 7. 409. 423. 426 between-finn collaborative 404 betweenness centrality 26. 41 bilateral governance 341 biotechnology 390-408. 432.438.447 Blau-tie 253-256. 258 Boeing 3. 66. 369 boundaries of organizations 43 boundary 410. 411

spanner 47. 351. 354 system's identity 52 transaction systems 51. 65

Boxman. Ed 6. 19. 60. 87. 181. 200. 211. 212. 213.214.331.431.487.494.499.507

Brass. Daniel J. 4. 14.40.41.42. 162. 167. 199. 232. 294. 324. 325. 327. 328. 329. 330. 333. 334.335.337.499.500.501.519.524

Breiger. Ronald L. 2. 92. 104. 119. 120. 122. 133. 136. 137. 147. 213. 265. 377. 399. 441. 484,485.500.512,522, 524, 541

bright side of social capital/ties 299. 304 buddy relationships 177 bull-whip effect 416 business

friendship 450. 452 group 30. 52. 54, 59. 64 relationships 358. 490 -risk view of trust 33 strategy 74

buyer-seller interactions 148-158

c calculativeness 83 capital viii, 73, 74 capital-in-general metaphor 86 career 174, 176. 178-180.336

capital 39 -cycle hypothesis 208 development 168. 334, 336 failure 165 help 163. 164 history 169 progress 162, 163

cartel 29,49, 52 Castellucci, Fabrizio 59.110.137.147.419.448.

465.468 causality 489 central firm 405 centrality 41. 42. 123. 125. 392. 397. 399. 400.

402-405,470.477,478 CEO

demographics 266-283 tenure 272. 278. 279

chaeboI30.49,50,54-58,loo challenge 163

Chandlerian principles 424 change 294 changes in the social system 221 changing network

patterns 32 structures 42

channel 431 behavior 424 management 423

Chicago lawyers 217. 223. 224. 226. 232 China 73, 82,95 chump pattern 153. 155, 157 clinical

networks 291. 292. 296 trials 399

close ties 453 closeness centrality 26. 330 closer ties 99 closure 94. 305, 355 cluster actors 27 co-attendance 23 coaching 163 codes of conduct 69 co-evolution of firms and networks 40 1 cognitive

corporate capital 107-110, 113. 115, 116 dimension of social capital 90 embeddedness 70 social capital 106

cohesion 27 collaboration 358-361. 363. 364. 369. 371. 373.

375.390-408 to patents 393

collaborative activities 399. 402 agreement 371 capital 390. 392, 394, 395,406. 414 experience 400, 403 relationships 414 social capital 406 ties 120

collaborator 390 collaborators in development 345 collective

action 182 dilemma 72. 84 dilemma 70 goods 80 social

capital 240 structure 241

collectivism 100 collegial organizations 238, 239. 241 co-management relations 119 commercial chain 416 comrnitment 332, 374 communication 326. 374

networks 34. 83, 334

communities of practice 344 of trust 91

community structures 464 competence 343-345, 348, 350, 357, 365, 370,

374,393,394 trust 346

competencies of motivation and morality 348 competition 155 competitive

advantage 27,182,379 corporate capital 107, 111, 113 corporate structure liS modalities 21 patterns 157 social capital 106

complementarity 84 complex ties 399 complications 362 composition of the network 304 concept of organization 45 conceptualization 43 confirmation 163 conflict 337,424 connectivity 98 consistency 59 conshr.rinment24O conshr.rint 248 consultation network 318 consumer loyalty 149 consumer-to-consumer relations ISO contingencies 352, 485 contractual arrangements 371 contribution to standing 141 control 219 controllability 80, 81 cooperation 155, 369, 374, 424 cooperative

behavior 152 modalities 21 patterns 157 relationships 356

coordination 326 failures 309, 311, 313, 315-317

core group 47, 48 corporate

acquisitions 267-270 actors 89 alliances 32 entities 20 leverage III liability 106-117 social

capital 1, 10-13, 106-134, 231, 238, 240, 291, 312, 320, 446-459, 483-493

capital management 491, 492 capital theory 458, 484, 485, 491

control lIS structure 106-117

strategy 359 superiority 113 technological prestige 378

corrective action 374 cost of financial capital 446-459 costs 420

of search 207 counseling 163 coworker ties 245 coworkers 248, 255, 263

conshr.rints 249 network 240, 251, 256, 263 teams 259

creation of social capital 76, 77, 83 credibility 290

Index -547

cross-firm economy of learning 345 cross-licensing 396 cross-stockholdings 64 esc 1,2,9, 10, 11, 12, 13,483,484,485,486,

489,491,493,509 cultural

conditions 99 context 99 embeddedness 70 organizations 436

customer-employee dyad 151 customer service dyads 148-157 customers 148

D dark side

of cooperation 357 of social capital 298-322

data interactions 153 Davis-Blake, Alison 181,497 decentralized organization 242 decision uncertainty 271 dedicated biotechnology firm (DBF) 390-391,

394,398,399,404,405 definitions of social capital 118 demographic

composition 377 variables 328

demographics 329 dense networks 110 density 98 depreciation 76 design 88-105 destruction of social capital 77 development of social capital 161-179 differentiated markers vii diffused reciprocity 77 diffusion of the hr.rining 333 dimension of social capital 103


direct approaches 183 social ties 181

dissatisfaction 415, 416 dissimilarity 328 dissolution 62 distribution 399

of exchange 457 of infonnation 5 of power 5 of strategies 100

disunion strategy 88, 89, 93-95, 98-100, 102, 104,493

diversity 406 of network activity 403

division of labour 70 domain-expanding acquisitions 269, 270 domination 40 Doreian, Patrick 14, 104, 136, 138, 149, 441,

485,500,505,521,542 downstream partners 357 dual mode

ego-network structures 458 networks 457

duplex ties 255 durability 76 duration of relationships 454, 458 Dutch managers 201-203, 205, 207, 210-212 dyad 162,240,312,342,399,446 dyad-specific social capital 77 dyadic

interaction model 154 interaction strategies 152 international corporate alliances 32 modeling 149 parameter patterns 155 relationship 150, 410, 413, 415, 416, 450,

451 supply relationships 418 ties 251, 253

dynamic task environments 318 dynamics of social capital accumulation 401

E Easton, Geoff 59, 69, 71, 74, 75, 76, 78, 285,

300,410,419,496,504,505 economic

exchange 68 performance 237-265 relations 21 resources 68, 69, 74, 75, 410 sociology 458 theory 347, 424 value 396

of ties 217 economy of time 460-479 economy-society interaction 68

education level 276 sector 135, 138, 141, 142, 144, 145

educational background 266-283 effective ties 115 effectiveness of social capital 320 egocentric

network 22 uncertainty 431-445

ego-network 446, 451453, 458, 459 size 449, 454, 456 structure 452

eigenvector centrality 123 elite ties 223, 228 elites 217, 220-224, 227, 228, 230, 231 embedded ties 45, 453, 458 embeddedness 36, 46, 58, 59, 64, 65, 68-70, 80,

86, Ill, 237, 238, 241, 245, 259-262, 377, 448, 459,465

approach 446459 embedding of relations 342 emerging technologies 367 emotional support 199 employability 38 employee-customer relations 149 endowments of social capital 79 end-users 148 entrepreneurial

community 462, 465, 468 firms 436, 437, 440 organizations 303, 320 research 90 strategies 93, 97, 100 successes 476 ventures 468

entrepreneurs45,220, 302,459,463,478 entrepreneurship 80, 88, 89, 320, 460, 461 equity 274

-based contracts 35 partnerships 28, 29 pressures 186 swap 29

equivalence 27 erosion of social capital 17 establishment of collaborative agreement 360,

363 Europe 71, 369 evaluability 78 evaluation of social capital 78, 79 exchange 449

of hostages 348 partners 378, 451 relations 377, 432, 450 system 241, 242

exchanges of resources 260 exclusivity 99 expatriates 174, 176 expectations 414416

expert based organizations 284 exposure 163, 164 external

corporate ties 271 economy of cognitive scope 345 networks 49, 280 pressures 186 relationships 358 resources 374

extra-organizational social capital 232

F face-ta-face groups 76 failure 165,462, 468,474 family 84 Far East 369 favoritism 183, 186 female managers 199 Ferlie, Ewan 14, 149, 288,289, 297,507,529 finance 280 financial

aid 24 capital 44, 53, 436, 442, 446, 459 capitallIlllrlcets 446 planning 451

finders 243 firm age 400 firm-environment interface 50 firm-level

social capital 237, 256 variables 400

firm social capital 240 firm's boundaries 47 firm-ta-firm relations 150,488 fixed-effects specification 40 1 Rap, Henk 6, 13, 19,60,87, 119, 181, 198,200,

202, 203, 211, 212, 213, 214, 215, 216, 238, 239,299, 300, 318, 331 , 431,487,494, 499, 500,501,504, 507,525,540

flexibility 352 flow of information 458 focal

firm 433 producer 431

focus 413, 416, 418 Foller 2-3 form of social capital 97 formal

differentiation 98 methods 183-185, 187 organizations 85, 230 peer review system 244 qualifications 232 relationships 9, 392 rules 243 rules for lIIllrlcet order 70 selection procedures 181

forms of capital 73-75 Forrester effect 416, 417 fosters trust 327, 450 France 198, 362

Index -549

Freeman, John 8, 24, 32, 66, 90, 93, 132, 293, 321 , 336, 448,452, 463,494,497,508, 512

frequency of an entrepreneurial strategy 97 friendship 163, 199,245,248,392

networks 263, 327 ties 249, 253-256, 259, 335

function of network ties 432 functional background 266-283 functions of metaphors 71

G Gabbay, Shaul M. 2, 4, 14, 87, 110, lIS, 117,

174, 215, 216, 219, 238, 264, 299, 300, 323, 325, 343, 408, 448, 449, 459, 486, 492, 493, 494, SOl, 509, 540

game theory 148, ISO, 155 gaming strategies 157 Gargiulo, Martin 3, 4, 58, 80, lIS, 117, 292, 294,

299, 300,309, 325, 492,493 gender 162, 177 gender-differentiated network 39, 40 general management 270 Genoese traders 70 gentlemen's agreement 243 geographical associates 70 Germany 198, 362 gift 82

exchange 73 Gillespie, James 1. 8, 87,418,447,450,459,539 global network structures 42 globalization 7 goal

definition 360, 362 specifity 3

ga-between 341-355, 374, 465 Gorman, Elizabeth H. 6, 201, 331 governance 348

mechanism 341 govemment355 grapevine 6 grinders 243 group

cohesiveness 305 dynamics 91

group-level social structure 237 social ties 241, 249 GSS 194,195 guanxi 82

H halfway model 126, 128


Han, Shin-Kap 104, 1l0, 117, 119, 122, 133, 136, 137, 147, 203, 399, 433, 441, 484, 485, 493,512

Harland Christine M. 7, 410, 411, 415,417,418, 420,422,424,488,513,516

Haunschild, Pamela R. 119,267, 268, 271, 274, 281,283,431,513

health care 286, 287, 289, 296

management 284 sector 135, 138, 141, 143-145

Heinz, John P. 14,220,222,223,224,231,232, 233,513

heterogeneity 400 Heineken 14,490 hierarchical

networks 177 relations 28, 85 status 239

hierarchy 98, 102, 164,248,285,326 Higgins, Monica 6, 9, 19, 178, 179, 209, 306,

319,336,489,514,538 high performers 254, 255 high prestige 377

firms 380, 387 high quality relationships 458 high status 432

actors 388 firms379,434,435,438,441~

organization 376 high-technology

firms 378 industries 379 organizations 284

Hitachi 66, 369, 383 Holland 166,200 home runs 468 homosocial reproduction 328, 331 homogeneous RC(l) model 126, 128 homophily 39, 327, 377, 387 honor 73 hostage keeping 348, 349, 353, 354 HRM implications 294 human capital 44, 53, 72, 84,164,177,182,203,

205,210-212,218,225,300,324,327,334 investments 20 theory 201

human resources 389 management323,324,332,334, 337

hybrid interfaces 458

I Iacobucci, Dawn 14,25,148,150, 154,514,515,

540 mM 50, 66, 95, 369, 379, 380, 383, 388, 396 ideal expectations 415 identity-making ties 114

immigrant communities 80 impatience 468, 477 IMP 409, 410, 424 implementation 360, 363, 374 income 205, 229, 231 indegree centrality 248 independent social capital 176 indirect social ties 181 individual

actions 182, 299 commitment 260 economic performance 260,261 networks 6 social capital 240, 248, 405 social network 240 social relationships 237 241,251

individual-collective distinction 46 individual-level social capital 240 individual-level social structure 237 individualism 100 individuals 91, 163, 167,391 influence 40, 219 informal

channels 183, 209 contacts 198, 205, 208, 211 interorganizational networks 288 methods 184-186, 191, 193, 197 procedures 185 recrui tment 183 relationships 9, 176 search 202, 212 social contact 220 staffmg 183, 184, 187, 189 ties 180, 203, 207, 337

information access 161 exchange 23, 416 network 406 transfer 82, 183, 219 - and communication technology (ICT)

344 infrastructure of the network 418 initial

mentoring 174 public offering of stock (IPO) 462, 467-

469,473,478 innovations 332 input-output models 136 institution building 69 institutional

conditions 214 context 88-105 institutionalization of ties 9 institutionalized trust 52 instrumental ends 83 integrity 83 intellectual

capital 390, 391, 394-3%, 403, 405 competencies 396 output 390, 399 property views 396 rights 395

Intel 38, 67, 96 intentional

sense 348 trust 346, 347, 350

interagency alliances 293 intercorporate

coalitions 377 relations 378, 379

interdependencies 317, 319, 374, 451 interdependent social systems 19 interests 119 inter-firm relationships 9, 46, 49, 50, 54, 149,

352, 450,491 intergrating mechanisms 176 interlocking of social relations 245 internal

management structures 290 styles 290

networks 49 rate of return (lRR) 465, 467 recruitment 182 social cohesiveness 221 staffing practices 180, 182, 187-193

intemet vii interorganizational

alliance network 378 boundaries 410 chains 416 collaboration 7,390, 392,395 connections 18 links 49,271,327 network 27, 36, 268, 269, 392, 406, 409,

410, 418,426,427 relationships 357 strategy 424, 427

relations (lOR's) 342, 343,377,391 , 392 social capital 282, 391 supply networks 409, 427 ties 42, 285 trust 35 interpersonal relations 347, 416 skills 185 transactions 152

intersubsidiary contacts 168, 169, 174 intervention of the state 77 intra-group relationships 57 intraorganizational

boundaries 177 connections 18 level 237, 238 mobility to careers 181

Index - 551

networks 5-7, 37,41,108,427 performance data 239 relations 116 ties 42, 166

inventory control 99 investment 399

banks 119, 120, 128, 131 , 132 lOR network (ION) 135, 137, 142 Israel110-115 Italy 77, 100, 306, 362

J Japan 71, 166, 342,362, 413, 419, 452 job 202, 208, 213

attainment 324 process 197

control 183 descriptions 195 -finding

method 204, 211 process 199, 200, 203, 209,214

placement 222 posting 183, 184, 186 satisfaction 205, 324, 329, 330 security 38

job-seeker 19, 20 job-search theory 201 joint

actions 23 problem solving 374 venture 29, 52, 399

judicial sector 135, 138, 142-145 jumping the gun 165, 166 just-in-time (JIT) 99, 414

K keiretsu 30, 49-51, 54-56, 58, 100 key technologies 368 Knoke, David 6, 7, 22, 23, 29, 38, 40, 41, 87,

105, 181, 187, 195, 196,219, 287, 330, 378, 383,470,494,501,516, 518,520,524

knowledge 393, 406 -based firms 116 production 390-408 spillover 406

Koput, Kenneth W. 7, 8, 50, 99, 132, 285, 293, 377, 378, 382, 391, 392, 394, 399, 400, 402, 405, 413,447,453,518,519,531

L Labianca, Giuseppe 4, 162, 232, 294, 324, 325,

337,519 labor

contracts 237 -265 markets 200, 205, 212


lack of social linkage 115 large-scale controls 70 Laumann. Edward O. 2. 14.22.219. 220. 222.

223. 224. 228. 231. 232. 233. 287. 513. 518, 520.533

lawyers 217-223, 225, 228, 229, 232, 246, 250 Lazega. Emmanuel 14,49,77,81,87,214,215,

218, 229, 239, 241, 242, 244, 246, 256, 257, 262,264,265,303,306,334,520,521

leadership 424 learning 333, 414, 435

by doing 463 Lee, Kyungmook 6, 7. 9, 14, 30, 50, 51, 63, 75,

77,87,132,215,285,347,408,502,529 Leenders, Roger Th.A.J. 4, 6, 14, 25, 32, 87, 104,

215, 216, 219, 286. 293. 294, 299, 325. 327. 328, 329, 333, 343, 389, 408, 449, 452, 459, 487,489,493,494,509,519,521

legislation 77 legitimacy 164, 479

of an entrepreneurial strategy 97 lending relationships 456, 459 length of relationship 449 level of social capital 97 levels of analysis 4, 43 liability 220, 240, see also social liability license alliance 379, 385 licensing 396. 399 link 50 local cycles 259 location effects vii logic of embeddedness 449 long term

buyer-supplier relationships 64 collaborative relationships 414 dependencies 358 licensing agreement 52 relationships 70,150,415,419,420,450 satisfaction 415

lottery ticket's value 395 low perfonners 254, 255 low-status firms 435, 442, 444

M macro-level sources of trust 34 Maghribi traders 70 magnitude of ties 25 maintenance of social capital 77, 83, 301, 354 management

of social capital 356-375 of supply 411

managerial background 267 managers 178-216,299,300,302,307-3\0,317,

318 , networks 199,300,306,310

manufacturing 399 marginal workers 47

market economies 69 exchanges 457 networks 444 opportunities 433 relations 28 stuff 431. 432 ties 458 uncertainty 433, 443

marketability 113 marketing focus 416 markets 285 Marsden, Peter V. 6, 22, 23, 38, 119, 181, 187,

195, 196,201,207,213,239,299,331,377, 378, 501, 511, 516, 518, 520, 522, 523, 524, 528,531

Mathew Effect 466 matrix hoppers 290 maximum likelihood

procedures 123 techniques 122

measure of standing 140 measurement 484 measurements of performance 253 measures

ofeSe 118 of performance 246

membership 392 mental health

agencies 135, 141, 145 sector 138,141,142,144

mentor 335 mentoring relationships 161-179, 256 metaphor 74, 86 Microsoft 49, 50, 67 minders 243 minority equity investment 29 mixed

coupling 459 ego-network coupling 453 mode of management 287

mobility 205 mode of acquisition 220 model

of independence 126 of quasi-symmetry 123

models of ese 118 modular cooperation 358 moral

borderlands 79 resources 19

morality 344 multi-group membership 47 multilevel

embeddedness 238, 260 social capital network 20

multinational subsidiary boundaries 161, 162, 176

multinationals 167, 168, 178, 306, 362 multi-participant alliances 28 multiple

cycles 259 levels 195 linear regression (MLR) 169

multiplex boundary system 52 combinations of dyadic ties 251 exchange system 238 fonns of organizational boundaries 49 generalized exchange system 240, 256-

258, 260,262 links 49, 57 networks relations 42, 46 ties 450, 451, 456 web of network connections 50

multiplexity 450, 454, 458, 486 multiplicative models 122 multi-site organization 195 mutual

N

competition pattern 152, 155 cooperation pattern 152, 155 cowork tie 253

national subsidiaries 167 natural paradigm 11 nature

of organizations 44 of relationships 448 of social capital 97

negative asymmetry 325, 332, 336 relationships 324-326, 328, 329, 331, 332,

336,337,434 Netherlands 198,342, 353,361,362 network 167, 198, 222,253,269, 299,341,352,

365, 374,39~8, 453

activity 400, 405 actors 18 alters 22 based management 285, 288, 289, 294, 296, 297 organizations 284, 293, 295 centrality 41,391,394,395, 402 concepts 17 configuration 98 cons~nt311,313, 315,316 coupling 449, 451 , 452, 455, 458, 459 data 237 design 103 diversity 403 dynamics 17,39 embeddedness 64 formation 18

fonns 25 fundamentals 21

Index - 553

generated (corporate) social capital 134, 136

hierarchy 311, 313, 315, 316, 318 in health care management 284, 295 level 446, 451 of coworkers 249 of lending relationships 458 of partial cooperation 345 of social relations 461 opportunities 169 organization 39, 338, 358 organizational design 98 patterns 32 portfolio diversity 400 relational contents 23 relations 356, 358, 400 size 451 structure viii, 5, 248, 304, 318, 352, 366,

427,446,452,484 sustainability 293 ties 120, 402,432 volume 25

networking 334 abilities 39 phenomenon 289, 290

networks 202, 246, 256, 285, 293-295, 298, 364 of civic engagement 77

new business organizations 463 ventures 45

niches 146 Nissan 30, 420, 423 Nohria, Nitin 6, 9, 14, 19, 97, 102, 103, 162,

167, 169, 209, 285, 286, 289, 300, 306, 319, 336, 338, 366, 489, 496, 497, 500, 510, 517, 518,519,527,529,531

nominalist strategy 22 non-consolidated network 459 non-equity partnerships 28 nonfriendly relationships 416 nonprofit sector 195 nonredundant contacts 391 nonsupervisory mentoring relationships 177 nontradability 61 Nooteboom, Bart 24, 57, 78, 87, 91, 92, 305,

342, 343, 345, 347, 350, 351, 352, 355, 366, 373, 377,432, 451,465,494

nonns 69, 77, 78,100 of generalized reciprocity 77

NOS (National Organizations Study) 187, 188, 194, 195

notion of social capital 72 number

of players in the supply network 422 of ties 118, 132, 136,454


o Obstfeld, David 92, 94, 493, 528 occupational categories 195 OECD countries 284, 286 old boy network 52 Omta,OMo 14,49,99,364,392,494,528,540 open system paradigm 11 openness 100 opportunities for access 377 opportunity 165, 299

costs 185, 325 optimal

embeddedness strategy 111 retaliation pattern 153

organization 's network of ties 452 stratification 48 -to-organization ties 392

organizational action 299 benefits 391 boundaries 43, 44, 461 career 166, 174, 176 competencies 344 employment policies and practices 180 goals 10 level of analysis 46 networks 17 performance 318 resource 44 role 168,374 standing 134-147

organizations 91, 176, 241, 306, 355 origins of social capital 17 output functions 270 outsourcing 367, 369, 413 overlapping membership 47 Owen-Smith, Iason 7, 8, 32, SO, 99, 132, 285,

293,378,382,395,397,413,531

p pacing technologies 368 panel-regression model 400 paradigms 11 paradox of embeddedness 351 partner selection 360, 363, 369 partnering

decisions 369 matrix 367

partners 244, 247 partnership

agreement 242 management 356,371 sourcing 414

patent activity 403 patenting 390-408

patents 390, 391, 393, 395-399, 402, 404, 405 patb-dependence 79 Pennings, JohaMes M. 6,7,9, 14,30,47,50,51,

63,75,77,87,132,215,285,347,408,529 perceived

constraints 169 similarity 327

perception of performance 415-417 performance 238, 239, 318, 319, 334, 415

assessment 374 of corrective actions 360, 364

performance benefits 36 implications 43, 53 measures 241, 242, 246

peripheral backgrounds 271 , 280, 281 functions 270

person based networks 45 personal

forms of organizational boundaries 49 goals 19 inclusion 48 networks 39 trust 52

personality characteristics 162 personalized ties 245 personalizing work ties 245 personnel department 195 pharmaceutical firms 357 philos relationships 81 physical

capital 72,84, 177 position in the supply network 422

piecewise exponential model 469, 472 pipeline 416 pitfalls 362, 394 player 422,441,490 Podolny, Joel M. 37, 59, 87, 94, 108, 110, 111,

114, 119, 128, 130, 131, 132, 137, 147, 181, 185, 215, 299, 300, 305, 318, 319, 321, 377, 378,379, 382,383,389,394,399,419, 432. 433, 434, 435, 444, 447, 448, 449, 458, 465, 466,468,509,530

political actor 44. 45 corporate capital 107, 110-113, 116 embeddedness 71 power 41 relations 40 social capital 106

poor customer service 149 popularity 125 position 27,119 positional advantage 341 poverty sector 135, 139, 141-145

Powell, WaIter W. 7, 8, 28, 32, 36, 49, 50, 57, 97,99,100, III, 132,215,285,293,294,295, 300, 377, 378, 382, 391, 392, 394, 395, 397, 399, 400,402, 405, 406, 413, 447, 448, 453, 504,507,518,519,524,530,531,534

power6,73,85,324,330,424 in the supply network 422 resources 40 structures 6

prejudice 183, 186 presence of ties 92, 103 prestige 125,376,387-389 primary contribution 141, 142 prism 431-445 prismatic perspective 432, 444 problem of order 120 process 359 product quality 113 production capacity 113 professional

elites 217, 220, 232 identity 176 networks 287 recognition 262 reputation 176 services firms 54

profitability 274 project teams 307, 313 promotion 39, 180-182, 184, 194, 199,324

practices 180 systems 181

property 390 protection 163 protege 161-166, 168, 176-178 proximity 118, 122, 125, 126, 128, 131-133 psychological help 163 public

goods 80 sector 195

public service organizations 284-297 settings 286

pyramidal vehicle networks 423

Q quality 59, 149,454

of a relationship 446 quasi-symmetry models 122, 125, 126, 128

R random

data patterns 157 interactive behaviors 156

random-competitive data 156 randomly generated pattern 156

rational actors 43, 45 paradigm 11

rationality pressures 186 RC(2) model 126, 128 realist approach 22

Index - 555

reciprocity 77-79, 83, 87, 100, 125, 290, 302, 303,317,327

recognition 23 recruitment procedures 182,214,331 redundant

market areas 110 ties 47

referrals 181, 183,201 regular group 47, 48 relational

competence 341, 344 dimension of social capital 90 inertia 303, 317 proximity 131 ties 44

relations 25 relationship

between strategy and cultural context 99 status and volume 130

duration 449, 456 formation 377 interfaces 458 multiplexity 446, 449 to actors 80

relationships viii, 50, 219, 221, 229, 300, 301, 319,351,399,409,414,446,447,450,490

between organizations 465 reliability 59 reputation 73, 164, 479 reputational status 136 research and development 395, 399, 400, 487

collaborations 356-375, 391 managers 357, 361, 370 partnership 51, 52, 57, 362-365 relationships 356, 358, 365, 366, 370 ties 402, 403, 405

resource 59, 68, 78, 119, 161, 256, 258, 299, 301, 341, 343, 344, 376, 409, 410, 432, 446, 460,465

allocation decisions 431, 433, 435 allocations 444 dependence 31, 32

theory 308, 309, 452 flows 432 interdependencies 31 value 61

resource-flow network 168 restraint 261 restricted access 377 retaliation 155,353

dyadicinteraction 157


revelation problem 341, 349, 354 rewards 180 risk sharing 58 role

s

modeling 163 of networks 431 of go-between 341 of trade 342

Sandefur, Rebecca L. 14, 219, 224, 228, 232, 233,533

satisfaction 415, 416 scale advantages 57 scarcity 61, 78, 79 search costs 202 secondary contribution 141 selection

methods 182,331 of ties 97

self-entrapment 262 semiconductor industry 380 seniority 183,239,247,249,250

lockstep system 243 services industries 148 sex-specific division 199 short-term satisfaction 415 should expectations 415 sidekick effect 161-179, 336 Silicon Valley 29, 82, 83,99,357,369,463,471 ,

473,475 Simmelian problem of order 120, 122 simplex

boundary system 52 web of network connections 50

size 195,274,326,400,402,454,456 in the supply network 422

small business entrepreneurs 449 businesses 8, 446, 454-456, 458, 459 companies 358 firm 45, 357 networks 29 group dynamics 65 towns 19

Smith-Doerr, Laurel 7, 8, 32, SO, 99, 132, 285, 293, 377, 378, 382, 391, 392, 394, 397, 399, 400,402,405,406,413,447,453,518,519, 531,535

social actors 18,91,93 capital I, 17, 20, 36, 44, 72-76, 89, 90,

162, 163, 203, 205, 210-212, 217-233, 251,262,284-297,300-302,324,459

creation 285 explanation of problems 364 management 491-493

metaphor 86 of ties 220 theory 202, 358

closure 200 clubs 392 cohesiveness 88, 89, 92-94,103 connections 19 constraints 78, 365 contagion 6 debts 82 domain 68 embeddedness 237-265 entrepreneurship 88, 90 exchange 416,426

system 261 groups 20 liability 3,5, 10, 11,31,42,59, 150, 186,

291, 298, 318, 323-338, 358, 374, 409, 410, 416,417,421,423, 449, 458, 479, 483,489

mobility 461 nature 460 network

analysis 459 theory 448

networks 163, 185, 194, 197, 198, 200, 208, 211, 215, 238, 249, 284-297, 299, 302,391

organization 222, 223, 230 power 40 psychological explanation of trust 33 relationships 108,209,329,330,334 resources 20, 240

management 323-338 rules 69 service organizations 134, 135 services 146 solidarity 69, 219 status 198 structure 3, 68, 217, 237-239, 249, 301,

304,305,447,448,478,485,489,490 systems 19,218,221 ties 2, 3,6,212,219,229, 241,281

socialization 332 societal sector 135 socioeconomic status 162 solidarity 69, 87 sourcing 417, 419 Spain 198,416 specific reciprocity 77 spill-over control 350, 352-354 sponsorship 163, 358 spread ofsocial capital 18 staffing methods 182-184 status 7, 118, 122, 125, 128, 131, 132, 162,249-

251,262,431-445 homophily 131

stimulation 262

stock of social capital 161, 163, 166, 168, 178 strategic

action 267 alliance network 377 alliances 37, 49, 97, 356, 374, 377 license alliances 378 models 424

strategies 88-105 strong tie 47, 110,248,256,259,318,319,331,

332 strategy 335 trust relations 35

structural conditions 98 dimension of social capital 90, 98 embeddedness 71, 446, 448, 449, 451 equivalence 27 hole theory 304, 317 holes 76, 88-92, 98, 102, 103, 110, 119,

305,306,310,317,318,335,391 structure

of advantage 1 of disadvantage 1 of institutional context 97 of social capital 92, 103 of the network 304, 364 of ties 119

structures 88-105 Stuart, Toby E. 8, 25, 32, 50, 99, 130, 132,378,

379, 382, 383, 389, 399, 408, 432, 433, 444, 464,468,494,530, 537

subnetworks 177,409,410 subordinates 183 subsidiary 168, 169, 174,178 success

of a collaboration 359 of an entrepreneurial strategy 97

supervisors 183, 334 supervisory mentoring relationships 177 supplier-customer partnerships 356, 409 supplier-manufacturer networks 453 supply 410, 411 , 413, 416, 418

base reduction 421 chain 416-418

management 412 focus 416 network 424

breadth 420 strategy 409-427 structure 418, 419

supply networks 410, 418, 421 , 423 support 24 surplus consultation 308 Sweden 200, 362 symbolic capital 73, 74 syndicate relations 432 syndicates 120 synergy 414

Index - 557

system boundaries 22

T talking shops 291 Talmud, Dan 107, 108, 109, 110, Ill , 112, 113,

114, 115, 116, 117, 501 , 515,537 task

environment 299, 306, 318 interdependence 309-310, 312, 313

teamwork 100,414 technological

collaboration 358-360, 362 prestige 376-389 spillover 395

technology management processes 374 networks 357

temporary or marginal category 48 workers 47

tenure 281 tenured partners 246 tertius gaudens 91, 93-95, 98, 99,102 theory

of corporate social capital 483 of knowledge 345

third parties 97 third-party enforcement 69 Third World 79 throughput fUnctions 270 ties 50, 128, 136, 146, 150, 155, 161-163, 166,

167, 169, 176, 179, 182,204, 217, 221, 222, 225,227, 231 , 238,240, 303

of firms 280 time 489

span in research procedures 25 Tobit regression model 456 top management team (TMn 268 total

network 22 quality management (TQM) 414

Toyota 7, 29, 30, 57, 58, 64, 369, 409, 420, 421, 423,426

tradeability 80 trade

association 392 expansion 70

trading partners 457 training 195, 332, 333

programs 176 transaction

cost economics (TeE) 342, 343, 345, 347,

348 theory 201,345, 452

costs 30, 238 interfaces 458


transactional economics 411 transactions 23, 450 transfer 180-184,194 transfers of resources 260 transfonnation of social capital 17 triads census 100 trilateral governance 341, 343, 348, 353, 354 triplex reciprocated ties 254, 255, 260 trust 24, 33-37,41, 52, 70, 78, 79, 81, 93, 100,

115, 290, 346, 350, 353, 354, 371, 373, 374, 414,416,465,479

trust-violation 36 turnover 336

U U.K. 284,287,288,362,416 uncertainty 274 understandability 85, 290 union

presence 195 strategy 88, 89, 94-97, 99, 100, 102-104,

493 universalism 183 university-industry collaboration 364 upstream partners 357 U.S. 71, 80, 119, 178, 180, 183, 188, 193, 198,

218, 237, 272, 273, 357, 362, 369, 382, 383, 392-394,396,419,432,452

Uzzi, Brian 8, 36, 45,59,65,66,67,71,76, 81, 87,95,96, 107, 108, 109, 111, 116,240,301, 302, 334, 342, 343, 347, 350, 351, 418, 435, 447, 448, 449, 450, 451, 452, 453, 454, 455, 456,459,539

v valence 220 valuation 78 value 78, 79, 149,217,300,359,379,394,396,

399,409,417,427,450,458 adding process 409 of conduct 344 of elite ties 222, 223, 228 of status 443

Van Rossum, Wouter 49,99,357, 364, 392, 494, 501

venture capital 436, 460-479 finn 438, 440, 442, 461-479 markets 431-445

versatility 83 virtual

corporation 358 organization 358

visibility 163-165 volume 118, 122, 128, 132,458

w weak tie 47, 110,203,208,213,331,335

strategy 336 web 358 will expectations 415 women 199,327 work

assignments 163 experience 164

work-related perceptions 41

CONTRIBUTORS

• Luis Araujo Senior Lecturer in Marketing, The Management School, University of Lancaster, Great Britain.

Wayne E. Baker Associate Professor of Organizational Behavior and Human Resource Management, Business School, and Faculty Associate, Institute for Social Research, University of Michigan.

Mario Benassi Assistant Professor, Department of Computer and Management Sciences, University of Trento, Italy.

Ed Boxman Consultant Operational Auditing, Dutch Post Office Organization Postkantoren BV., The Hague, The Netherlands.

Daniel J. Brass Professor of Organizational Behavior, Smeal College of Business Administration, The Pennsylvania State University.

Ronald L. Breiger Goldwin Smith Professor of Sociology, Department of Sociology, Cornell University.

Fabrizio Castellucci Ph.D. Candidate, Graduate School of Business, Stanford University.

Alison Davis-Blake Associate Professor of Management, Graduate School of Business, University of Texas-Austin.

Patrick Doreian Professor of Sociology, Department of Sociology, University of Pittsburgh.


Geoff Easton Professor of Marketing, The Management School, University of Lancaster, Great Britain.

Ewan Ferlie Professor of Public Services Management, The Management School, Imperial Management School, Great Britain.

HenkFlap Associate Professor of Sociology, Department of Sociology and Interuniversity Center for Social Science Research and Methodology, University of Utrecht, The Netherlands.

John Freeman Helzel Professor in the Haas School of Business, University of California, Berkeley

Martin Gargiulo Assistant Professor of Organizational Behavior, INSEAD, France.

James J. Gillespie Ph.D. Candidate, Department of Organization Behavior, J.L. Kellogg Graduate School of Management, Northwestern University.

Elizabeth H. Gorman Ph.D. Candidate, Department of Sociology, Harvard University.

Shin-Kap Han Assistant Professor of Sociology, Department of Sociology, Cornell University.

Christine M. Harland Senior Research Fellow, Centre for Research in Strategic Purchasing and Supply (CRiSPS), School of Management, University of Bath, Great Britain.

Pamela R. Hauoschild Associate Professor of Organizational Behavior, Graduate School of Business, Stanford University.

John P. Heinz Owen L. Coon Professor of Law, School of Law, Northwestern University, and Distinguished Research Fellow, American Bar Foundation.

Andrew D. Henderson Assistant Professor of Management, Graduate School of Business, Columbia University.

Contributors - 561

Monica C. Higgins Assistant Professor of Organizational Behavior, Harvard Business School, Harvard University.

Dawn Iacobucci Professor of Marketing, J.L. Kellogg Graduate School of Management, Northwestern University

David Knoke Professor of Sociology, Department of Sociology, University of Minnesota.

Kenneth W. Koput Associate Professor, Department of Management and Policy, University of Arizona.

Giuseppe Labianca Assistant Professor of Organizational Behavior, A.B. Freeman School of Business, Tulane University.

Edward O. Laumann George Herbert Mead Distinguished Service Professor of Sociology, The University of Chicago

Emmanuel Lazega Associate Professor of Sociology, University of Versailles and LASMAS-CNRS, France.

Kyungmook Lee Assistant Professor of Management, College of Business Administration, Seoul National University, Korea.

Peter V. Marsden Professor of Sociology, Department of Sociology, Harvard University.

Nitin Nohria Professor of Organizational Behavior, Harvard Business School, Harvard University.

Bart Nooteboom Professor of Industrial Organization, School of Management and Organization, University of Groningen, The Netherlands.

David Obstfeld Doctoral Candidate, Organizational Behavior and Human Resource Management, Business School, University of Michigan.


Onno (S.W.F.) Omta Associate Professor of Innovation and Business Development, School of Management and Organization, University Groningen, The Netherlands.

Jason Owen-Smith Ph.D. Candidate, Department of Sociology, University of Arizona.

Johannes M. Pennings Professor of Management, The Wharton School, University of Pennsylvania.

Joel M. Podolny Associate Professor of Strategic Management and Organizational Behavior, Graduate School of Business, Stanford University.

Walter W. Powell Professor of Sociology, Department of Sociology, University of Arizona.

Wouter van Rossum Professor of Commercial Management, Marketing, and Innovation Management, University of Twente, The Netherlands.

Rebecca L. Sandefur Ph.D. Candidate, Department of Sociology, University of Chicago.

Laurel Smith-Doerr Ph.D. Candidate, Department of Sociology, University of Arizona.

Toby E. Stuart Assistant Professor of Organizations and Strategy, Graduate School of Business, University of Chicago.

Dan Talmud Lecturer, Graduate Program in Applied and Organizational Sociology, Department of Sociology and Anthropology, University of Haifa, Israel.

Brian Uzzi Associate Professor of Organization Behavior and Sociology, J.L. Kellogg Graduate School of Management, Northwestern University.

THE EDITORS

Shaul M. Gabbay Lecturer. Davidson Faculty of Industrial Engineering and Management. Technion. Israel Institute of Technology. Israel.

Roger Th.A.J. Leenders Assistant Professor of Business Development and Business Research Methods. School of Management and Organization. University Groningen. The Netherlands.