jason owen-smith and walter w. powell -...

Networks and Institutions

Jason Owen-Smith and Walter W. Powell

25

INTRODUCTION

Research on institutions and networks hasproceeded on largely separate trajectoriesover the past few decades. The former ismore associated with work in organizationaland political sociology, and the latter servesas the wellspring of research in economicsociology. To be sure, a number of loose link-ages exist between the subfields. For exam-ple, many institutional studies presume thatprofessional or inter-organizational networksserve as conduits for the diffusion of appro-priate practices and ideas. Indeed, muchinstitutional research conflates ‘simple’ dif-fusion with ‘deep’ institutionalization.Meanwhile, research on networks often con-siders how categorical or status variations innetwork structures shape social comparisonand stratification processes. But these pointsof intellectual cross-fertilization haveremained undertheorized.1

We think there is much to be gained froma more analytically driven dialogue betweenthese literatures. We argue that networks andinstitutions mutually shape one another. Overtime, this co-evolutionary process creates,

sustains, and transforms social worlds. The cognitive categories, conventions, rules, expectations, and logics that give insti-tutions their force also condition the forma-tion of relationships and thus the networkstructures that function as the skeletons offields. But networks are more than just thescaffolds and circulatory systems of organi-zational fields. They are also the source of‘horizontal’ distinctions among categories ofindividuals, organizations, and actions, aswell as ‘vertical’ status differentials. Whileinstitutions shape structures and conditiontheir effects, networks generate the cate-gories and hierarchies that help define insti-tutions and contribute to their efficacy. Thus,any effort to understand institutionalprocesses must take networks into account,and vice versa.

Our argument draws on core conceptsfrom institutional and network theory that areutilized across both lines of work. Despitetheir portability, however, they have rarelybeen theoretically integrated. Using institu-tional theory, we highlight the under-theorized relational aspects of both fields andlogics. Starting in network theory, we

9781412931236-Ch25 12/1/07 4:35 PM Page 594

emphasize the important institutional fea-tures of embeddedness and social capital.Our goal is to do more than review existingpoints of contact or stimulate joint discus-sion. We aim to provide a roadmap for futureresearch that will directly address two criticalanimating questions. First, how do institu-tional practices and forms emerge from networks? Second, how do institutionalizedcategories and conventions shape the structure and effects of networks?

We begin with a brief excursion throughseveral canonical works in institutionalanalysis, highlighting the implicit, butnonetheless strong, network underpinningsof these theoretical arguments. For symme-try, we select several well-known empiricalstudies that directly measure network effectsto account for the transmission of institu-tional practices and structures. Next werevisit four foundational ideas – organiza-tional field, institutional logic, embedded-ness or the non-contractual basis of contract,and social capital – which contain both network and institutional insights that are, we contend, indissoluble. We thendevelop answers to our key questions about emergence and constraint against thisbackground.

NETWORKS ARE CARRIERS OFINSTITUTIONAL EFFECTS

In their classic paper, Meyer and Rowan(1977) observed that the formal structures oforganizations ‘dramatically reflect the mythsof their institutional environments.’ Theyargued that organizations are driven to incor-porate practices and procedures defined andbuttressed by widely prevalent, rationalizedconcepts in the larger society. These prac-tices were institutionalized through profes-sional standards and prestige hierarchies, andreinforced by public opinion. Consequently,Meyer and Rowan contended that the build-ing blocks of formal organization ‘litter thesocietal landscape.’

This canonical article incorporated net-work ideas in several key ways, althoughsubsequent work has tended to overlook itsstructural aspects. In part, this neglect maybe traced to a contrast that Meyer and Rowanemphasized between organizations wheresurvival depended on managing the contin-gencies of boundary-spanning relations andothers that had to respond to ceremonialdemands which were present in their environments. This continuum suggested thatmanaging relational networks involved mat-ters of coordination and control, while moreinstitutionalized settings necessitated effortsat symbolic management. But Meyer andRowan also emphasized that all organiza-tions are embedded in both relational andinstitutionalized contexts. They stressed thatthe complexity of relational networks gener-ated ‘explosive organizing potential,’ and thisgreatly increased both the spread and numberof rationalized myths. Central to this processof transmission and standardization weretrade and professional associations and inter-organizational coalitions.

The generative potential of networks astransmission channels is readily apparent inthe Meyer and Rowan paper. Similarly,DiMaggio and Powell (1983) argued that thegreat rationalizers of the latter half of thetwentieth century were the professions andthe modern State. The growth and elabora-tion of professional networks spanningorganizations contributed, they argued, to therapid spread of various models of organizing.Networks were also essential components ofDiMaggio and Powell’s conception of anorganizational field, which emphasized bothconnectedness (Lauman, Galaskiewicz, andMarsden, 1978) and structural equivalence(White, Boorman, and Breiger, 1976). Theinstitutional development of an organiza-tional field hinged on: (1) increased interac-tion among participants; (2) the developmentof well-defined status orders and patterns ofcoalition; (3) heightened information shar-ing; and (4) mutual awareness and respon-siveness. The twin imprints of the relationalsociologies of Harrison White and Pierre

NETWORKS AND INSTITUTIONS 595

9781412931236-Ch25 12/1/07 4:35 PM Page 595

Bourdieu clearly stamp this account of fieldevolution and institutional formation.

From DiMaggio and Powell’s perspective,status orders shaped patterns of informationexchange, creating a core and peripherystructure that channeled the flow of news andpersonnel within organizational fields. Thepolicies and structures of the most centralorganizations in a field were more likely tobe emulated by others. While many subse-quent researchers picked up on the mimeticaspects of this phenomenon, the underlyingstructural elements received less attention(see discussion in Mizruchi and Fein, 1999).Nevertheless, this account of field structura-tion emphasized how shared meanings andtypifications, as well as stable role structures,emerged out of repeated interaction.

In one of the most comprehensive empiri-cal studies of institutional transformation,Scott and colleagues (2000) analyzed theprofound changes that occurred in healthcare delivery in the San Francisco Bay Areabetween 1945 and 1990. They demonstratethe effects differing forms of legitimacy haveon hospital survival rates. In the period fol-lowing World War II, physicians and theirprofessional code of conduct dominatedhealth care standards. Federal financing andthe regulation of health care arose in the late1960s and greatly expanded in the 1970s and1980s. That growth was accompanied byincreasingly salient technical forms of legiti-macy. In recent decades, the health careindustry became more intensely competitive.For-profit entities entered the field in largenumbers, and managerial legitimacy increas-ingly shaped evaluative standards. Scott andcolleagues’ rich analysis documented thatearlier professional and regulatory standardswere not extinguished by the new managerialand market orientations; rather, each succes-sive era displayed more heterogeneus formsof legitimacy.

Consequently, as the health care fieldevolved, ‘three logics – professional, public,corporate – were all present, active and contending with one another’ (Scott et al.,2000: 316). Federal funding and oversight of

health care eroded professional sovereignty,opening the door for more market-based cri-teria. The key point, however, is not that anew managerial logic replaced physicians orbureaucrats, but that health care became acomplex, multi-level field in which both thenumber and novelty of inter-organizationalconnections between hospitals and othertypes of health care institutions expandeddramatically. ‘Managers appear to have beenthe beneficiaries, not the agents’ of deinstitu-tionalized professional power (Scott et al., 2000: 328).

The forces that transformed the healthcare field were varied and numerous, rangingfrom policy legislation to medical specializa-tion to the increasing complexity of servicedelivery. These broad changes were typicallyushered in by new linkages, formed byaccreditation bodies, shifting organiza-tional control structures, inter-organizationalalliances and coalitions, as well as new affil-iations with purchasers, intermediaries, serv-ice providers, and government. Thesenetwork realignments not only brought withthem participants who changed the bound-aries of the health care field, but the newentrants were also carriers of novel ideas thatprofoundly altered the meaning of healthcare. Scott et al. (2000) captured this changein relationships and meanings aptly in theirdiscussion of the shift from thedoctor–patient relationship to one of a healthcare provider–consumer transaction. Thisupsurge in linkages and connections werecritical to accreditation, health care provi-sion, and fiscal solvency; but these new rela-tionships also remade the taken-for-grantedunderstandings of the medical field.

Each of these pillars of sociological insti-tutionalism argues that social networks trans-mit ideas and practices in distinctive ways.Networks also reflect key micro-level inter-actions that influence institutional dynamics.To illuminate the recursive nature of institu-tional and network influences, we turn toseveral notable empirical studies that demon-strate the potent force of social networks.Though the orientations of the authors differ

596 THE SAGE HANDBOOK OF ORGANIZATIONAL INSTITUTIONALISM

9781412931236-Ch25 12/1/07 4:35 PM Page 596

and the objects of inquiry vary, we argue thatthese prominent papers carry a common mes-sage that networks are shaped by social com-parison processes in which institutionalizedcategories are highly influential.

NETWORKS ARE STAMPED BYINSTITUTIONAL CATEGORIES

Galaskiewicz and Burt (1991) analyze howcorporate officers evaluated nonprofit organ-izations in the Minneapolis-St. Paul regionand decided whether to make significantcharitable contributions to them. In the TwinCities, the networks of high-profile corporatephilanthropists and leaders of the nonprofitcommunity were closely inter-connected.The authors tested to see whether the spreadof evaluative standards operated through themechanisms of cohesion or through struc-tural equivalence. The frame of reference inthe former is the dyad, stemming from a his-tory of past experiences, while the latter isthe larger social system. Dyadic influenceprocesses operate on a one-to-one basis,while structural equivalence effects draw onperceptions of similarity rather than directcommunication. Consequently, structuralequivalence processes are driven by whatofficers presume others in comparable posi-tions are doing.

In a community that is closely knit and inregular contact, one might expect directinteraction and cohesion to trump structuralequivalence. Instead, Galaskiewicz and Burtfind the opposite. In the Twin Cities, theimportance of structural equivalencereflected common norms and standards mag-nified within a professional communitywhile demonstrating how the field’s informalstratification orders conditioned individualacceptance of these norms. As the authorsput it, when an opinion comes to be sharedwithin ego’s profession, ‘ego is expected tofollow rapidly to avoid the embarrassment ofbeing the last to espouse a belief that hasbecome a recognized feature of occupying

his or her position in the contributions com-munity’ (Galaskiewicz and Burt, 1991: 90).

Galaskiewicz and Burt recognize thestrong parallels between their account of theTwin Cities nonprofits community andDiMaggio and Powell’s (1983) explanationfor how and why organizations, and thestructure of organizational fields, changeover time. Indeed, they explicitly note that‘an important component of DiMaggio andPowell’s argument is the network of contactsamong organizations or their agents’(Galaskiewicz and Burt, 1991: 88). Withinthis philanthropic community, the manner inwhich evaluative categories varied acrossorganizations could be predicted by how corporate contributions offices were strati-fied within the status hierarchy of their profession.

Palmer, Jennings, and Zhou (1993) ana-lyzed the influence of institutional, political,and economic factors on the adoption of themultidivisional form (MDF) by large U.S.corporations in the 1960s. To assess howinstitutional factors influenced the transitionto an MDF, they measured the professionaltraining and social network connections ofkey organizational decision makers, focusingspecifically on elite business school trainingof corporate chief executive officers andinterlocks among corporate boards of direc-tors. They also assessed economic and polit-ical considerations, including corporatestrategy and performance, as well as theinfluence of managerial rivalries, both insidecompanies and in external coalitions.

This impressive effort to test rival theoret-ical arguments found ample support for botheconomic and institutional factors, but littlefor an explicitly political view. Both corpo-rate industrial diversity and geographic dis-persion stimulated the adoption of themultidivisional structure. Differentiatedcompanies (i.e., firms involved in multipleunrelated lines of activity) and those withfacilities spread across the nation, encounterproblems that a multidivisional structure pur-ports to solve. Institutional variables alsoproved to be robust. Most notably for our


9781412931236-Ch25 12/1/07 4:35 PM Page 597

purposes, networks were critical factors inthe transition to a multidivisional form.Corporations in which CEOs had graduatedegrees from elite business schools weremore likely to adopt the MDF than firms withexecutives who did not hold elite degrees.Boards of directors with interlock ties tofirms that had already adopted an MDFstructure also influenced adoption. Not sur-prisingly, corporate board connections tonon-MDF firms did not (Palmer et al., 1993:120). Thus, institutional backgrounds andsocial connections jointly condition corpo-rate strategies.

Davis and Greve (1997) analyzed the dif-fusion of two practices that were adopted bycorporations in the 1980s as a defenseagainst hostile takeovers. The ‘poison pill’and the ‘golden parachute’ were embraced bycompanies and their managers to raise thecosts of an unwanted takeover bid. Both of these practices were initially controversialbut came to be adopted by the majority of U.S. corporations. Yet despite their simi-larities, the ‘pill’ spread quickly and the‘parachute’ diffused more slowly.Interestingly, the channels of social influencevaried too. Pills spread through cohesive tiesamong members of corporate boards ofdirectors, while parachutes were adopted on the basis of geographic proximity.Corporations adopted golden parachutes asother firms in their local metropolitan areadid so.

This intriguing analysis revealed a puzzle:the same individuals – members of corporateboards – decided to adopt both practices, butthe tools spread at different speeds throughdivergent routes. Davis and Greve ask whatfactors accounted for these different patternsof diffusion. Social networks provide onecompelling answer. Pills spread from onecorporation to another across the nationbecause the corporate director network has anational reach. Boards that shared directorswere the conduits through which this mecha-nism to deter hostile raiders spread. Goldenparachutes, in contrast, circulated locally.Their diffusion was rapid in some areas, but

in other regions the practice never took hold.In the status-bound corporate world of NewYork City, for example, protecting the CEOwas seen as a duty of boards, but in the morerough-and-tumble entrepreneurial world ofSilicon Valley, parachutes were eschewed.The mechanism at work here was socialcomparison among local elites, who lookedto their regional reference groups for a signof whether their CEO should be protectedagainst unexpected job loss. This ‘parochial’social comparison process resulted in aslower rate of diffusion than the morenational and cosmopolitan transmission ofpills through director networks.

Davis and Greve point out that these localand national network channels were alsocharacterized by different normative stan-dards. The poison pill was couched in a lan-guage of fending off unscrupulous raiders.This defense was perceived as appropriateand legitimate by board members; and thuschampioned by them in different corporatesettings. Contact with directors in similarindustry sectors and in corporations of com-parable status became the venues for diffu-sion of a practice that came to be regarded asaccepted and necessary.

Parachutes were much more difficult tolegitimate. They were perceived by some toreflect naked managerial self-interest, whileothers saw them as payoffs for weak man-agers. Questions about the appropriateness ofparachutes were answered locally, by look-ing to the behavior of central individuals andcompanies in the regional economy. In short,the relevant networks for diffusion and thepace at which they communicated specificpractices were shaped by the broader institu-tional context in which they were situated.Davis and Greve’s analysis affords keeninsight into how network configurations are conditioned by institutional forces.Understanding why the same boards take sig-nals about adopting pills from distant, butconnected rivals while turning to local com-munity members for signals about the legiti-macy of parachutes requires not onlyattention to networks, but to the meaning of


9781412931236-Ch25 12/1/07 4:35 PM 98

practices, categorical distinctions and statushierarchies.

We find strong common analytical under-pinnings in these different, notable accountsof institutional influences and networkeffects. Numerous scholars identify networksas the channels through which institutionaleffects flow, and see networks of like-mindedindividuals as central reference groups thatpromote widely emulated practices. Thepresence of these common elements in ahandful of important empirical studies lendscredence to our claim that networks andinstitutions mutually influence one another.To pursue this argument, we move fromempirical studies to conceptual claims, andoffer a brief exegesis of four core ideas thatare widely used in both network and institu-tional analyses. We demonstrate that coreconcepts in institutional and network theoryare analytically richer and more useful whenthey take each other into account.

FOUNDATIONAL CONCEPTS

Organizational fields

Fields are simultaneously master conceptsand fundamental empirical sites for institu-tional analysis. Much recent (and some notso recent) work has focused on genesis andchange in such diverse fields as politics, thearts, law, gastronomy, and the chemicalindustry (Clemens, 1997; DiMaggio, 1991;Dezalay and Garth, 1996; Ferguson, 1998;Hoffman, 2001; Rao, Monin, and Durand,2003). These rich narrative efforts, as well asmore abstract treatments, rely heavily onrelational language to describe the contoursand characteristics of fields. Perhaps moreimportantly, organizational action withinfields is understood largely in terms of affili-ation, competition, and shared membership,all features that emphasize how social rela-tions shape institutions (Fourcade 2007).

Consider two related definitions of fields. In one view, an organizational field is

a community of organizations that engage incommon activities and are subject to similarreputational and regulatory pressures(DiMaggio and Powell, 1983; Scott et al.,2000). Through a more politically filteredlens, a field is seen as a space of positionswhose characteristics are jointly defined bythe configuration of their inter-relationshipsand by the struggles of actors who seek toclaim them (Bourdieu and Wacquant, 1992).These two definitions bracket contemporaryinstitutional parlance, where ‘field’ carriesthree distinct, but only partially decompos-able connotations. We highlight each senseof the term, and then argue that their indissol-ubility results from the relational threads thatcross-cut them. Networks both structure andintegrate fields.

The fields of institutional theory are recog-nizable arenas of social action, as such theyare fields of endeavor. A more dynamic viewsuggests differentiated fields of play wheremore or less attractive positions conveyopportunity, and constrict the possibilities ofvarious social groups. Finally, fields aremolded into their characteristic shapes byrules, conventions and expectations thatdefine appropriate activities and legitimatepositions. Thus, this view emphasizes fieldsof force that regulate social action.

Relationships are moves in games and fun-damental components of the fields on whichthey are played. As such, concrete networkstructures map past struggles, and shape pos-sibilities for the future by differentially chan-neling resources to contestants. Networksalso push and pull players into finite sets ofpositions. Thus, the positions and affiliations participants claim are only partially undertheir control. The presence and absence of ties render a confusing struggle clear to observers and participants alike by allowing them to classify and order boththe players and their moves into categories(forms, identities, strategies) and hierarchies(status orders). A fundamental insight of network theory, often neglected in institu-tional analyses, is that relationships are bothpipes that channel resource flows as well as


9781412931236-Ch25 12/1/07 4:35 PM Page 599

prisms that help render action sensible(Podolny, 2001).

Networks are essential to fields in at leasttwo senses: they are both a circulatorysystem and a mechanism for sensemaking.Fields are shaped by networks, which condi-tion the formation of relationships and helpestablish their consequences. But it is onlyagainst the backdrops of particular fields thatrationalities and strategies of action are sen-sible. The relational aspects of fields are thethreads that weave together the term’s dis-parate meanings.

Institutional logics

Logics constitute the rules and conventionsof a particular organizational field. In broadterms, an institutional logic is the constella-tion of beliefs and associated practices (theschemas and scripts) that a field’s partici-pants hold in common. These packages ofbeliefs and practices are organizing princi-ples and recipes for action. They have instrumental, normative, and cognitive impli-cations (Friedland and Alford, 1991;Whitley, 1992; Thornton, 2004). Logics pro-vide rationales for action. They are mostinfluential when they are consistent andeasily taken-for-granted. But when multiplecompeting logics are in play in the same set-ting, they can trigger conflict and/or generatenew accounts of activity.

Three different approaches to the idea oflogics rely on relational underpinnings, buttheir structural features are rarely elaborated.Consider first the idea, drawn from work inthe Carnegie School tradition of organizationtheory, that organizational action is routine-based, rule-governed, and triggered by con-ventions that match concrete situations andactions to the needs of particular positions(Cohen, March and Olsen, 1972; March andOlsen, 1984). These logics of appropriate-ness do more than simply set the grounds forconcrete action in particular situations. Whenstrung together across roles, they representthe authority structure of an organization by

‘defining the relationships among roles interms of what the incumbent of one roleowes to the incumbents of other roles’(March and Olsen, 1984: 23). In contrast toclassic Weberian notions of authority, it is thelinkages among conventional recipes foraction that are central, defining characteris-tics of organizations. When spread across the categorical distinctions provided by orga-nizational roles, logics comprise formalstructures.

Clemens (1993, 1997) extends the idea of logics by recognizing that the social worldis rife with alternative models for organizingany particular endeavor. In her view, organizational repertoires are templates thatstructure concrete relationships within organ-izations and convey scripts for behavior thatlink forms of organization to cultural expectations (Clemens 1993: 758). In thissense, logics offer a mechanism by whichinstitutions direct the formation and mobilization of networks, while providing ameans for expectations and regulations toexert force upon the participants in a field. The analytic link she makes betweeninstitutional logic and organizational form is an important one that deserves furtherexplication.

Institutional logics, then, are inextricablytied to concrete structures that define theauthority relationships that characterize orga-nizational forms. Logics do more, however,than forge collections of roles into formalorganization. Friedland and Alford (1991)offered a now widely held view of logics ascentral, distinguishing features of fields. Intheir view, the content of a field’s dominantlogic renders networks much more than mereaffiliations. Without institutional logics, ‘itwill be impossible to explain what kinds ofsocial relationships have what kind of effecton the behavior of organizations and individ-uals’ (Friedland and Alford, 1991: 225).Logics make networks meaningful featuresof social and economic worlds precisely bydisciplining (though not determining) theformation and implications of relationships.The presence or absence of connections and


9781412931236-Ch25 12/1/07 4:35 PM Page 600

the resources that flow through them as wellas the meanings that participants andobservers attribute to relationships dependupon prevailing logics. The same relation-ship or affiliation may exist under the aus-pices of multiple institutional logics, whichprovides leverage to elaborate the ways inwhich relationships carry the social intoinstrumental exchanges.

Embeddedness: the non-contractual basis of contractClassic research in organizational theory(Dalton, 1959; Gouldner, 1954) and a foun-dational work in economic sociology(Macaulay, 1963) demonstrated that evenhighly purposive economic exchanges areenmeshed in and freighted with social expec-tations. Organizational and economic actionsresult from a complex lamination of motiva-tions and meanings that participants drawfrom the various fields in which they partici-pate. Macaulay’s (1963) key finding thatbusinessmen often disregard the legal rightsand responsibilities inherent in contract infavor of more social means of dealmakingand dispute resolution underscored howsocial relations cemented economic transac-tions. His study offered a starting point forGranovetter (1985) who, drawing onPolanyi’s (1957) insight that market relation-ships are embedded in both economic (con-tract) and non-economic (friendship, familial)institutions, illuminated how concrete socialrelationships shape economic activity.

Granovetter’s (1985: 500) argument thatsocial relationships are fundamental to eco-nomic processes has been highly influential.Nevertheless, we concur, to a degree, withcritics of the embeddedness perspective(Krippner, 2001; Lie, 1997) who argue that a purely relational view of market activityloses some of the evocative features ofPolanyi’s original, more institutional defini-tion. Social ties are fundamental to eco-nomic relations, certainly. What we findmore interesting is the insight that economicrelationships (as well as any other collectivesocial activity) can be understood in terms of

multiple institutional arrangements andlogics, only some of which are instrumental.Relationships matter precisely because theirmeanings are variable and depend on the ori-entations of participants to the various logicsand contexts them sensible.

Return, for a moment, to Macaulay’s (1963:61) discussion of contract and consider theoft-cited example of a businessman whonotes: ‘You don’t read legalistic contractclauses to each other if you ever want to dobusiness again. One doesn’t run to the lawyersif he wants to stay in business because onemust behave decently.’ There is certainly astory about trust, forbearance, and the shadowof the future implicit in these statements.Much less explored, however, is the idea that‘behaving decently’ is defined against a partic-ular social and institutional backdrop.

The idea that lawyers should be excluded isnot because they are personal strangers butbecause they view the same relationshipthrough a different institutional lens, whichhelps explain why they find the business-man’s approach ‘startling’ (Macaulay, 1963).As Macaulay noted, where businessmen seeorders that can legitimately be cancelled,lawyers see the same exchanges as contractswhose violation carries strongly negative con-sequences. Economic exchange has a non-contractual basis, but that bedrock is bothrelational and categorical. The meaning of arelationship and the actions appropriate to itdepend jointly on the parties to the tie and thebroader institutional and professional milieusto which they belong. Put differently, rela-tionships are multiply embedded and thesocial entanglements that make economicexchange possible are the joint outcome ofboth networks and institutions. We make asimilar claim about our final concept, socialcapital, which more closely situates individ-ual activities in both fields and networks.

Social capital

Like ‘embeddedness,’ the voluminous and disparate uses of ‘social capital’ have


9781412931236-Ch25 12/1/07 4:35 PM Page 601

rendered the concept slippery. At a basiclevel, capital is a resource that can grow withinvestment and use. Social capital, then, iscapital derived from relationships external tothe individual (Lin, 2001). In other words,social networks convey an array of resourcesto individuals at differential rates. Viewedinstrumentally, that capital can be investedwith some expectation of returns. The networks that convey social capital can beeither concrete, measurable relationships ormore diffuse affiliations based on groupmembership.

The latter sense of membership owesmuch to Durkheim and treats the collectiveeffervescence and shared identification ofsocial groups as both a public and a privategood that can be harvested for personal andcollective benefit (Putnam, 2000). InAlejandro Portes’ terms (1998: 52), ‘involve-ment and participation in groups can havepositive consequences for the individual andthe community.’ Other scholars take a nar-rower view, focusing more explicitly on con-crete relationships of exchange. Burt (2005:4), for instance, notes ‘One’s position in the structure of ... exchanges can be an assetin its own right. That asset is social capital, in essence, a conception of location effects indifferentiated markets.’ In this formulation,social capital derives from the structure ofthe collective, but the returns are to the indi-vidual, based on differential positions withinnetworks, rather than on the interplay of affiliation and identification emphasizedin more categorical, membership-based treatments.

Two often-cited general definitions ofsocial capital combine both these aspects in afashion that is instructive for our effort. Bothsituate social capital within a particular context, while treating it as an imperfectlyfungible resource. Coleman (1990: S98) rec-ognizes that social capital is plural: ‘Socialcapital is defined by its functions. It is not asingle entity, but a variety of different enti-ties, with two elements in common: they all consist of some aspect of social struc-tures, and they facilitate certain actions of

actors ... within the structure.’ Note three fea-tures of this definition. First, there are multi-ple ‘social capitals.’ Second, social capitaldoes not equally facilitate all activities.Third, the activities for which this capital isan efficacious resource are located within the structure that defined it. Put more suc-cinctly, social capital is contextual: it derivesfrom and only pays dividends in certain situations.

Pierre Bourdieu offers a subtly differentdefinition of the concept. He emphasizes abroader notion of social structure and a moreexplicit emphasis on resources, distinguish-ing social capital from both cultural andhuman forms of capital: ‘The aggregate ofthe actual or potential resources which arelinked to possession of a durable network of more or less institutionalized relationshipsof mutual acquaintance or recognition’(Bourdieu, 1985: 248). This definition is alsoa complicated one, but note that it makessocial capital an outcome of both direct tiesand recognizable membership. Perhaps moreimportantly, Bourdieu’s emphasis on institu-tionalized relationships returns us to the consideration of fields, a complementaryconcept in his theory of practical action.Here, capital (of whatever form) is derivedfrom the arrangements that characterize par-ticular fields, and it is within those fields thatdifferent varieties of capital can be mobilizedto serve disparate ends.

We argue that this close look at key con-cepts from institutional theory and networktheory demonstrates that the approaches areindissoluble. Fields are fairly barren withoutthe interpretive lenses and resource channelscreated by networks. Logics render networksand organizational structures sensible in par-ticular fields, but many, if not most, activitiesare amenable to multiple logics. Thus, theability of logics to shape social actiondepends intimately on the structures in whichactivities take place and the partners withwhom they are undertaken. Expanding thereach of all four concepts to more fruitfullycapture such relationships will be particu-larly important to the growing number of


9781412931236-Ch25 12/1/07 4:35 PM Page 602

studies that examine the genesis and dynam-ics of institutions and networks. In the fol-lowing section, we abstract from ourforegoing discussion to sketch an analyticframework that takes up this challenge.

CONTEXT OR CO-CONSTITUTION?

We have documented a set of analytic con-nections between networks and institutions.Canonical works in neo-institutional theoryrely explicitly on network imagery andmechanisms, while exemplary empiricalpieces demonstrate that networks are centralto explanations of institutional phenomena.Likewise, four master concepts – field, logic,embeddedness, and social capital – mix bothrelational and categorical claims. We believethese interdependencies can be understood intwo ways. The first, less radical view treatsnetworks and institutions as mutually reinforcing, contextual features of social systems. The second line of argument exam-ines how networks and institutions co-consti-tute one another. Put differently, the firstview sees institutions as the landscape andnetworks as the social relations on that field.The second view argues that fields influencewhich relations are possible, and how these relations are forged can alter the landscape in profound ways. We hold thesecond view.

Many institutionalists have recognized thatnetworks are important contexts for under-standing institutional process (Jepperson,1991; Davis, Diekmann, and Tinsley, 1994;Dobbin and Dowd, 2000). Contextual effectsare key, but more important, we believe, is theidea that networks and institutions are co-constitutive. In other words, networks shapeinstitutions but institutions sculpt networksand direct their growth. Genesis and change,not just context, are at stake in the merger ofstructural and cultural approaches to complexsocial systems.

Our argument rests on the idea that categorical distinctions are at the heart of

institutions, and the concrete relation-ships that are the basis of networks have adual character. Like other well-known duali-ties – between persons and groups (Brieger,1974), meanings and structures (Mohr,1998), organizations and environments(Stinchcombe, 1965) – we take meaningfulsocial categories to be defined in large partby relationships’ participants from withinand across them (White et al., 1976). At thesame time, the likelihood and implications ofparticular relationships stem from the categories that collaborators occupy andspan. As a result, categories and relationshipsjointly bound and determine action in socialsystems.

Understanding how networks and institu-tions co-evolve to shape social and economicarrangements requires us to attend to themyriad ways that relationships and cate-gories influence each other. We argue thatone force behind that shaping is organiza-tions and individuals who strive to navigatesettings where multiple institutional logicseither co-exist or collide.2 If logics offer tem-plates for action and organizing while ren-dering existing and potential relationshipsmeaningful, then settings where multiplelogics overlap will be particularly fertileground for institutional entrepreneurship.Some of those in structural locations thatengage multiple logics – as in art and com-merce, patient care and administrative efficiency, or altruistic medical donationsand income generation – can use their circumstances to forge new opportunities orcraft multivocal identities. In settings where numerous logics reflect conflicting orincompatible demands, ambiguous identitiesand multiple networks offer room to maneu-ver. Still, the tensions that are generated by ambiguity, multiplicity, and contradictioncan be daunting to individuals and organizations.

Practical action draws on both relation-ships and categories, and, in so doing, linksnetworks and institutions. Such efforts aremost visible in settings characterized by con-flicting logics, multiple audiences, and


9781412931236-Ch25 12/1/07 4:35 PM Page 603

ambiguous categories. Participants in spe-cific fields draw on categories and associatedlogics to make sense of their worlds anddirect their relationships and affiliations. Atthe same time, relationships and affiliationsoffer participants disparate types andamounts of capital, depending on their insti-tutional context. Continuity and change incategorization systems and network struc-tures alike depend on discernable patterns inthe formation of ties and affiliations.

How, then, do we explore the generativerelationship between networks and institu-tions? A thorough-going elaboration of a‘network-institutional’ research program isbeyond the scope of this chapter. We optinstead to reconsider some of our own workon the evolution of the human therapeuticand diagnostic biotechnology industry andon the institutional changes that surround thecommercialization of academic research. Inthe former setting, multiple logics of discov-ery associated with different types of organi-zational partners encourage biotechnologyfirms to create and maintain diverse networkties in order to innovate and develop novelproducts. These ties span multiple types oforganizations to form a field where relation-ships and outcomes alike are stamped by the categorical features of partners. In the latter setting, logics associated with the commercial use of science are imported intothe established field of public science, spark-ing both structural and institutional transfor-mations. We address a set of researchquestions that emerge from treating skilled,but constrained, performances as a mecha-nism linking relationships and categories.

The key questions we consider are:

(1) How do the meaning and consequences of relationships depend on the character of the participants?

(2) How do the effects of macro-structures dependon the types of participants that comprise them?

(3) How do locally situated individuals pull downglobal categories and draw on external relation-ships in their daily activities?

(4) How does situated action escape its local context toalter global categories and external relationships?

These questions do not exhaust the con-nections between networks and institutions.Nevertheless, we believe that initial answersto these queries will aid in developing atheory of social and economic life that treatsnetworks and institutions as flip sides of thesame analytic coin.

We begin by revisiting our work on theevolution of inter-organizational collabora-tion in human therapeutic and diagnosticbiotechnology. The commercial field of thelife sciences provides us with fertile groundto answer our first two questions. We firstdiscuss how the same collaborative activi-ties, for instance joint R&D efforts, havevery different implications for biotech firmsdepending on the organizational form ofpartners. Here categorically different formsof organization bring different logics to thesame activities. As a result, the likelihoodand effects of any particular tie depends oninstitutional features of the partner.

R&D undertaken with pharmaceuticalfirms, for instance, differs dramatically from scientifically comparable research con-ducted with academic, university-based collaborators because pharmaceuticals anduniversities operate in different selectionenvironments under different institutionallogics. Moreover, as the field developedbiotech firms and partner organizationsbecome relational generalists. In addition tolearning to manage multiple types of activi-ties across stages of product development,biotech firms developed the capacities neces-sary to conducting the same kinds of endeav-ors with different types of partners. Theirefforts to develop and maintain network port-folios that include diverse activities and part-ners accounts for the characteristic structureof the industry-wide network.

We next turn to analyses of innovation intwo densely populated biotechnologyregions, Boston and the San Francisco Bayarea, to address our second question. Thesetwo regional communities are highly produc-tive, but one (Boston) is anchored in a net-work that grew from public sector origins.The other community (SF Bay) is centered


9781412931236-Ch25 12/1/07 4:35 PM Page 604

on a network that emerged from startingpoints in venture capital (VC) initiatives. Thedifferent institutional anchors in these tworegional networks result in divergentapproaches to innovation. Both clusters arehighly successful and network structures arefundamental in both places, but the types ofsuccess and the ways in which networksmatter vary with the organizational form andassociated logics of key participants.

We dramatically shift levels of analysisand go ‘microscopic’ to examine our nexttwo questions. We first consider the ways inwhich broad logics (of appropriate skepti-cism), salient categories (such as academicdiscipline) and concrete relationships (of col-laboration and mentorship) are pulled downinto the daily life of a scientific laboratory.The institutional and relational features ofacademic science shape laboratory life, butthey do so imperfectly because they alsooffer researchers avenues for resistance.Finally, we turn to an analysis of decisionmaking in a high-profile technology licens-ing office to consider how local action canescape its immediate context to reshapebroader categories and relationships. In this instance, situated efforts to resolve con-tradictory logics at the boundary betweenacademe and the market drive licensing officers to create complicated deals that canentangle participants from different marketand technological categories, thus shiftingthe character of the field.

THE RECURSIVE NATURE OFNETWORKS AND INSTITUTIONS

In their study of the commercialization of thelife sciences, Powell and colleagues (2005)offer a co-evolutionary analysis of how fieldsand networks influence one another. Theirstarting point is the view that fields emergewhen social, technical, or economic changesexert pressures on existing relations, andreconfigure models of action and socialstructures. In this respect, they follow

Bourdieu and Wacquant (1992) in viewing afield as a center of debate in which compet-ing interests negotiate over resources and theinterpretation of key rules and conventions.Their study focuses on the interaction ofmultiple overlapping networks through timeby examining how the formation, dissolu-tion, and rewiring of network ties from 1988to 1999 shaped the opportunity structure ofthe biomedical field. By linking an evolvingnetwork topology and field evolution, Powellet al. demonstrate that social change is not aninvariant process that affects all participantsequally. Rather, field-level transfor-mations are multi-dimensional phenomena.Organizations feel the reverberations ofchange in different ways depending on theirinstitutional status and location in the overallnetwork. But the status orders and structuresof the field change over time.

The analytical aim of the 2005 paper wasto illuminate how patterns of network inter-action emerged, took root, and transformedthe field, with disparate ramifications for allof the varied participants. The empirical set-ting was the field of biotechnology, whichdeveloped out of university laboratories inthe 1970s, saw the founding of dozens of sci-ence-based companies in the 1980s, andmatured in the 1990s with the release ofdozens of novel medicines. The field isnotable for both scientific and commercialadvances and a diverse cast of organizationsranging from universities, public researchorganizations, venture capital firms, dedi-cated biotech firms, and giant multinationalpharmaceutical corporations. Because thesources of scientific leadership were widelydispersed and developed rapidly, and the rel-evant skills and resources needed to producenew medicines were broadly distributed, theparticipants in the biomedical field havefound inter-organizational collaborationessential (Powell, Koput, and Smith-Doerr,1996). By analyzing the evolving structure ofinter-organizational networks, we demon-strated how the larger field and its conven-tions changed both the meaning of ties andthe practice of collaboration.


9781412931236-Ch25 12/1/07 4:35 PM Page 605

In the early years of the industry, from1975 to the late 1980s, most biotech firmswere small companies that relied heavily onexternal support. No biotech firm had thenecessary skills or resources to bring a newmedicine to market in the early days, thusthey became involved in an elaborate lattice-like network of relationships with universi-ties, hospitals, and large multinational firms.The large corporations, despite well-established internal capabilities, lackedaccess to the cutting edge of university sci-ence. Deficient in a knowledge base in thenew field of molecular biology, large firmswere drawn to the biotech start-ups that hadmore capability at basic and translational sci-ence. This diverse distribution of technologi-cal and organizational resources was a keyfactor driving early collaborative arrange-ments in the industry. A number of institu-tional factors undergirded this collaborativedivision of labor.

The breakneck pace of technical advancehas rendered it difficult for any single organ-ization to remain scientifically abreast onmultiple fronts, hence linkages to universitiesat the forefront of basic science have beennecessary. The availability of funding alsoincreased rapidly, as biomedicine became amajor force in modern global society. Thebudget of the U.S. National Institutes ofHealth, a key funder of basic research, nearlydoubled in the 1990s during the Clintonyears. Venture capital financing flowed intobiotech somewhat irregularly in the 1990s,but over the course of the decade grewmarkedly. Biotech financing by venture cap-ital has always been somewhat counter-cycli-cal. When there was great enthusiasm for theinternet and telecommunications start-ups,interest in biotech waned. But when thebloom fell off the internet rose, financing forbiomedical ventures went on the upswing.

Two factors stood out in shaping the earlystructure of the field and the nature of its net-works. One is that the different members ofthe field had varying abilities and competen-cies. Some of the participants were highlyspecialized, while others had a hand in

multiple activities. For example, universitiesand public research organizations specializedin basic science and in early stages of drugdevelopment. Venture capital firms special-ized in financing. Biotechnology companies,and especially large multinationals, tended tohave a hand in many more activities. Morerecently, public research organizations suchas universities have greatly broadened theirrange of endeavors in the biomedical field.The most dramatic finding of this researchwas that, over time, all participants in thefield had to learn to master a wider array ofrelationships and move from specialist togeneralist roles. That move makes the needto navigate multiple potentially competinglogics a key feature of the field.

Second, as the field gained coherence andthe pattern of reliance on networks solidified,various institutions emerged to facilitate andmonitor inter-organizational collaboration.Offices were established on university cam-puses to promote technology transfer, lawfirms developed expertise in intellectualproperty issues, and various angel investorsand venture capital firms provided financing,along with management oversight and refer-rals to a host of related businesses. As theserelations thickened and a relational contract-ing infrastructure grew, the reputation of aparticipant came to loom large in shapingidentities (Powell, 1996).

There are two aspects of this analysis thatare highly relevant to our current discussionof network and field evolution. One, notablechanges in the nature of the actions pursuedby the field’s participants accompanied shiftsin the field’s characteristic practices, logics,and norms. Two, both the cast of participantsand the rules of the game changed as newlogics of affiliation emerged and spread. Webriefly summarize these two co-evolvingtrends, and refer the reader to the more exten-sive discussion in the 2005 paper.

In the late 1980s, the most active partici-pants in the emerging biotechnology industrywere the dedicated biotech firms, pharma-ceutical corporations, and key governmentagencies such as the National Institutes of


9781412931236-Ch25 12/1/07 4:35 PM Page 606

Health. In these early years, biotech firmslacked the capability to bring novel medi-cines to market, while large firms trailed inunderstanding new developments in molecu-lar biology (Gambardella, 1995; Powell andBrantley, 1992; Henderson, Orsenigo andPisano, 1999). Venture capital activity inbiotech was limited, and most small compa-nies supported their research and develop-ment activities by selling their lead productsto large corporations, which subsequentlymarketed the medicine and pocketed thelion’s share of the revenues (Powell andBrantley, 1996).

A handful of emerging dedicated biotechfirms with considerable intellectual propertyand strong translational research ability werehighly sought after as collaborators. This firstwave of biotechs founded in the 1970s andearly 1980s included Genentech, Centocor,Amgen, Genzyme, Biogen and Chiron, andthe most active large corporate partners werefirms such as Eastman Kodak, Johnson andJohnson, and Hoffman La Roche. While thecommercial logic of young firms selling theirlead products to major corporations domi-nated the landscape of the 1980s, a new set ofrelationships was quietly emerging.

The National Institutes of Health beganforging R&D relationships with new entrantsto the industry, and linking university scien-tists and start-up firms. As the science under-girding biotechnology expanded by leaps andbounds, the intellectual property associatedwith the science became more codified andlegally secure. This, in turn, attracted greaterinterest from venture capital. By the early1990s, biotech firms not only had highlyprestigious science, evidenced by publica-tions in top-tier journals, but they also hadsecure legal rights to their intellectual prop-erty in the form of patents. The networks ofaffiliation began to change, in some respectsquite dramatically. By the mid 1990s, the most active participants in the field con-tinued to be dedicated biotech firms, but thelarge pharmaceutical companies were pushedto the sidelines by the entrance of venturecapital firms and universities. Moreover, the

primary locus of activity shifted from com-mercialization to research and developmentand finance.

The industry expanded geographically aswell, moving from its early origins in the BayArea and Boston to San Diego and a handfulof other key regions in the United States andEurope (Owen-Smith, Riccaboni, Pammolli,and Powell 2002). Growth in the number ofnew firms, new partnerships, and new ideaswas greatly enhanced by an increase in finan-cial linkages and government research fund-ing. The combination of the growth of privateequity markets and national funding forR&D replaced the former reliance on largecorporations for support.

The older relationship with giant multina-tionals for commercialization activity was avery restrictive one. A small handful of firmshad the ability to take a drug to late-stagedevelopment and a small set of dominantmultinationals could manufacture and dis-tribute the drug worldwide. This commer-cialization arrangement was a downstreamactivity, involving the sale of a new medicalproduct. One might consider it the last dancein the product life cycle. In contrast, financeis an upstream activity, which fueled researchand development, licensing, and subsequentcommercialization. Consequently, it enrolledmany more participants into the industry net-work. With the addition of more participants,a wider array of organizational forms joinedthe field. Diversely anchored, multi-con-nected networks are much less likely tounravel than are networks that are reliant ona few forms of organization.

Most notably, multiple logics were now atplay. Pharmaceuticals began to recognizethat they had to learn skills other than devel-opment and commercialization in order tocompete with university researchers and ven-ture capitalists for access to cutting edgeideas. The network structure of the field con-tinued to expand throughout the 1990s, asboth the number of entrants and the numberof ties linking incumbents and new entrantsexpanded greatly. Indeed, in 1998 more than 1100 new ties were forged. All of the


9781412931236-Ch25 12/1/07 4:35 PM Page 607

participants – from federal funding agenciesto universities to biotech firms to pharmaceu-ticals – had begun to engage in a wide arrayof activities and were no longer specialists.We refer to the ability of participants to par-ticipate effectively in multiple kinds of tieswith diverse parties as multivocality, a domi-nant pattern that emerged whereby highlycentral participants were involved in adiverse array of collaborations with an exten-sive set of partners of different types (Padgettand Ansell, 1993).

As the cast of participants grew , and asdiversity in both organizational form andactivity became more important, new logicsof affiliation took hold. In the early years,there was a powerful influence of accumula-tive advantage. Those entrants who had themost visibility attracted the most attentionand the greatest sponsorship. In short, therich got richer in the Mertonian sense(Merton, 1968). As the field grew, homophilybecame more important, particularly in termsof geographic location as firms located nearone another connected. Particular regions ofthe country became known for their biotechclusters. Through time, a logic of appropri-ateness developed, in which assumptionsabout what a biotechnology firm looked likebecame widely accepted. A new canonicalfirm excelled in translational science andtypically had ties to a research university, aventure capital firm or two, and a large cor-porate partner. A highly successful firmwould add affiliations with a noted researchhospital and perhaps a federal agency, suchas one of the branches of the NationalInstitutes of Health. But note that each ofthese affiliations was for a specific type ofrelationship, the venture capital tie forfinance, the university tie for research, thelink to the hospital for clinical trials, and thepartnership with a large corporation for com-mercialization.

As the field evolved, the diversity of participants began to reshape the range ofactivities that the participants undertook. Askey participants became relational general-ists, the logic of affiliation that we dub

combinatorial or multivocal took root andbegan to diffuse to the field’s periphery.Neither money, market power, nor the sheer force of novel ideas dominated thefield. Rather organizations with diverse port-folios of well-connected collaboratorsbecame the most cohesive, central partici-pants in the field and played the largest role in shaping its evolution. The tight density of the expanding network and theopen scientific trajectory combined toenhance the importance of the various partic-ipants’ reputations. The pattern of cross-cut-ting collaboration meant partners on oneproject were often rivals on another. As aresult, networks were frequently rewired.Thus participants had to learn how to exitrelationships gracefully, so as not to damagetheir future collaborative prospects. The co-evolution of networks and categories in thefield created a social structure in which exter-nal sources of knowledge and resourcesbecame widely differentiated and a prefer-ence for diversity and affiliation with multi-ply connected partners had powerfulmobilizing consequences.

INSTITUTIONAL EMBEDDEDNESSSHAPES CATEGORIES ANDPRODUCTS

In more recent work we have moved fromanalyzing the evolution of the macro-networkto a more fine-grained study of the two mostactive biotechnology clusters, the SanFrancisco Bay Area and the Cambridge/Boston region. The attributes and successesof these clusters are widely studied, and theirefforts have been broadly emulated world-wide (Powell, Owen-Smith, and Colyvas,2007). But interestingly, despite their similar-ities in scale and reputation, each regionemerged through distinctive patterns of col-laboration that appear to influence their char-acteristic processes of discovery and types ofinnovation. We explored the relationshipbetween the forms of affiliation and the types


9781412931236-Ch25 12/1/07 4:35 PM Page 608

of innovative activity pursued in these regions(Owen-Smith and Powell, 2006).

In the conceptual terms used earlier, ouranalyses of innovation in Boston and BayArea biotechnology draw on two core ideas.First, our earlier work demonstrates that theorganizational form of the dominant playersin a network shapes the character of socialcapital in a community. Where universitiesdominate, a logic of discovery that favorsopenness and information diffusion prevailsand membership alone suffices to increaserates of innovation. In contrast, when for-profit organizations are key players in thenetwork and more ‘closed,’ proprietary logicsare at the fore, a central network position isessential (Owen-Smith and Powell, 2004). Inaddition to shifting the ways that organiza-tions extract benefits from their networks, thedifferent logics associated with partners ofdisparate form shape strategies for innova-tion, the kinds of connections firms forge andthe markets they seek to serve.

Recall our description of the different typesof organizations — including VC firms, gov-ernment agencies, large multinationals, largepublic research organizations, and dedicatedbiotech firms – that comprise this field. Thesediverse organizational forms were linked bymultiple types of affiliations: R&D connec-tions for shared research and development,finance ties reflecting investment, licensingrelations that transfer the rights to intellectualproperty across organizations, and commer-cialization partnerships that include productdevelopment, clinical trials, manufacturing,and sales and marketing.

We find two notable differences betweenthe Bay Area and Boston regional networks.The Bay Area is larger organizationally andgeographically, with many more biotechfirms, several major universities, includingStanford and the Universities of California(UC) at Berkeley and at San Francisco(UCSF), and numerous venture capital firms.The Boston network, while denser and some-what smaller, has many more public researchorganizations, including MIT, HarvardUniversity, Massachusetts General Hospital,

Dana Farber Cancer Center, and Brighamand Women’s Hospital among others. The Boston area had many fewer venturecapital firms in the 1970s and 1980s, andVCs arrive in the Boston region much later(Powell, Koput, Bowie, and Smith-Doerr,2002). Neither region housed a large multi-national pharmaceutical corporation duringthe period stretching from 1970s through the1990s. Both clusters have structurally cohe-sive networks, but they differ in the demog-raphy of their organizational types.

We have shown that the Boston networkgrew from early origins in the public sector,and that public science formed the foundationor anchor for subsequent commercial applica-tion (Owen-Smith and Powell, 2004; Porter,Whittington and Powell, 2005). Because theBoston biotechnology community was linkedby shared connections to public researchorganizations early in its evolution, this clus-ter manifested a more open technological tra-jectory than a cluster that relied more heavilyon industrial R&D. By contrast, the Bay Areawas much influenced by the prospecting andmatchmaking efforts of venture capitalists,the multidisciplinary science of the UC SanFrancisco medical school, and the novelefforts at technology transfer at Stanford(Colyvas 2007). The San Francisco Bay Areaevolved out of this more commercial andentrepreneurial orientation. Interestingly,both Boston and the San Francisco Bay Areadeveloped from dependence upon a non-biotech organizational form, and thesediverse forms, whether they are public science organizations or highly engagedentrepreneurial financiers, helped catalyzethe development of the respective clusters.

Do these different relational componentsand logics influence the nature of researchand the kinds of medical products thatemerged from the companies in these tworegions? We explored this question in twoways, by examining the nature of patentingamong the participants in the two regions andthrough a paired comparison of two compa-rable treatments for multiple sclerosis. In a2006 study, we found a significant difference


9781412931236-Ch25 12/1/07 4:35 PM Page 609

in the patenting activity of biotech firms inBoston and the Bay Area, with Bay Areafirms producing roughly 3,800 U.S. utilitypatents over the period 1988–1999, whileBoston area firms generated 1,376. Bay Areafirms appear to be much more prolific paten-tors. The highly skewed distribution of patents, however, suggests that the differ-ence results from a small number of excep-tionally productive Bay Area companies(Owen-Smith and Powell, 2006).

More interestingly, however, are the con-siderable differences we observe in the cita-tions Boston and Bay Area firms make intheir patents. These data suggest that Bostonbiotechs more routinely engage inexploratory innovative search, which typi-cally yields a few very high impact patents atthe expense of numerous innovations withlower than average future effects (Flemingand Sorenson, 2001). In contrast, the domi-nant Bay Area patenting strategy appears tobe a more directed and incremental,‘exploitation’ strategy that is what one mightexpect of companies supported by investornetworks that demand demonstratedprogress. Companies that pursue exploitativestrategies generally develop numerousrelated improvements on established compo-nents of their research trajectories.Exploratory Boston area companies aremuch more reliant on citations to prior artgenerated by universities and public researchorganizations, while Bay Area companiesrely more on citations to their own prior art.Indeed, 71 percent of the patent citations byBoston companies are to prior art developedoutside at biotech firms.

How might such differences in patentingbe reflected in the kinds of products releasedby the companies? We used the Food andDrug Administration (FOA) approval recordsto identify the 58 new drugs developed byBoston and Bay Area biotech firms. Fifty-three of these medicines were approvedbetween 1988 and 2004. All of the drugs thatappeared on the market prior to 1988 weredeveloped by just two Bay Area firms, Alzaand Genentech. These early approvals no

doubt reflect the commercialization strategypursued in the Bay Area region. We find thatthe Boston-based companies had a strongerfocus on orphan drugs intended to treat rarediseases for patients with relatively smallmarkets. In 1983, the Orphan Drug Act wascreated to speed the development of thera-pies for rare diseases by offering tax breaksand regulatory assistance to organizationsthat would develop medicines for smallmarket medical needs. Many Boston-basedfirms have chosen to focus on orphan drugs,as one might expect of companies that areenmeshed in networks that are dominated byuniversities and hospitals. In contrast, BayArea biotech firms have pursued medicinesfor larger markets in which the potentialpatient populations run into the millions, andfor which there is likely to be stiff productcompetition. This high-risk, high-rewardstrategy shows the imprint of the venturecapital mindset.

We did a paired comparison of two drugs,Betaseron developed by the Bay Area firmCetus, which was eventually acquired byChiron, a Berkeley-based biotech firm, andAvonex, developed by Boston-based Biogenin tandem with Berlex Laboratories, anAmerican subsidiary of the German pharma-ceutical firm Schering-Plough. We comparedthese similar drugs by looking at FDA label-ing information and patenting citations toprior art. These two drugs are biologicallyand chemically comparable. Both are thera-pies for the same disease, recurring andremitting multiple sclerosis. Betaseron relieson a set of four patents, three initiallyassigned to Cetus and reassigned to Chironfollowing the merger of the two firms, aswell as one process patent, which was reas-signed to Berlex Labs. These four patentscite a small group of prior art patents, and inturn, a larger group of second generationcitations. In sum, Betaseron rests on a historyof some 55 interlocking patents, almost all ofwhich are based on intellectual propertyowned by companies.

Avonex, developed by Biogen, is based ona single compound patent, but it reached


9781412931236-Ch25 12/1/07 4:35 PM Page 610

more broadly into the prior art, relying on155 separate pieces of intellectual property.Not a single piece of the prior art on whichAvonex depends is owned by Biogen, sug-gesting that Biogen developed its market-leading therapeutic drug without the benefitof a thicket of intellectual property rights,and relied instead on a mix of public domainscience and its partners’ intellectual property.Obviously, internal R&D was critical to thedevelopment of both drugs, Biogen’s muchheavier reliance on public science reflectedthe local network characteristics of Boston.relied much more heavily on public science,which characterized the local network inBoston, more notably than is the case in theBay Area. Indeed, among the holders of thepatents for the prior art for Avonex are MITand the Massachusetts General Hospital.

Citation network comparison for similardrugs offers an interesting natural experi-ment that holds constant technical, clinical,and regulatory features of the innovationprocess. Even when such factors are quitesimilar, the patent citation networks underly-ing these two drugs differ in a manner thatreflects the larger institutional environmentof the regional innovation system. The BayArea-based drug relies more heavily on inter-nal R&D and on research efforts of otherfirms, while the Boston-based therapy drawson a broad cross-section of prior intellectualproperty owned by a wide range of differenttypes of organizations.

Our comparison demonstrates that the twonetworks bears a strong institutional foot-print. Bay Area firms were faster, more pro-lific in terms of new product development,and more likely to pursue novel medicinesfor large markets. In contrast, Boston firmswere more deliberative in their commercialstrategies and more likely to focus on medi-cines for identifiable patient populations inneed of relief from specific illnesses. Weconjecture that the organizational develop-ment and innovation processes were signifi-cantly influenced by the surroundinginstitutional environments. Boston is home toMIT, a powerful basic science institution that

lacks a medical school, Harvard, a world-class institution rich in basic science with anotable medical school, and numerous lead-ing research-oriented hospitals and healthinstitutes. The upshot of this institutional mixappears to be a corporate focus on expansivescience and new treatments for definablepatient populations.

In contrast, the biotech community in theBay Area had its earliest origins in the part-nership of Herbert Boyer, the UC SanFrancisco scientist, and Robert Swanson, aprominent venture capitalist, who joinedtogether to create Genentech, one of the firstbiotech companies, and long a bellwether ofthe industry. UCSF is an unusual institutionthat lacked disciplinary departments in thefull range of research programs. The organi-zational model at UCSF was interdiscipli-nary, with a cross-functional approach tomedicine and an emphasis on translatingbasic science into clinical applications(Varmus and Wineberg, 1992). Genentechadopted and refined UCSF’s interdisciplinaryteam model, adding the impatience and rest-lessness of venture capital financiers withtheir focus on swinging for the fences. Thus,the company has pursued new medical prod-ucts for illnesses suffered by millions. Thiscontrast of Boston and the Bay Area, the mostprolific biotechnology clusters in the world,gives considerable insight into the manner inwhich the institutional field shapes the forma-tion of networks. Our examinations of twoimportant biotechnology regions demon-strate that logics of action area shaped bygrowing network structures that influence thehabits of mind and the type of products thatcompanies develop.

SEEING THE FIELD IN PERFORMANCE

Linking relationships and categories throughsituated action requires us to understand howindividuals draw on and modify seeminglystable, persistent networks and classifications


9781412931236-Ch25 12/1/07 4:35 PM Page 611

in daily practice. We draw on examples from Owen-Smith’s (2001) ethnographicwork in a multidisciplinary neuroscience laboratory to explore how scientists draw onexisting categories and relationships to make sense of and maneuver within theirfields. The particular logics associated withdifferent scientists, technicians, researchers,and students, and the varieties of capital(resources) that can be derived from theirpositions and relationships have primacy inshaping both identities and opportunities.

In the H-lab3 – a large, multidisciplinaryacademic neuroscience group that conductsfundamental research on olfaction in themoth Manduca sexta – Owen-Smith foundthat collective opportunities for skepticismwere shaped by relative positions within thelaboratory as well as expectations based onthe ascribed skill and disciplinary affiliationsof participants. In the scientific field, skepticism is a core aspect of the logic bywhich novel claims are validated.

In an arena where tacit knowledge madedirect replication problematic and the multi-disciplinary insights needed to pursue theresearch nevertheless resulted in widely dif-ferent competencies, public performances ofskepticism and resistance became key fea-tures of training and knowledge production.Even though public episodes of skepticismwere clearly improvisational, they occurredagainst the backdrop of well-understood(though rarely articulated) norms of appro-priateness. Those standards were structuredby the relationships (of mentorship, collabo-ration, and sponsorship), categories (disci-plines, methods, audiences), and hierarchies(status) that characterize scientific fields. Thestage for particular skeptical performanceswas thus neither flat nor neutral. Scientists’career trajectories and the fate of new knowl-edge claims owed much to the skilled per-formance and reception of skepticism.

Fligstein (2001) identifies social skill withthe ability to induce cooperation fromothers.4 This view of skill is apparent in indi-vidual scientist’s abilities to convince skepti-cal peers of the quality and validity of their

findings. Such skilled performances, Owen-Smith’s field-work suggests, are differen-tially enabled by scientists’ disciplinaryaffiliations. Disciplines have different sta-tuses in the academic field and participantsdraw on broad expectations of disciplinarycompetence in assessing scientific claims.Discipline is a means to position an individ-ual and his or her claims in a general statushierarchy and a tool for making sense of theircompetencies. Bio-physicists, for instance,are accorded different degrees and types ofleeway in skeptical interactions thanchemists or ecologists, even if their claimsdo not rest explicitly on the particular com-petencies associated with their discipline. Inpractice, comparable findings presented byscientists whose disciplines differ can meetwith disparate skeptical reactions.

Relationships matter equally as much asdisciplinary categories, however. New find-ings are also evaluated in light of the individ-uals whose research produced them.Collaboration is a clear example. The vastmajority of empirical articles in neurosciencehave multiple authors. In the life sciences,authorship apportions credit according to awell-understood formula.5 Co-authorship haslong been understood as a means for scien-tists to invest established stores of profes-sional credibility in findings and colleagues(Latour and Woolgar, 1976), but in categori-cally and hierarchically differentiated fields,such investments can be double-edged. Here, relationships serve as important con-duits of information and material resources,but they are also key prisms for evaluatingnew claims under conditions where directmonitoring and replication are implausible orimpossible.

Consider a skeptical performance thattakes discipline, status, and relationships intoaccount simultaneously (Owen-Smith 2001:445 fn). In addition to serving as a means tovalidate findings developed ‘locally’ in thelaboratory, collective skepticism was a meansby which members of the laboratory deter-mined which ‘external’ claims could betrusted and whether findings that contradicted


9781412931236-Ch25 12/1/07 4:35 PM Page 612

their own needed to be taken into account. Inone instance, Owen-Smith observed a longdiscussion of a working paper from outsidethe H-lab that purported to contradict one ofthe group’s primary findings. The paper gen-erated heated discussions, including oneinterchange between Beth (a technician in thelab) and Jim, the group’s principal investiga-tor. They discussed a paper authored byBlanca, a post-doc in a Scandinavian labora-tory, and Bill, that laboratory’s principalinvestigator (PI) and a former student in theH-Lab. In this interaction the categoricalimplications of Blanca’s discipline (she is achemist) and of her collaborative relationshipwith Bill loomed large, as does Bill’s relativestatus in the field and the legacy of his time inthe H-lab.

Beth: You might also want to ask her abouther method. Before she came here sheworked on really small beetles. That is areally difficult animal. She is an expertwith these methods and she has techniquesthat we do not. Also, she is really goodwith chemistry. She has a really strongbackground, stronger than anyone here. Sothe answer to your implication that shehasn’t thought through her controls is thatshe probably has!

Jim: There’s no question about the chem-istry, but she is working in Bill’s lab andwe know that Bill is a little too flamboyantwith his methods.

This snippet of conversation is part of alarger, collective skeptical performance thatdrew on categorically based expectations(e.g. chemists are good at structuring experi-mental controls), ascribed levels of individ-ual skill (e.g. working with a difficult modelanimal results in better technique), and pastrelationships (e.g. evaluations of Bill’s scien-tific competency based on his time in the H-lab), as well as present ones (e.g. Jim’sinsistence that Blanca’s discipline and skillbe interpreted in light of her senior co-author). Such performances were disci-plined by logics of action native to the

broader field (e.g. standards of presentation,means of apportioning credit via authorship),but relied on a mix of local and global stan-dards of appropriateness.

In instances like this one, local action andsituated performances bring categories, hier-archies, and relationships together in mean-ingful efforts to navigate a field. Globalfeatures of the field of neuroscience – a mul-tidisciplinary endeavor that plays out on astatus-differentiated pitch where collabora-tions are fundamental to claims-making andevaluation – are apparent in the local interac-tions of skilled scientists. Convincing oneanother of the validity (or lack thereof) ofparticular claims required both carefulrhetorical effort and the ability to draw thebroader field and its conversations into spe-cific performances.

PERFORMANCES CAN CHANGEFIELDS

Observations in the H-lab clarify some of theways that categories and relationships getimported into local performances. We alsosuggest there are (perhaps fewer) instanceswhere situated action can shift the categori-cal and relational features of fields. We offera pair of examples drawn from Owen-Smith’s field work in a high-profile univer-sity technology licensing office (TLO)(Owen-Smith 2005, 2007)6.

Like the H-lab, the TLO is a university-affiliated workplace situated in a differenti-ated, hierarchical field. Where work in theH-laboratory is focused on scientificattempts to understand the neuroscience ofolfaction, the TLO’s goal is to identify,manage and market potentially valuabletechnologies for ‘society’s use and benefitwhile generating unrestricted income forresearch and education.’The TLO is a bound-ary-spanning administrative unit, the effortsof which are framed as a service to facultyresearchers and industrial partners. Theoffice’s staff comprises individuals who


9781412931236-Ch25 12/1/07 4:35 PM Page 613

typically hold bachelor’s (and in some casesmaster’s) degrees in technical fields.

Because this office spans the boundarybetween academe and industry, its work canbe understood in light of multiple logics. Inmost instances, licensing associates can drawon one or more ‘appropriate’ logics as theymake decisions. Because the office opts notto employ legal counsel, the staff has fewnormatively ‘correct’ approaches to prob-lems encoded in their training. These featuresand the prestige of this office make the TLOa fertile site for local action that can reshapebroader arrangements.

The TLO is one of the oldest and mostaccomplished offices of its kind. As a resultthe office, its staff, and (especially) its direc-tor occupy prominent positions in the rela-tively new field of university technologytransfer. Unlike academic life science, theprofessional field of university technologytransfer is still developing. Where skepticalevaluations in the H-lab draw explicitly onthe broader landscape of scientific norms,similar collective performances in the TLOare overwhelmingly local and only rarelyreach beyond the university (Owen-Smith,2007). The TLO’s highly visible position inthe field, combined with its enviable recordof success, results in its being widely emu-lated. Thus, the outcomes of actions taken inthe confines of the office often get trans-ferred out into the broader arena where theyalter the shape of relationships and help tocreate or modify emerging categories of pro-fessional action.

Much of the daily work of the TLO isinformed by routine meetings characterizedby improvisational efforts to make sense of,evaluate, and respond to scientific findingscouched as ‘invention disclosures.’ In theTLO, those evaluations take the form ofefforts to determine what kind of technologi-cal innovation is embedded in a scientificdiscovery and what sort of market that inno-vation might reach. Both of these decisionsare acts of classification, and once they aremade, TLO staff members (singly and, often,collectively) develop a plan for marketing

and licensing the invention.7 Marketing planstypically begin by ‘shopping’ an invention topotential licensees. Classifying a technologyin terms of existing markets and productstriggers licensing officers’ efforts to searchtheir ‘mental rolodexes’ for appropriate partners. Prior licensing relationships arehighly salient to that process.

Collective licensing discussions in the TLOtypically address difficult cases. Morestraightforward deals are the province of indi-vidual staff members. The most common formof difficulty arises as a result of the conflictinglogics under which university technologytransfer operations function. Recall the TLO’smission, which combines a focus on incomegeneration from licensing and efforts toensure broad public access to technologiesthat are often developed with federal R&Dfunds. This mission puts the TLO and otheroffices like it squarely at the intersection ofbusiness logics that emphasize revenues andacademic logics that emphasize open accessand the public good. That tension is palpablein discussions about whether federally funded,university-developed technologies should belicensed exclusively or non-exclusively. Theformer can be particularly lucrative (espe-cially if equity ownership in a start-up com-pany is a condition of the deal), but comes atthe expense of access. While they sometimesgenerate extensive revenues, non-exclusivelicenses often forgo a high financial upside tokeep a new technology accessible.

The technologies with the greatest poten-tial value often have the broadest appeal. Asa result, university technology managers routinely find themselves adjudicatingbetween more academic and more commer-cial approaches to their deals. In the TLO,this dilemma is commonly solved by writing‘field-of-use’ (FOU) deals that grant exclusive rights to different aspects or uses ofa technology to disparate licensees. As theTLO’s director noted in an interview (Owen-Smith, 2005: 83):

Almost everything we do is field-of-use. The posi-tive side of that is that you can get more than one


9781412931236-Ch25 12/1/07 4:35 PM Page 614

license in different fields. But there is also a nega-tive side. If there is a problem with a patent, or arelationship then you have compounded your difficulties if you have licensed it to multiple entities.

This brief description of a common responseto the one of the TLO’s primary institutionalcontradictions implies both the local chal-lenges and the global effects of widespreadreliance on this strategy. First, locally, TLOassociates who are often ill-prepared to dealwith highly technical inventions must partition early-stage technologies into multi-ple fields of use that can be separately andindependently licensed. Such efforts at dis-tinction are often imperfect, and raise subse-quent problems. More globally, when FOU licenses convey rights to the same tech-nology to licensees in widely disparateindustries, one effect is to make the TLO(and the university that houses it) a network‘short-cut’ between firms that might other-wise share no (or at least few) connections oraffiliations.

In other words, local decisions in the TLOcan yield changes in the relationships and cat-egories of larger fields by bridging otherwiseseparate licensees. To the extent that imperfect efforts to define fields of use arelikely to create later problems, FOU dealsmay be more likely to deeply entangle differ-ent licensees. Consider the example of an invention disclosed by a prominent biochemist. The technology – a compoundthat interrupts the metabolic processes of a particular bacterium and kills it – hasmultiple uses.

The bacterium in question produces asticky plaque that, if found in veins, has beenimplicated in heart attacks. If the bacteriuminhabits a washing machine, however, theplaque results in smelly clothes. The compound, then, can be understood as acomponent in laundry detergent or as a phar-maceutical aimed at the cardiovascularmarket. Both are potentially profitable usesbut, technically speaking, it is difficult to define separable fields of use because the mode of action of the detergent and the

therapy is difficult to distinguish. As a result,efforts to partition this technology to allowfor exclusive licensing to a pharmaceutical orbiotechnology company, and non-exclusivelicensing to manufacturers of laundry deter-gents may be imperfect. Problems that resultfrom trying to forge this separation may havethe unintended effect of creating relation-ships across industrial categories that other-wise might remain unconnected. Localevaluations of scientific findings in the H-labdraw broader categorical and relational char-acteristics into situated performances. Incontrast, similar efforts in the TLO havesome potential to alter the relationships andcategories of the larger field. Clearly, effortsin the H-lab could also remake its field by,for instance, making fundamentally noveldiscoveries about the neuroscience of olfac-tion. We do not wish to argue performanceson some stages can only draw upon theirlarger contexts while others can alter them.Instead, we wish to suggest that a network-institutional theory sensitive to genesis and,particularly, change should have some way toaccount for when and why local action shiftslarger fields. We believe that attending to theexisting stability of a field and to the relativepositions of actors within them offers someuseful starting points. It matters, for instance,that the H-lab is an important but not domi-nant player on a large and established fieldwhile the TLO is (arguably) one of the dominant players on a relatively young andgrowing field.

CONCLUSION AND FUTUREDIRECTIONS

We argue that understanding the characteris-tics and effects of social and economic systems requires simultaneous attention tothe categorical and relational features offields. Institutions and networks are inter-twined in canonical theoretical and empiricalworks in sociology and organizationaltheory. Perhaps more importantly, master


9781412931236-Ch25 12/1/07 4:35 PM Page 615

concepts in institutional and network theory – fields, logics, embeddedness, andsocial capital – are shot through with bothinstitutional and relational terminology.Throughout our discussion, we return to thedual relationship between social relation-ships and social categories that we take to beat the heart of a unified, ‘network-institu-tional’ approach to social and economic life.

Networks are essential to fields becausethey are both the pipes through whichresources circulate and the prisms thatobservers use to make sense of action. Fields,though, are associated with particular logicsof action and it is those logics that make net-works efficacious by determining whichsorts of relationships participants can con-ceive. More tellingly, the dominant logics ina field define which sorts of connections willhave what types of effects for different kindsof partners. In this sense, the relational andstructural embeddedness of economic actiondepends not just on networks but also on theorientations of participants to the fields andlogics that render ties sensible and helpdetermine the shape and effects of structures.Social ties and affiliations are not all of apiece. Similar activities and structures mayhave different implications depending on theinstitutional character of participants. Eventhe most purely structural definition of, forinstance, social capital must take institutionalcontext into account. Institutions and net-works jointly determine when various sortsof capital can be invested, by whom, andwith what expectations of returns.

We argue that networks and institutionsare co-constitutive. They set the conditionsof possibility for each other. At base, we takethis co-evolutionary relationship to rest on akey duality between relationships (the build-ing blocks of networks) and categories (thebuilding blocks of institutions). The situatedand often improvised performances of highlybounded, but nonetheless purposive, organi-zational and individual agents breathes lifeinto this duality, and, over time, provides amotor for evolution and change. We expectthe link between practical action, networks,

and institutions will be particularly apparent(and important) in situations where roles andidentities are ambiguous, logics and institu-tions are conflicting or multiple, and net-works span diverse audiences.

We revisited some of our empirical workon biotechnology, scientific collaborationand university technology transfer to exam-ine a few specific implications of ourapproach. Our discussion highlights fourissues where we believe further research canforward a network-institutional theory. Wesummarize those analytic questions and thensuggest some concrete methodologicalimplications for future studies of networksand institutions.

First, we contend that studies of evolutionand change in social systems must take into account the recursive nature of networksand institutions. Explaining the contempo-rary character of biotechnology requiresattention to the field’s history and to the par-ticular tensions that regulatory regimes andmarket competition create. More impor-tantly, we suggest, such an explanationrequires that we attend to the process bywhich collaborative relationships and network structures alike come to have theimplications and effects they do because oftheir institutional context. As a result, weargue that studies of social dynamics mustintegrate network and institutional conceptsand constructs.

Networks, then, must be understood in thecontext of institutional arrangements and the institutional embeddedness of networksshapes categories and products. Our exami-nation of innovation in the San FranciscoBay Area and Boston biotechnology regionsreinforces the important role network connections play in explaining outcomes. We add to that a recognition that the institu-tional characteristics of a network alter both the character of what participants produce and the process of production. Thefeatures and evolution of social systems, aswell as their substantive outcomes, areshaped by the joint pressures of networks andinstitutions.


9781412931236-Ch25 12/1/07 4:35 PM Page 616

Both of these arguments rest on a beliefthat participants’ strategies and rationalitiesare shaped by their network-institutionalcontext. We suggest that macro-organiza-tional efforts should attend more closely tobehavior. Whether that attention focuses onthe practical and situated performances ofindividuals or on re-integrating a behavioraltheory of organizations into network andinstitutional analyses depends primarily onlevels of analysis and topics. We shift focusfrom the dynamics of industry-wide net-works to situated action in bounded organi-zational settings to emphasize the linksbetween activities on the ground and broadercategorical and relational constraints.

We argued that attention should be paid toseeing fields in performances. Individualsand organizations act in contexts structuredby relationships, categories, and hierarchies.But such contexts cannot completely deter-mine action and sources of constraint canoffer unexpected opportunities to playerswhose positions and characteristics offerthem room to maneuver. If, as we contend,networks and institutions are yoked togetherby situated action, then studies of localaction must take relevant relationships, categories, and logics into account.

We do not wish to suggest, however, thatthe flow of influence is uni-directional. Fornetworks and institutions to be recursivelyrelated through action, the endeavors of par-ticipants must have some possibility of influ-encing their larger social environment. Moreeffort should be put to uncovering the situa-tions and conditions under which local per-formances shift the structural and institutionalfeatures of fields. Because the macro-socialworld is obdurate, action in some localesmust be more likely to effect change thanefforts in others. More attention should bepaid, then, to the relational and categoricalsources of innovation in fields. Even withinlocations that are situated to effect shifts intheir wider contexts, not all actions or partic-ipants are equally likely to have a broaderinfluence. As a result, we suggest that morestudies should focus on the conditions under

which particular practices and innovationsdiffuse or fail by making rules and practicesthemselves the unit of analysis.

Finally, our sketch of a network-institu-tional approach to social organization carriesmethodological implications. We note thatstudies at multiple levels of analysis – rangingfrom practices, to activities, organizations,dyads, collectives, structures and fields – arenecessary. More importantly, temporality anddynamics are at the center of our analyses.Seeing the interplay of networks and institu-tions, we contend, requires more than cross-sectional explanations of variation. Instead,efforts to track change in the categorical andrelational features of social worlds over timeare needed. Finally, we include a call forcomparison. Each of the studies we discussattends either to locally fluid behaviors or tochange over time in a single field. Variationsin networks and institutions, which can all tooeasily be treated as ubiquitous and invariantcharacteristics of social realms, may be moreapparent when we adopt a lens that empha-sizes comparative dynamics.

NOTES

1 Obviously, there are several notable exceptions,where more explicit conceptual connections areoffered. See Mizruchi, Stearns and Marquis, 2006;Strang and Meyer, 1994; Zuckerman, 1999;Zuckerman,Kim, Unkawa, and von Rittman, 2003;Powell, White, Koput, and Owen-Smith, 2005.

2 In this regard, see the chapters on the micro-level roots of institutional theory in this volume(Powell & Colyvas, Chapter 10 and Barley, Chapter20), as well as Barley and Tolbert (1997); Hallett andVentresca (2006).

3 In order to maintain confidentiality we refer tothe H-Lab and its occupants using pseudonyms.

4 In this regard, see also Callon (1986) and Latour (1987) whose formulations of actor-networktheory emphasize the differential abilities of individu-als to enroll disparate allies and maintain the stability of diverse constellations of relationships. Thissense of skill was also at play in the H-lab where sci-entists whose ‘golden hands’ routinely yielded partic-ularly compelling experimental data were accordedgreater deference than their less dexterous col-leagues.


9781412931236-Ch25 12/1/07 4:35 PM Page 617

5 The first author is usually a junior scientistresponsible for the bulk of the ‘bench’ work that sup-ports a particular claim. The last author is typically asenior scientist who ‘owns’ the lab in which the workoccurred and may have played a significant role indesigning experiments and framing questions. Otherauthors are typically arrayed alphabetically or in afashion that places the least important participantnearest the center of a long list. While they are rarelyexplicitly articulated, such authorship rules representanother characteristic logic that lets participants eval-uate claims and scientists by attending to researchers’relative positions in author-lists and to the strength ofa finding’s association with high-profile scientists.

6 All names are pseudonyms.7 A license is a deal that transfers the right to use

an invention or material that is protected by someform of intellectual property from the property’sowner (in the case a university) to a licensee (mostoften a firm) that hopes to develop it. Licenses canconvey exclusive rights, non-exclusive rights, or somelimited form of exclusivity.

ACKNOWLEDGEMENTS

We are grateful to Royston Greenwood forcomments on an earlier draft.

REFERENCES

Barley, Stephen R. and Pamela S. Tolbert. 1997. ‘Institutionalization and structura-tion: Studying the links between action andinstitutions.’ Organization Studies, 18:93–117.

Bourdieu, Pierre. 1985. ‘The social space andthe genesis of groups.’ Theory and Society,14(6): 723–44.

Bourdieu, Pierre, and Loïc J. D. Wacquant.1992. An invitation to reflexive sociology.Chicago: University of Chicago Press.

Breiger, Ron L. 1974. ‘The duality of personsand groups.’ Social Forces, 53: 181–90.

Burt, Ronald S. 2005. Brokerage and closure:An introduction to social capital. New York:Oxford University Press.

Callon, Michel 1986. ‘Some elements of a soci-ology of translation: domestication of thescallops and the fishermen of St Brieuc Bay.’Pp. 196–233 in John Law (ed.), Power, action

and belief: A new sociology of knowledge.London: Routledge & Kegan Paul

Clemens, Elisabeth. 1993. ‘Organizationalrepertoires and institutional change:Women’s groups and the transformation ofAmerican politics, 1890–1920.’ AmericanJournal of Sociology, 98(4): 755–98.

Clemens, Elisabeth. 1997. The people’s lobby:Organizational innovation and the rise ofinterest group politics in the United States,1890–1925.

Cohen, M. D., March, J. G., and Olsen, J. P. 1972.‘A garbage can model of organizationalchoice.’ Administrative Science Quarterly,17(1): 1–25.

Coleman, James S. 1988. ‘Social capital in thecreation of human capital.’ American Journalof Sociology Supplement, 94: S95–S120.

Colyvas, Jeannette A. 2007. ‘From divergentmeanings to common practices: The earlyinstitutionalization of technology transfer inthe life sciences at Stanford University.’Research Policy, 36: 456–77.

Dalton, Melville. 1959. Men who manage.New York: John Wiley and Sons.

Davis, Gerald F. and Henrich R. Greve. 1997.‘Corporate elite networks and governancechanges in the 1980s.’ American Journal ofSociology, 103: 1–37.

Davis, Gerald F., Kristina Diekmann, andCatherine H. Tinsley. 1994. ‘The decline andfall of the conglomerate firm in the 1980s:The deinstitutionalization of an organiza-tional form.’ American Sociological Review,59: 547–70.

Dezalay, Yves and Bryant G. Garth. 1996.Dealing in virtue: International commercialarbitration and the construction of atransnational legal order. Chicago: Universityof Chicago Press.

DiMaggio, Paul J. 1991. ‘Constructing an organi-zational field as a professional project: U.S. art museums, 1920–40.’ Pp. 267–92 inW. Powell and P.J. DiMaggio (eds.), The new institutionalism in organizational analysis, Chicago: University of Chicago Press.

DiMaggio, Paul J. and Walter W. Powell. 1983.‘The iron cage revisited: Institutional isomor-phism and collective rationality in organiza-tional fields.’ American Sociological Review,48: 147–60

Dobbin, Frank and Timothy J. Dowd. 2000. Themarket that antitrust built: Public policy,


9781412931236-Ch25 12/1/07 4:35 PM Page 618

private coercion and railroad acquisition.’American Sociological Review, 65: 631–647.

Ferguson, Priscilla Parkhurst. 1998. ‘A culturalfield in the making: gastronomy in 19th cen-tury France.’ American Journal of Sociology,104: 597–641.

Fleming, Lee and Olav Sorenson. 2001.‘Technology as a complex adaptive system:Evidence from patent data.’ Research Policy,30: 1019–39.

Fligstein, Neil. 2001. ‘Social skill and the theoryof fields.’ Sociological Theory, 19: 105–125.

Fourcade, Marion. 2007. ‘Theories of market,theories of society.’ American BehavioralScientist, 50: 1015–34.

Friedland, Roger, and Robert R. Alford. 1991.‘Bringing society back in: Symbols, practices,and institutional contradictions.’ Pp. 232–63in Walter W. Powell and Paul J. DiMaggio(eds.), The new institutionalism in organiza-tional analysis, Chicago: University ofChicago Press.

Galaskiewicz, Joseph and Ronald S. Burt. 1991.‘Interorganization contagion in corporatephilanthropy.’ Administrative ScienceQuarterly, 26: 88–105.

Gambardella, Alfonso. 1995. Science and inno-vation: The U.S. pharmaceutical industryduring the 1980s. Cambridge, U.K.:Cambridge University Press.

Gouldner, Alvin. 1954. Patterns of industrialbureaucracy. New York: Free Press.

Granovetter, M. 1985. ‘Economic action andsocial structure: The problem of embedded-ness.’ American Journal of Sociology,91(November): 481–510.

Hallett, Timothy and Marc Ventresca. 2006.‘Inhabited institutionalism: Social interac-tions and organizational forms in Gouldners’patterns of industrial bureaucracy.’ Theoryand Society, 35: 213–36.

Henderson, Rebecca, Luigi Orsenigo, and GaryP. Pisano. 1999. ‘The pharmaceutical indus-try and the revolution in molecular biology:Interactions among scientific, institutional,and organizational change.’ In D.C. Moweryand R. R. Nelson (eds.), Sources of industrialleadership: Studies of seven industries.Cambridge, UK: Cambridge University Press.

Hoffman, Andrew J. 2001. From heresy todogma: An institutional history of corporateenvironmentalism. Stanford, CA: StanfordUniversity Press.

Jepperson, Ronald L. 1991 ‘Institutions, institu-tional effects, and institutionalism.’ Pp. 143–63 in W.W. Powell and P.J.DiMaggio (eds.), The new institutionalism inorganizational analysis. Chicago: Universityof Chicago Press.

Krippner, Greta. 2001. ‘The elusive market:Embeddedness and the paradigm of economicsociology.’ Theory and Society, 30: 775–810.

Latour, Bruno. 1987. Science in action: How tofollow science and engineers through society.Cambridge, MA: Harvard University Press.

Latour, Bruno and Steve Woolgar. 1976.Laboratory life: The construction of scientificfacts. Princeton, NJ: Princeton University Press.

Laumann, Edward O., Joseph Galaskiewicz,and Peter V. Marsden. 1978. ‘Communitystructure as interorganizational linkages.’Annual Review of Sociology, 4: 455–84.

Lie, John. 1997. ‘Sociology of markets.’ AnnualReview of Sociology, 23: 341–60.

Lin, Nan. 2001. Social capital: A theory ofsocial structure and action. New York:Cambridge University Press.

Macaulay, Stewart. 1963. ‘Non-contractualrelations in business.’ American SociologicalReview, 22: 55–67.

March, James G. and Johan P. Olsen. 1984.Rediscovering institutions. New York: TheFree Press.

Merton, Robert K. 1968. ‘The Matthew effectin science.’ Science, 159: 56–9.

Meyer, John W., and Brian Rowan. 1977.‘Institutionalized organizations: Formal struc-ture as myth and ceremony.’ AmericanJournal of Sociology, 83: 340–63.

Mizruchi, Mark S. and L.C. Fein. 1999.‘Coercive, normative, and mimetic isomor-phism: A study of the social construction ofsociological knowledge.’ AdministrativeScience Quarterly, 46: 29–56.

Mizruchi, Mark S., Linda B. Stearns andChristopher Marquis. 2006. The conditionalnature of embeddedness: A study of borrow-ing by large U.S. firms. AmericanSociological Review, 71(2): 310–33.

Mohr, John. 1998. ‘Measuring meaning struc-tures.’ Annual Review of Sociology, 24:345–70.

Owen-Smith, Jason. 2001. ‘Managing laboratorywork through skepticism: Processes of evalua-tion and control.’ American SociologicalReview, 66: 427–52.


9781412931236-Ch25 12/1/07 4:35 PM Page 619

Owen-Smith, Jason. 2005. ‘Dockets, deals, and sagas: Commensuration and therationalization of experience in universitylicensing.’ Social Studies of Science, 35:69–97.

Owen-Smith, Jason. 2007. ‘Where does profes-sional knowledge come from? Licensing talk,problem resolution, and rationalization inuniversity licensing.’ Working paper,University of Michigan.

Owen-Smith, Jason and Walter W. Powell.2004. ‘Knowledge networks as channels andconduits: The effects of spillovers in theBoston biotechnology community.’Organization Science, 15: 5–21.

Owen-Smith, Jason and Walter W. Powell.2006. ‘Accounting for emergence and novelty in Boston and Bay area biotechnol-ogy.’ in P. Braunerhjelm and M. Feldman(eds.), Cluster genesis: The emergence of technology clusters and their implicationsfor government policy. Cambridge, UK:Cambridge University Press.

Owen-Smith, Jason, Massimo Riccaboni, FabioPammolli, and Walter W. Powell. 2002. ‘Acomparison of U.S. and European univer-sity–industry relations in the life sciences.’Management Science, 48: 24–43.

Padgett, John F. and Christopher K. Ansell.1993. ‘Robust action and the rise of theMedici, 1400–1434.’ American Journal ofSociology, 98: 1259–319.

Palmer, Donald, P. Devereaux Jennings andZueguang Zhou. 1993. ‘Politics and institu-tional change: Late adoption of the multidi-visional form by large U.S. corporations.’Administrative Sciences Quarterly, 38(1):100–31.

Podolny, Joel M. 2001. ‘Networks as the pipesand prisms of the market.’ American Journalof Sociology, 107: 33–60.

Polanyi, Karl. 1957. ‘The economy as an insti-tuted process.’ Pp. 243–70 in K. Polanyi, C. Arensberg, and H. Pearson (eds.), Tradeand market in the early empires: Economiesin history and theory. Chicago: Free Press.

Porter, Kelley, Kjersten Bunker Whittington,and Walter W. Powell. 2005. ‘The institu-tional embeddedness of high-tech regions:Relational foundations of the Bostonbiotechnology community.’ Pp. 261–96 inStefano Breschi and Franco Malerba (eds.),Clusters, networks, and innovation. Oxford,UK: Oxford University Press.

Portes, Alejandro. 1998. ‘Social capital: Its ori-gins and application in modern sociology.’Annual Review of Sociology, 24: 1–24.

Powell, Walter W. 1996. ‘Inter-organizationalcollaboration in the biotechnology industry.’Journal of Institutional and TheoreticalEconomics, 120(1): 197–215.

Powell, Walter W. and Peter Brantley.1992.’Competitive cooperation in biotech-nology: Learning through networks.’ Pp. 366–294 in R. Eccles and N. Nohria (eds.),Networks and organizations. Boston, MA:Harvard University Press.

Powell, Walter W. and Peter Brantley. 1996.‘Magic bullets and patent wars: New prod-uct development and the evolution of thebiotechnology industry,’ pp. 233–60 inToshihiro Nishiguchi (ed.), Competitive prod-uct development, Oxford, UK: OxfordUniversity Press.

Powell, Walter W., Jason Owen-Smith, andJeannette Colyvas. 2007 ‘Innovation andemulation: Lessons from American universitiesin selling private rights to public knowledge,’Minerva, 45: 121–42.

Powell, Walter W., K.W. Koput, J.I. Bowie, andL. Smith-Doerr. 2002. ‘The spatial clusteringof science and capital: Accounting forbiotech firm–venture capital relationships.’Regional Studies, 36: 3: 291–306.

Powell, Walter W., Kenneth Koput, and LaurelSmith-Doerr. 1996. Interorganizational col-laboration and the locus of innovation:Networks of learning in biotechnology.Administrative Science Quarterly, 41(1):116–45.

Powell, Walter W., Doug White, Ken Koput,and Jason Owen-Smith. 2005. ‘Networkdynamics and field evolution: The growth ofinter-organizational collaboration in the lifesciences.’ American Journal of Sociology,110(4): 1132–205.

Putnam, Robert. 2000. Bowling alone: The col-lapse and revival of American community.New York: Simon and Schuster.

Rao, Hayagreeva, Phillipe Monin, and RudolpheDurand. 2003. ‘Institutional change in ToqueVille: Nouvelle cuisine as identity movementin French gastronomy.’ American Journal ofSociology, 108: 795–843.

Scott, W. Richard, Martin Reuf, Peter J. Mendel,and Carol Caronna. 2000. Institutionalchange and health care organizations.Chicago: University of Chicago Press.


9781412931236-Ch25 12/1/07 4:35 PM Page 620

Strang, David and John W. Meyer. 1994.‘Institutional conditions for diffusion.’ Pp. 100–12 in W.R. Scott and J.W. Meyer(eds.), Institutional environments and organ-izations: Structural complexity and individu-alism. Thousand Oaks, CA: Sage.

Stinchcombe, Arthur C. 1965. ‘Social structureand organzation.’ Pp. 142–93 in James G.March (ed.), Handbook of organizations.Chicago: Rand McNally & Co.

Thornton, Patricia. 2004. Markets from culture:Institutional logics and organizational deci-sions in higher education publishing.Stanford: Stanford University Press.

Varmus, Harold E. and Robert A. Weinberg.1992. Genes and the biology of cancer. NewYork: Scientific American Press.

White, Harrison, Scott Boorman, and RonaldBreiger. 1976. ‘Social structure from multiplenetworks I: Blockmodels of roles and positions.’American Journal of Sociology, 81: 730–79.

Whitley, Richard, (ed.). 1992. ‘Societies, firmsand markets: the Social structuring of busi-ness systems’. Pp. 5–45 in European BusinessSystems. London: Sage.

Zuckerman, Ezra. 1999. ‘The categorical imper-ative: Securities analysts and the illegitimacydiscount.’ American Journal of Sociology,104: 1398–438.

Zuckerman, Ezra W., Tai Y. Kim, Kalinda Unkawa,and James Von Rittman. 2003. ‘Robust iden-tities or non-entities? Typecasting in the fea-ture film labor market.’ American Journal ofSociology, 108: 1018–74.


9781412931236-Ch25 12/1/07 4:35 PM Page 621

jason owen-smith and walter w. powell -...

Documents