© copyright 2001, phil e. stetz
Post on 19-Nov-2021
3 Views
Preview:
TRANSCRIPT
TOWARDS A RECONCILIATION OF THE DIVERSIFICATION-
PERFORMANCE PARADOX: AN EXAMINATION OF STRATEGIES
ACROSS THE SPECTRUM OF DIVERSIFIED CORPORATIONS
by
PHILIP EDWARD STETZ, B.A., B.S., M.B.A.
A DISSERTATION
IN
BUSINESS ADMINISTRATION
Submitted to the Graduate Faculty of Texas Tech University in
Partial Fulfillment of the Requirements for
the Degree of
DOCTOR OF PHILOSOPHY
Approved
May, 2001
© Copyright 2001, Phil E. Stetz
ACKNOWLEDGMENTS
This journey would not have been possible without the help and support of many
people, and I would like to first express my appreciation to my dissertation committee:
Dr. Robert Phillips, Chair; Dr. Roy Howell, Co-chair; Dr. Kimberiy Boal; Dr. Peter
Westfall; and Dr. Duane Ireland. Each of you have provided sound guidance, intellectual
challenges, and invaluable assistance for which I am very grateful. I truly am fortunate
for your willingness to be part of my dissertation committee.
I especially acknowledge my mentor/nemesis, Dr. Phillips, whose support and
guidance, grounded in relevance, was very instrumental in my development. BBM, your
willingness to help at any time, day or night, is an inspiration. Dr. Boal, I have always
been impressed with your breadth and depth of knowledge and Dr. Howell, your
perception and insights are rare indeed. Finally, Dr. Ireland, you have been a steady
beacon and pillar of support for which I will always be grateful.
Special thanks also goes to the many who have contributed to my achieving this
end. Without your help when needed, the task would have been insurmountable.
Last, but by no means least, I wish to thank my two sons for their understandmg
and tolerance of why I was not always available for them. I have literally watched you
become young men and wish that I would have been able to spend more time with you
these past years. As a small token of appreciation and gratitude, I wish to dedicate this
dissertation to you both equally, Philip and Steve.
11
TABLE OF CONTENTS
ACKNOWLEDGMENTS ii
ABSTRACT v
LISTS OF TABLES vii
LIST OF FIGURES viii
CHAPTER
L INTRODUCTION 1
"Six Blind Men and An Elephant": A Fable 1
Purpose of Study 6
Organization of the Study 8
n. THEORETICAL DEVELOPMENT 9
Diversification 10
History 10
Theoretical Perspectives 14
Empirical Studies 24
Corporate Effects: Historical and Empirical Development 27
m. HYPOTHESES 46
IV. RESEARCH DESIGN 54
Level of Analysis 56
Data 57
Diversification Measures 58
Rumelt's Typology 59
Entropy Measure 60
Classification Methodology for Level of Diversification 63
iii
Performance Metric 65
Controls 68
Model 70
Summary 73
V. ANALYSIS 84
Results 84
Hypothesis 1 86
Hypothesis 2 87
Hypothesis 3 88
Additional Confirmation of Results 90
Discussion 93
Implications 108
VL CONCLUDING COMMENTARY 119
Limitations of Study 119
Caveat 122
Contributions 123
Future Research 124
REFERENCES 128
APPENDIX 144
IV
ABSTRACT
Most empirical research examinuig the value of diversification explores the
linkage between economic performance and the level of diversification at the corporate
level of analysis. However, without comparing the returns to diversification to business
units operating within a corporation's governance system to the retums of stand-alone
businesses or to other business units embedded in other diversified corporations, the
analysis can not directly address a fundamental question underpinning the research on
diversification, "Do corporations make businesses better off?" Furthermore, few studies,
uivestigating the relationship between diversification and performance, have controlled
for variables that have demonstrated effects on business unit performance. To address
these criticisms, this study focuses on the business unit level of analysis and employs a
general linear mixed model to investigate the linkage between the level of diversification
on business unit performance (fixed effects) while controlling for industry, corporate, and
business unit factors (random effects). Results show that the relationship between
business unit performance and the level of corporate diversification, m which the
business unit is embedded, is an inverted U-shaped relationship. Additionally, business
unit performance, for most levels of diversification, was significantly different firom that
of stand-alone firms, suggesting that diversification strategies may add value to
businesses over that which a business may achieve without corporate parentage. Business
units, within low to moderately diversified corporations, earned a 60% greater return, on
average, than that of single stand-alone firms. However, differences in performance of
business units embedded in diversified corporations, firom dominate through highly
diversified corporations, were non-significant.
VI
LIST OF TABLES
2.1 Theoretical Perspectives and Suggested Level of Diversification
to Exploit Their Respective Theoretical Premise 36
2.2 Empirical Fmdmgs on the Diversification-Performance Linkage 38
2.3 Review of Corporate, Industry, and Busmess Effects Studies 40
4.1 Total Diversification Scores: Cluster Analysis 79
4.2 Descriptive Statistics of Sample by Level of Diversification 81
4.3 Examples of Corporations as Classified by Level of Diversification 82
5.1 Mean and Standard Error Estimates of Business Unit ROAs Across the Spectrum of Diversified Corporations 112
5.2 Test of Differences in Means Between Diversified and Non-Diversified Corporations 113
5.3 Test of Differences in Means Among Business Units Embedded Within Corporations of Varying Levels of Diversification 114
5.4 Tests of Differences in Means Between Business Units Embedded Within Diversified Corporations and Single Stand-AIone Businesses 115
5.5 Test of Significance of Identifiable Assets, as a Fixed Effect, and ROA 117
A.l Hierarchical Cluster Analysis of Total Diversification Scores: Dendrogram 146
A.2 Key for Correspondence Between Case Number and TDS Scores 154
Vll
LIST OF FIGURES
2.1 Graphical Representations of Diversification Theories 37
4.1 Rumelt's Typology 76
4.2 Classification Methodology of Corporations
as to Level of Diversification 77
4.3 Total Diversification Scores: Histogram 78
4.4 Comprehensive Review of Classification System 80
4.5 Performance Metric: Accounting Based 83
5.1 Plot of Mean ROAs and Number of Business Units Across the Spectrum of Diversified Corporations I l l
5.2 Plot of Mean ROAs and Differences in Means Between Single Firms and Business Units of Dominant and MBC 116
5.3 Total Diversification Scores of MBC Operationalized
as Continuous Measure and as Categories 118
6.1 SIC/NAICS Classification System 127
A. 1 Hierarchical Cluster Analysis of Total Diversification Scores: Dendrogram Overview 145
Vll l
CHAPTER I
INTRODUCTION
"Six Blmd Men and An Elephant": A Fable
A long time ago in the valley of the Brahmaputra River in India there lived six men who were much inclined to boast of their wit and lore. Though they were no longer young and had all been blind since birth, they would compete with each other to see who could tell the tallest story.
One day, however, they fell to arguing. The object of their dispute was the elephant. Now, since each was blind, none had ever seen that mighty beast of whom so many tales are told. So, to satisfy their minds and settle the dispute, they decided to go and seek out an elephant.
Having hired a young guide, Dookiram by name, they set out early one morning in single file along the forest track, each placing his hands on the back of the man in front. It was not long before they came to a forest clearing where a huge bull elephant, quite tame, was standing contemplating his menu for the day.
The six blind men became quite excited; at last they would satisfy their minds. Thus it was that the men took turns to investigate the elephant's shape and form.
As all six men were blind, neither of them could see the whole elephant and approached the elephant from different directions. After encountering the elephant, each man proclaimed in turn:
'O my brothers,' the first man at once cried out, 'it is as sure as I am wise that this elephant is like a great mud wall baked hard in the sun.'
'Now, my brothers,' the second man exclaimed with a cry of dawning recognition, 'I can tell you what shape this elephant is - he is exactly like a spear.'
The others smiled in disbelief.
'Why, dear brothers, do you not see,' said the third man - 'this elephant is very much like a rope,' he shouted.
'Ha, I thought as much,' the fourth man declared excitedly, 'This elephant much resembles a serpent.'
The others snorted their contempt.
'Good gracious, brothers,' the fifth man called out, 'even a blind man can see what shape the elephant resembles most. Why he's mightily like a fan.'
At last, it was the turn of the sixth old fellow and he proclaimed, 'This sturdy pillar, brothers' mine, feels exactly like the trunk of a great arecapalm tree.'
Of course, no one believed him.
Their curiosity satisfied, they all linked hands and followed the guide, Dookiram, back to the village. Once there, seated beneath a waving palm, the six blind men began disputing loud and long. Each now had his own opinion, firmly based on his own experience, of what an elephant is really like. For after all, each had felt the elephant for himself and knew that he was right!
And so indeed he was. For depending on how the elephant is seen, each blind man was partly right, though all were in the wrong.
(Riordan, 1986, pp. 30-33)
The dispute about the shape and form of the elephant is analogous to the debate
concerning the linkage between diversification and performance within the field of
strategic management. For example, one author suggests no performance differences
exist between diversified and non-diversified firms, while another argues that the
performance of corporations with high levels of diversification exceeds that of firms that
are low to moderately diversified, and yet another claims that firms with low to moderate
levels of diversification are the most effective m achieving high performance.
The linkage between diversification and performance has received considerable
attention within strategic management over the last thirty years. Although considered the
most researched linkage in the literature (Chatterjee & Wemerfett, 1991), Reichers and
Schneider (1990) would suggest that theory development in diversification has not
reached Stage Three, where controversy wanes and reviews of the literature state what is
and what is not known. Markides and Williamson (1994) suggest there is "considerable
disagreement about how and when diversification can be used to build long-run
competitive advantage" (p. 149). Or, to rephrase the above statement, the consensus is —
there is no consensus conceming the linkage between diversification and performance.
However, there is consensus conceming methodological issues in that construct
measurement and the lack of controlling for important variables have aided in the
fi-agmentation of findings.
Hoskisson and Hitt (1990, cited in Hoskisson, Hitt, Johnson, & Mosel, 1993)
suggest that the confusion regarding the diversification-performance relationship is
"partially theoretical and partially methodological, although both are inextricably woven
because the methods employed to measure diversification often are associated with a
specific theoretical perspective" (p. 216). Support for this argument was found in my
review of empirical studies on corporate effects m that researchers, with an industrial
organization perspective, exclusively operationalize diversification using a corporate
focus measure (Wemerfelt & Montgomery, 1988; McGahan, 1998). Witiiout provoking a
debate conceming the tiieory ladenness of observation (Hunt, 1994), what may be gleaned
fi-om Hoskisson and Hitt's comment is the importance of operationalizing constructs by
measures tiiat have been assessed as to tiieir objectivity, reliability, and validity. In this
study, diversification is operationalized through a refined entropy measure, a measure
with established psychometric properties.
Palich, Cardinal, and Miller (2000) examined over thirty years of research on
diversification, and in their meta-analysis of over fifty studies, commented that "very few
of the studies accounted for the impact of firm size; firm leverage; and advertising,
capital, and R«&D intensities, each of which have demonstrated effects on performance in
prior research" (p. 169). Furthermore, they suggest that "adjusting or accounting for
these variables in future research may further clarify diversification-performance
relationships" (p. 169).
To address the above criticisms of past research efforts, this study draws fi-om the
research stream on corporate effects, both past (Rumelt, 1991; Roquebert, Phillips, &
Westfall, 1996; Chang & Singh, 2000) and emergent (Bowman & Helfat, 2001).
Although the effects literature has a long history and addresses similar questions as that
of diversification research, such as, "Does corporate strategy matter?" these bodies of
research have developed somewhat independently of each other. For example, few, if
any, of studies that are focal to the corporate effects literature were examined in the meta
analysis of fifty-five studies on diversification (Palich et al., 2000).
A possible explanation for this finding is that effects research, which utilizes a
unique research design, takes into account entire classes of effects, such as industry.
corporate, and business effects, in uivestigating the determinants of firm performance.
For example, corporate effects, defined as the impact on profitability of factors associated
with membership of muhi-businesses with individual corporations, may be composed of
several corporate-level factors, which mclude (1) scope — referring to the extent and
nature of diversification and vertical integration (Rumeh, 1974; Williamson, 1975, 1985),
(2) core competencies and resources (Prahalad & Hamel, 1990), (3) organizational
structure (Teece, 1981), (4) climate, (5) systems of planning and control (Miller &
Cardmal, 1994), (6) corporate financing, and (7) corporate management (Bowman &
Helfat, 2001).
The importance of modeling entire classes of effects was initially demonstrated by
Scott and Pascoe (1986) by showmg that a class, representing multiple factors, accounted
for the majority of the variance in profitability in their model over that explained by the
operationalization of specific constructs. Furthermore, it may be argued that an effect
may be a sufficient proxy for the net effect of all the individual factors that make up that
specific effect. For example, the variance attributable to corporate effects may be
partitioned into uidividual factors, such as diversification, that are conceptualized as
comprising that effect. Additionally, the effect literature has converged on the
significance of all three effects — industry, corporate, and business unit, as important
determinates of business unit profitability. Thus, m parallel with the diversification
research, the research on corporate effects suggests that not only are industry and business
level factors important, but also corporate level factors as well.
In an extension of the variance component modeling technique, which is often
used in the corporate effects research, a general linear mixed model offers a means
through which a researcher can investigate the impact on performance of the phenomenon
of interest while accounting for industry, corporate, and business effects, and thus
answers and extends the call of previous critiques (e.g., Palich et al., 2000).
Furthermore, the use of mixed models has two other important advantages. First,
one is able to integrate research across two or more levels of strategy (Dess, Gupta,
Hermart, & Hill, 1995). In controlling for industry, corporate, and business level effects
in the model, a researcher is addressing three different levels of strategy. Second, by
including multiple levels in the model, a researcher is also integrating multiple theoretical
firameworks, such as industrial organization economics (industry effects) and strategic
management (corporate effects) (Hitt, Hoskisson, & Kim, 1997).
Finally, in the examination of the efficacy of diversification with respect to
profitability, in addition to measurement and control issues, past studies have been
plagued by small sample sizes as well as the inadvertent selection of firms in superior
industries in terms of higher ROAs (Christensen & Montgomery, 1981). These issues are
also addressed in this study.
Purpose of Study
This study's purpose, through the use of objective measures along with a large
sample (19,724 observations and 3,243 corporations) drawn fi-om an entire sector, is to
demonstrate how the literature on corporate effects may mform and be mtegrated into the
research on the linkage between diversification and performance to address many of the
limitations of past studies on diversification. In this effort, I derive a (I) parsimonious
means through which performance of firms, across the spectrum of diversified
corporations — single stand-alone businesses through highly diversified corporations —
may be modeled and tested while accounting for industry, corporate, and business effects;
(2) in so doing, determine the shape and form of the diversification-performance
relationship across the diversification spectrum; and (3) determine if statistically
significant differences exist between the level of diversification and performance, as
measured by ROA.
Finally, most empirical research examining the value of diversification explores
the economic performance of diversification at the corporate level of analysis without
comparing the retums of diversification to business units operating within a corporation's
governance system to the retums of stand-alone businesses (Barney, 1997). It could be
argued that the most fundamental question underpinning the research on diversification
is, "Do corporations improve business performance?" (Bowman & Helfat, 2001; Rumelt,
Schendel, & Teece, 1994; Porter, 1987). This question echoes Barney's observation and
suggests that a more appropriate level of analysis may be to focus on the business unit
(BU). This focus has two major research advantages ui that it allows for the assessment
of the effects on business unit performance of (1) corporate stirategy; i.e., strategic choices
conceming the domain and scope of the business unit, and (2) a business unit competing
with other business units across the diversification spectrum. Therefore, to answer their
call, the level of analysis for this study is the business unit.
In summary, this study does not attempt to put forth any theory that is grand or
new. However, to my knowledge, it does represent the first integration of the research
streams on corporate effects and diversification into a single model (or even a single
study), and demonstrates how each may uiform each other, and in tum, lead towards a
reconciliation of the diversification-performance paradox.
Organization of the Study
This study is organized as follows: a review of the diversification literature and
the corporate effects literature is presented in Chapter n. The linkages between these two
streams of research lay the foundation for the generation of the stated hypotheses in
Chapter EI. Chapter FV discusses the research design through which the research
questions will be addressed. Chapter V presents the results of the analysis and discusses
the implications of the findings. In the final chapter. Chapter VI, the limitations and
contributions of the study along with directions for future research are presented.
CHAPTER n
THEORETICAL DEVELOPMENT
The literature on diversification spans over three decades of research with roots in
multiple business disciplines, including economics, sociology, and strategic management.
However, even though the domain may be considered vast and embodies muhiple
theoretical perspectives, the field is not mature. Scholars have not reached a consensus as
to the superiority of one theoretical perspective over another, nor has there been
consistency in the empirical findings on the Imkage between diversification and
performance. In sum, there is considerable disagreement about how and when
diversification can be used to build long-run competitive advantage.
To begin to bring about a reconciliation of the diversification-performance
paradox, this study draws upon an emerging research stream on corporate effects.
Although this research addresses similar questions to the diversification literature, such
as, "Does corporate strategy matter?" these streams of research have developed somewhat
independently of each other. For example, few, if any, of the studies that are focal to the
corporate effects literature were examined in a meta-analysis of over thirty years of
research on diversification.
To fully explore these related research streams, this chapter is divided into two
main sections, with the first section discussmg the diversification literature, beginning
with its history, and then reviews the various theoretical perspectives and empirical
findings, and the second section discusses the corporate effects literature and how it may
contribute to the research on diversification.
Diversification
History
Chandler (1962), in his seminal account of the history of American business
enterprise, suggests that the multi-divisional form first appeared m the United States
shortly after World War I and was independently developed at about the same time by a
number of major companies such as du Pont and General Motors. The new form that
emerged was the decentralized or M-form structure, whose chief advantage was the clear
separation of strategic firom operational decision making. An era beginning ui the 1950s
and continuing into the 1960s witnessed the rise and growth of conglomerates fueled by
the notion fi"om "the science of management" that professional managers could run
widely diversified corporations through the application of a common set of financial
controls, capital appraisal systems, human resource management policies, and decision
rules (Grant, 1995). However, by the late 1960s, there was an increasing awareness that a
new approach to the management of diversity was needed. What became apparent was
that sound principles of organization and financial contiol, coupled with a corporate
objective of growth, was not alone sufficient to ensure satisfactory performance in highly
diversified companies.
10
With the emergence of the ideas of corporate strategy as being more than long-
range planning or objective setting, a more resti-ained view of the ability of individual
corporations to diversify across multiple, unrelated sectors emerged. Andrews (1971)
defined tiie main task of corporate level sti-ategy as identifying the businesses m which
firms would compete. Although industrial organization economics provided fi-ameworks
and models to evaluate the attractiveness of industries and how to competitively position
the firm within an industry, business consulting groups, such as the Boston Consulting
Group, were the most influential on business practices through portfolio planning
techniques. These tools provided corporate managers with a common fi-amework to
compare many different businesses. In many companies, portfolio planning became more
than analytical tools to help chief executives direct corporate resources toward the most
profitable opportunities: they became the basis for corporate strategy itself (Goold &
Luchs, 1993). However, it is important to note that "the use of these techniques in this
manner exceeded the intentions of those developing the technique/models" (Duane
Ireland, personal communication, March 18, 2001). In other words, basing corporate
strategy on these models, e.g., McKinsey 7-S firamework, far exceeded the capability of
the models' design and intended purpose.
As increasing numbers of corporations turned to portfolio planning, problems in
managing balanced portfolios became apparent (Bettis & Hall, 1983). The recognition
that different types of businesses had to be managed differently undermined the premise
that general management skills and the use of portfolio planning techniques were
11
sufficient. Many companies discovered tiiat common systems and approaches, when
applied to different kinds of businesses, could minimize, rather than maximize, the value
fi-om those businesses (Goold & Luchs, 1993).
In response to the poor financial and stock market performance of many highly
diversified companies and the disappointing results of many diversifying mergers, the
dominant theme of the 1980s was shareholder value maximization. Value-based
planning techniques gained many adherents, especially among American corporations.
The rigorous application of these tools resulted in the divestment of businesses that were
failing to create "economic value added."
The concept of corporate success based on "stick to the knitting" also gained
popularity during this time. Peters and Waterman's In Search of Excellence (1982)
suggested that successful corporations did not diversify widely. Firms tended to
specialize in particular industries and focused intently on eight principles of excellence.
Although the book became one of the most often quoted sources in the popular
management literature and many business firms reportedly attempted to conform to the
eight principles of excellence, Hitt and Ireland (1987) found that "many of the firms
Peters and Waterman designated as excellent may not have been excellent performers. In
addition, many did not exhibit adherence to the excellence precepts to a greater extent
than did a general sample of tiie Fortune 1,000 firms" (p. 96). In sum, only three of the
excellent firms performed better than the average of the general sample, and several firms
fi-om the general sample outperformed all excellent firms.
12
Two important points may be made. First, In Search of Excellence is an example
of how the fundamental prmciples of diversification can "become susceptible to the
whims of strategic fashion which dictate terms on how the strategy should be adopted, to
what extent, and in what form" (Reed & Luftinan, 1984, p. 29). Second, all types of
corporations receive all kinds of advice that is sometimes conflicting in nature and some
advice, rather than grounded in sound management principles or science, may be based
on advocacy.
Nevertheless, fi-om the mid-1980s onwards, a goal of many corporations has been
to rationalize their portfolios (refocus) to overcome the perceived disadvantages of broad
diversification (Goold & Luchs, 1993). Grant (1995) suggests that the 1990s have seen
the reemergence of the logic that drove the conglomerate diversification of the 1960s —
synergy, a concept based in part on economies of scope. However, rather than financial
synergy, value is created by the sharing of common, integrated sets of resources and
capabilities. Nevertheless, the logic of shareholder value, under the rubric of EVA,
MVA, or CFROI, has also grown in popularity in the 1990s, with over 350 corporations
usmg some variant of this metric (Myers, 1996).
Witiiin the manufactiiring sector from 1991 through 1997 (COMPUSTAT®
segment file), very moderate decreases m the level of diversification of corporations have
occurred. Diversification, as measured by the number of a corporation's unique product
markets, has decreased from 3.68 segments per corporation in 1991 to 3.26 segments in
1997 (other than single or dominant corporations). This evidence suggests that
13
corporations within tiie manufacturing sector, on average, are becoming slightly less
diversified. However, Montgomery (1994) suggests that for die 500 largest US public
companies, diversification has actually increased. For example, CEO L. Dennis
Kozlowski, in just over five years, has quietly transformed Tyco Intemational Ltd. from
an unheralded $4.5 billion company into a $29 billion multi-industry conglomerate
(Verespej, 2000). For the fiscal year ended September 30, 1999, revenues were over $22
billion and market cap has grown from $75 billion in mid-April 2000 (Kaback, 2000), to
over $78 billion as of March 18, 2001 (Quote.com, 2001).
Theoretical Perspectives
The literature on diversification spans over thirty years with genesis in multiple
business disciplines. In review, I identify and explicate the major theoretical perspectives
that pertain to diversification and, drawing from their respective premises, suggest a level
of diversification through which cost savings or revenue enhancements may be obtained
among or between a mix of business units within a corporation.
Whether a firm chooses to diversify its operations beyond a single industry or to
operate business in several industries because of intemal incentives (sfrengths or
weaknesses) or extemal incentives (threats or opportunities), the firm is pursuing a
corporate level sfrategy of diversification. Following Ramanujam and Varadarajan
(1989), diversification is defined as tiie entry of a firm or business unit into new Imes of
activity, either by processes of intemal business development or acquisition, which entail
14
changes in tiie firm's administrative structure, systems, and other management processes.
Thus, diversification is a corporate level strategy that mvolves tiie management of a mix
of businesses competing m several industries or product markets. Furtiiermore, level of
diversification is defined as tiie extent to which firms are simultaneously active in many
distinct businesses. This definition parallels tiiat for diversity (Pitts & Hopkins, 1982);
however, I use a different term to signal that the measure of diversification, in this study,
is different than that used in the literature for operationalizing diversity, which is
addressed in the research design section.
Multiple theoretical perspectives address the reasons for diversification and the
implications of the type and/or level of diversification to performance. Some
perspectives suggest that diversification is implemented to create value over and above
that which may be attained by a single stand-alone business — economies of scope
through operational synergies or financial economies; others suggest that diversification
is a means through which to gain market power, while others suggest that diversification
is implemented to reduce management employment risk or increase managerial
compensation. (See Figure 2.1 for a summary of theoretical perspectives.)
Economies of scope suggest that value may be created at the corporate level
through the selection and management of a particular group of businesses that are worth
more under the ownership of the company than as a single stand-alone busmess unit
(CoUis & Montgomery, 1998a, 1988b). Multibusiness corporations may create value
either by exploiting synergies or financial economies of scope between business units;
15
however, the level of diversification tiirough which a firm may achieve optimum
performance varies as to the theoretical perspective.
Operational Synergies
Synergy Theories argue that benefits may accrue to multibusiness units through
the sharing of activities or the leveraging of core competencies that are not available to
single stand-alone businesses. Thus, synergy exists when the value created by business
units working together exceeds the value those same units create when working
independently.
The concept, sharing of activities (Porter, 1985, 1987), with origins in industrial
organization economics and locus in the, frequently termed, industry-structure
perspective (Cormer, 1994), is based on value chain analysis. Through such an
evaluation, a firm may identify ways m which activities can be shared across several
different businesses within a diversified corporation. Porter (1985), Rumelt (1974), and
Ansoff (1965) suggest ways in which activities can be linked between and among
business units embedded within a multibusiness corporate structure. As Barney (1997)
notes, shared activities may reduce costs or enhance revenues and are quite common in
corporations that are low to moderately diversified.
Leveraging of core competencies has genesis in the resource-based view of the
firm (Wemerfelt, 1984; Dierickx & Cool, 1989; Barney, 1991; Conner, 1991; Peteraf,
1993) and resource advantage tiieoty (Hunt, 1995, 1997) and suggests tiiat revenue
16
enhancements or cost savings may be achieved through tiie sharing of less tangible
resources, such as knowledge, experience, or brand name (Grant, 1988). As Argyres
(1996) points out, because the expense of developing a core competence is a sunk cost,
and competencies based on intangible resources are less visible and more difficult for
competitors to understand and imitate, transferring these types of competencies from an
original business unit to another may reduce costs and enhance an entire firm's strategic
competitiveness.
Although research has shown that sharing resources and activities contributed to
post-acquisition performance increases and higher retums to shareholders (in the banking
industry) (Brush, 1996; Zhang, 1995), I argue that there are limits to the degree to which
the sharing of activities or the leveraging of resources can create value. Davis, Robinson,
Pearce, and Park (1992) as well as Chandler (1962, 1977, 1991) suggest tiiat tiie
coordination and managing the sharing of activities can lead to excess bureaucracy,
inefficiency, and organization gridlock as well as the loss of flexibility because of the
interdependencies between and among business units. Barney (1997) suggests that the
leveraging of core competencies may be limited by the way the firm is structured as well
as by the ability to transfer intangible assets (i.e., tacit knowledge) to other business units.
In sum, synergy theories suggest that a firm may achieve benefits from low to moderate
levels of diversification through the sharing of activities or leveraging of competencies
among its business units — up to a point, and then would be faced with higher marginal
costs respective to increased marginal benefits (Markidas, 1992). Thus, tiiis interplay
17
between synergies and limits would suggest an inverted U-shaped relationship between
the level of diversification and business unit performance (as depicted in Figure 2.1.1a.).
Financial Economies
An intemal capital market (ICM) — one that is more efficient than the extemal
capital market — is a theoretical perspective that focuses on financial advantages
associated with diversification rather than with operational synergies. Intemal capital
markets are a "natural extension of the M-form to manage less closely related activities"
(Williamson, 1985, p. 288). In such firms, business units are treated primarily as profit
centers: the prime criteria for their continuation and support are their current or future
profitability, and corporate headquarters fiinctions primarily as an intemal capital market
by which cash flows are directed to high-yield uses (Scott, 1995).
In an intemal capital market, Williamson (1975, 1979, 1985) suggests efficiency
gains are derived from the amount and quality of uiformation that corporate headquarters
possesses concerning the operations and performance of business units embedded within
its corporate structure. For example, an extemal capital market may fail to allocate
resources adequately to high potential investments, as compared to corporate office
investments, because it has limited and less accurate information. Additionally, capital
allocation can be adjusted according to more specific criteria than is possible with
extemal markets. Another advantage that accrues to firms implementing an intemal
18
capital market sfrategy is that corporate headquarters can more effectively perform such
tasks as disciplining underperforming management teams (Kochhar & Hitt, 1998).
The premise of implementing an intemal capital market to manage less closely
related activities (Williamson, 1985) suggests that as a firm becomes more diversified
(moderate to high levels of diversification), the more likely the firm may create value
through financial economies of scope. This relationship suggests that the linkage
between diversification and performance would be linear and an increasing fiinction, with
performance premiums increasing as the firm becomes more diversified as depicted in
Figure 2.1.1b, line a.
Risk-spreading is a theoretical perspective grounded in the finance literature and
suggests that the risk of engaging in two businesses will be lower as long as the retums
from the two businesses are not perfectly and positively correlated. Through
diversification into different product markets, a dowoitum in one market may be buffered
by an upturn in another. A fundamental premise of this perspective is that firms can
reduce their overall risk by engaging in muhiple businesses with imperfectly correlated
retums over time (Copeland & Weston, 1983).
This perspective is similar to the intemal capital market view in that the benefits
of diversification would accrue to those corporations that are moderately to highly
diversified. This also suggests a linear and positively increasing function between
performance and the level of diversification and is depicted in Figure 2.1.1b, curve a.
19
Market Power Economies
In a different vein from the above efficiency theories, tiie market power
perspective (Caves & Porter, 1977; Caves, Porter, & Spence, 1980) suggests tiiat firms
may create value through anticompethive economies of scope. The phenomenon of
market power in the diversification process is characterized by multibusiness firms
leveraging their size and diversity to exert market power and in tum, gain a strategic
advantage. Market power exists when a firm is able to sell its products above tiie existing
competitive level or reduce the costs of its primary and support activities below the
competitive level, or both (Shepherd, 1986).
Two common mechanisms through which firms may exert market power is
through cross subsidization or mutual forbearance in multipoint competition (Grant,
1995). Cross subsidization involves transferring resources from one business unit to
another to give it an unfair advantage in the marketplace. Hence, a firm might choose to
take heavy losses in a particular business using profits from other businesses to subsidize
the losses, in order to force competition out and enjoy higher profits in the long run.
Mutual forbearance is a relationship between two or more firms in which excessive
competition leads to a situation whereby the firms see that such competition is self-
destructive and, without formal agreement, cease the rivalry (Tirole, 1988; Gimeno &
Woo, 1999; Hitt, freland, & Hoskisson, 2001), thus allowing the respective business units
within the firm's portfolio to achieve profitability levels that exceed those of a equivalent
stand-alone business.
20
Altiiough the optimum level of diversification to exploit these perspectives is
debated in the literature, h is interesting to note that in the 1960s and 1970s, it was
claimed that conglomerates were more likely to exercise market power (Grant, 1995). I
concur and suggest that highly diversified firms, because of their size and diversity,
would be more likely to engage in market power strategies than would low to moderately
diversified firms. Therefore, the suggested relationship between market power strategies
and performance would be a linear and positively increasing function as a firm becomes
more diversified. This view is also depicted in Figure 2.1.1b, curve a.
In sum, both the financial and market power perspectives suggest a linear function
between increased diversification and performance. However, one could argue that this
relationship may be constrained or limited as to the degree to which a firm can accrue all
the benefits as it becomes more and more diversified. Therefore, a plausible linkage
between diversification and performance conceming the financial and market power
perspectives may be more of a decreasing fiinction in which the marginal benefits to
diversification is a decreasing fiinction (Markidas, 1992). This relationship is depicted in
Figure 2.1.1b, curve b.
Behavioral Motives
Still other theoretical perspectives suggest that firms may diversify for reasons
other than profit maximization. Two such theories are tiie power perspective and
institiitionalism (Hoskisson, Hill, & Kim, 1993). The power perspective (Pfeffer &
21
Slancik, 1974; Pen-ow, 1970; Pfeffer, 1981) argues that organizations must allocate
scarce resources, and it is not always apparent as to what might be tiie optimal
mechanism for such allocation. Since diversification can be viewed as a mechanism tiiat
allows for grov^ through some form of diversification strategy, its implementation
would be favored by tiiose who stood the most to gam. Therefore, managerial motives
for diversification, such as managerial risk reduction and a desire for increased
compensation (Cannela & Monroe, 1999; Finkelstein & Hambrick, 1996), may exist
independent of other incentives and resources. For example, diversification may reduce a
top-level manager's employment risk; that is, corporate executives may miplement a
diversification strategy m order to diversify their employment risk, as long as profitability
does not suffer excessively (Amihud & Lev, 1981).
Diversification also provides an additional benefit. Diversification and firm size
are highly correlated and, as size increases, so does executive compensation (Gray &
Cannella, 1997; Tosi & Gomez-Mejia, 1989). Large firms are more complex and harder
to manage and thus, managers of larger firms are more likely to receive higher
compensation (Finkelstein & D'Aveni, 1994). Govemance mechanisms may not be
strong and, in some instances, managers may diversify the firm to the point that it fails to
eam even average retums (Hoskisson & Turk, 1990). Nevertheless, it is overly
pessimistic to assume that managers will usually act in their own self-interests as opposed
to their firm's interest (Finkelstein &. D'Aveni, 1994).
22
Bamey (1997) suggests that firms motivated by a power perspective are more
likely to pursue a strategy of high diversification. However, being that the motives for
diversification are other tiian performance, it could be argued tiiat tiie effect of
diversification on performance may be neutral or even negative. Thus, I suggest tiiat the
correlation between diversification and performance is zero and that, as the level of
diversification mcreases, performance would remain at about the same level, on average.
A graph depicting this relationship would be a linear horizontal line, as shown in Figure
2.1.1c.
Institutional theory (DiMaggio & Powell, 1983; Fligstein, 1985, 1990) suggests
that organizations are likely to come to resemble one another due to pressures from their
environment; and, when organizations face uncertainty in their environment, they may
mimic other, more successful organizations. By so doing, firms may gain legitimacy, but
the adoption of a type or level of diversification does not necessarily ensure performance
gains. In reality, performance may decline. Because firms will mimic other firms that
appear successful, regardless of the level of diversification of the successful firm, the
correlation between diversification and performance would be approximately zero. As
suggested m the discussion on the power perspective, the relationship between
diversification and performance may be depicted as a linear and horizontal straight line as
graphically demonstrated in Figure 2.1.1c.
In summary of tiie theoretical perspectives on diversification, two broad
categories emerge. One category, tiie rational perspective, suggests that implementing a
23
diversification sfrategy is a means through which a firm may achieve higher rettims while
tiie otiier category, the behavioral perspective, suggests diversification is a means tiu-ough
which legitimacy or a solidification of power may be obtained ratiier tiian improving
performance.
Furthermore, of the theories embodied within the rational perspective, each
framework suggests different mechanisms and level of diversification through which a
firm may obtain a higher return. Given the breadth of theories on diversification, it may
be concluded, that on the basis of theory alone, it is difficult to come to a definitive
conclusion regarding the performance superiority of one diversification strategy over
anotiier (Setii, 1990).
Empirical Studies
Palich, Cardinal, and Miller (2000) suggest that the threat of fragmentation of
findings on the relationship between diversification and performance is great owing to the
myriad approaches and frameworks from which this research has been generated. With
this caution in mind, I turned to the empirical research to determine if some consensus
has been reached conceming the linkage between diversification and performance.
Empirical studies on diversification may be categorized into three types: those that
study the performance differences between (1) diversified and non-diversified firms, (2)
related and unrelated diversified firms (low versus high levels of diversification), and (3)
24
stand-alone, related (low/moderate diversification), and unrelated diversified firms (high
levels of diversification). (See Table 2.2 for a summary of empirical results.)
The first category of studies investigates whether or not diversification may lead
to higher performance. The findings by Weston and Mansinghka (1971) suggest tiiat
diversification does lead to higher performance; however, the difference in performance
is not significantly different. Lang and Stulz (1994) come to an ahemative conclusion
and argue that diversification is not a successful path to higher performance.
Nevertheless, Levit (1975) and Jose, Nichols, and Stevens (1986) argue tiiat diversified
firms outperform non-diversified firms with the latter authors further suggesting that the
difference in performance is statistically significant.
The second group of studies looks at diversification in a more fine grained manner
by delineating diversified firms as to their level of diversification and then compares the
retums to diversification among all categories. Rumelt (1974) suggests performance
differences exist across levels of diversification, with dominate and low to moderately
diversified firms particularly profitable. However, Bettis and Hall (1982), investigating
the performance differences in Rumelt's study, found no significant differences in
profitability once (emphasis added) they accounted for the influence of industry
(pharmaceutical industry, an industry earning above average retums). In a very recent
meta-analysis, Palich, Miller, and Cardmal (2000) found an mverted U-shaped
relationship between the level of diversification and performance, with low to moderate
levels of diversification outperforming both single and highly diversified corporations.
25
However, as a meta-analysis is a sttidy of shidies, the authors noted tiiat many of tiie
studies did not control for variables tiiat have had a demonstrated effect on firm
performance.
The final group of studies looks at only those corporations that are diversified,
without comparing the retums of diversified corporations to that of single stand-alone
firms. Again, tiie findings are mixed and even contradictory. Hoskisson (1987) and
Michel and Shaked (1984) argue that highly diversified firms are able to generate
statistically superior retums while Grant and Jammine (1988), Grant, Jammine, and
Thomas (1988) and Simmods (1990) suggest low to moderate levels of diversification do
not outperform highly related diversified firms. Although there is no consensus as to
performance benefits that may accrue to levels of diversification, other studies have
shown that diversification may be a vehicle through which mangers may reduce their
unemployment risk (Amihud & Lev, 1981), and there may be a host of variables that have
determinant effects on firm performance (Chenhall, 1984).
The findings, as presented above, support Grant's (1995) conclusion that the
inconsistency of the empirical evidence on diversification points to the impossibility of
generalizuig about the performance outcomes of diversification. Palich et al. (2000) also
concur by suggesting that, the research domain, altiiough large, has not reached maturity,
in that the field has not reached a consensus conceming the linkage between
diversification and performance.
26
Conjorate Effects: Historical and Empirical Development
The roots of the effects literature, defined as studies that take into account entire
classes of effects, date back to tiie late tiiirties witii Ed Mason (1939), tiie father of
industrial organization economics, who argued that there was a rather deterministic
association between market stmcture and profitability. However, Nourse and Dewry
(1938) suggested that influences specific "to the firm" determined performance. In otiier
words, management mattered.
These two perspectives are groxmded in different ontological and epistemological
assumptions and underpinned the either/or debate on the determinant of profitability.
Most studies in these early years, as well as into the early nineties, gave rise to what
Leventiial (1995) calls the "Holy Wars" between the "High Church" and "Low Church"
views on the relative importance of firm and industry effects. The High Church (HC)
assumes the widespread prevalence of firm rationality and market equilibrium while the
Low Church (LC) rejects rationality and equilibrium as accurate descriptions of the
competitive environment. The Low Church argues that firms are fiindamentally different
and that firm effects dominate performance. Furthermore, this debate is not only
complicated by the different philosophical assumptions, but also by the level of analysis
with tiie HC (lO) being at the industry level and the LC (strategists) at the firm level.
In the middle of the eighties, scholars began to try to resolve the debate as
reflected in tiie work of Schmalensee (1985), Scott and Pascoe (1986), Wemerfeh and
Montgomery (1988) and tiien later by Rumeh (1991). Schmalensee was one of tiie first to
27
directly address tiie issue of relative importance of firms, modeled as market share, and
industry effects by decomposing the profitability of lines of businesses using botii
hierarchical regression and variance decomposition techniques. The results of the study
suggested tiiat industry effects accounted for the largest fraction of business profitability,
18.5% to 19.5%, while market share (business effects) accounted for less than 1% of the
variance in profitability. (Due to negligible importance in the regression model, corporate
effects were not included in the variance decomposition analysis.) Schmalensee
concluded that his resuhs supported the classical focus on the importance of mdustries;
however, the unexplained variance in ROA in the line of business was approximately
80%.
Table 2.3 on corporate effects summarizes studies from this time period
(beginning with Schmalensee's study) forward and will be referred to by the author in the
remaining discussion on corporate effects. Studies included in the table are primarily
variance decomposition studies, and results are reported that are directly related to the
manufacturing sector and 4-digit SIC classifications, although other results may be
reported because of the significance to the corporate effects literature. In general, the
table reports for each study: data sources; range of years of the data set; definition of an
industry; types of industries included; definition of manufacturing sector; sizes of firms;
number of firms, businesses, and businesses per firm. Additionally, the dependent
variable of interest is reported along with the statistical techiuque employed, estimate of
28
industiy, firm and corporate effects, and whether or not an interaction term was included
within the analysis.
Scott and Pascoe (1986) addressed the issue of the relative importance of the
different effects in a somewhat novel but very important way by dividing their analysis
into two models, a null and full model. The null model operationalized the various
effects by specific variables. That is, industry effects were operationalized as
concentration, minimum efficient scale, import competition, geographic size of markets,
growth of market demand, and cost of capital. Corporate effects were operationalized as
diversification and leverage, while line of business effects were operationalized as the
share, advertising, and concentration times advertising. The full model, which the
authors termed the complete model, consisted of dummy variables for industry and
busmess effects "in addition to" the variables in the null model, termed the traditional
model. The most important finding of this study was that industry and business effects,
modeled as a class representing multiple factors (dummy variables), accounted for the
majority of the variance in profitability in the full model over that explained by the
operationalization of specific constructs; i.e., market share, concentration, economies of
scale and the like. A second important conclusion tiiat may be drawn from tiiis study was
that both business and industry factors were found to be important.
In a similar analysis to tiiat of Schmalensee, Wemerfeh and Montgomery (1988)
conducted a study of the importance of industry, market share (business effects), and
corporate effects. Corporate effects were operationalized by a corporate focus measure
29
which portrays the degree of relatedness in diversification of a corporation. Two
mteresting aspects of their model was the use of Tobin's q as the dependent variable and
the partitioning of firm effects uito Ime of business and corporate effects. Their findings
also supported the structure-conduct-performance paradigm with industry effects
explaining 10% to 20% of the variance in the market's perceived prospects for firm
profitability, while market share and corporate focus explained 0.0% to 2.3% and 0.2% to
3.7%, respectively.
In a study somewhat related to the previous studies but with more of a focus on
internally oriented organizational factors, Hansen and Wemerfelt (1989) used an
integrated model that included both economic and organizational factors. Their findings
suggested that organizational variables accounted for the majority of the explained
variance, but moreover, the two sets explained relatively independent portions of
performance.
In a seminal study, Rumeh (1991) respecifies Schmalensee's model by
decomposing line of business profitability variance within manufacturing firms over time
into corporate, business, industry, and other effects. The results of his variance
decomposition study suggested that industry and business effects were important;
however, corporate effects were quite small, accounting for 0.0% to 1.6% of the variance
in ROA. Several implications of this study were (1) the degree to which business effects
explained performance with regards to industry effects, 44% to 47% and 7% to 4%,
respectively; (2) the trivial amount of variance explained by corporate effects; and (3) the
30
size of the line of business accounted for differences in tiie amount of explained variance
of all effects (except for corporate effects because tiiey were insignificant). These results,
especially those of corporate effects, led Rumeh (1991) to comment tiiat, "h is surprising
to find vanishing small corporate effects in tiiese data given tiie extent of tiie literature on
corporate strategy, the number of corporate management consuhing firms, and tiie focus
on senior corporate leaders in the business world" (p. 182). The results, as viewed by
Rumeh as well as interpreted by otiiers (Carroll, 1993; Ghemawat & Ricart I Costa, 1993;
Hoskisson, Hill, & Kim, 1993), suggested that the relatively small size of corporate
effects to that of the other effects indicates that corporate strategy is relatively
unimportant for explaining busmess performance (Bowman & Helfat, 2001).
The empirical findings at this pomt in history had not resolved the debate between
the different schools of thought, with one paradigm arguing that their perspective was the
primary determinant in explaining firm performance while the other argued that other
factors — business and corporate factors — were important determinants of performance,
in addition to industry effects. Nevertheless, some important findings had contributed to
the continued growth in understanding of the complex phenomena conceming
organizational performance. First, it was shown that each perspective — industry and
firm specific — contributed to explaining variance in profitability and that each explained
relative "independent portions." In other words, there was the beginning of accumulated
evidence that the two perspectives were complementary rather than one perspective being
more important than the other. Second, with the decomposmg of firm effects into
31
corporate- and business-level effects, the findings suggest that corporate effects were
trivial in explaming addhional variance in firm performance relative to that explained by
mdustry and business level effects. In other words, corporate sfrategy did not matter.
From a methodological perspective, several important models laid the foundation
for future studies. First, the modeling of entire classes of effects was shown to be more
robust in explaining variance m the dependent variable rather than operationalizing
individual constructs. Second was the use of variance decomposition models. This
modeling technique enabled researchers to use a multitude of dummy variables in an
efficient and parsimonious manner, although computing capacity still limited the
researcher in modeling entire data sets.
In a replication and extension of the previous work of Rumelt (1991), Roquebert,
Phillips, and Westfall (1996) contributed to the research on corporate effects by finding
as well as supporting the importance of corporate effects as an important determinant of
firm performance in conjunction with industry and business-specific effects. In their
study of manufacturing firms, classified witiiin tiie 2000 through 3099 SIC codes, the
authors found that all three effects were unportant with corporate effects, business effects,
and industry effects explainmg 17.9%, 37.1%, and 10.1%, respectively, tiie variance in
firm performance. Addhionally, tiie authors were able to reconcile their findings with
that of Rumelt conceming the magnitude of corporate effects. Through a sensitivity
analysis, the authors demonstt-ated tiiat as tiie number of SBUs witiiin a firm increased,
tiie size of corporate effects decreased, tiius suggestuig some type of curvilinear
32
relationship. In other words, once the size of multibusiness firms was taken into account,
the results could be reconciled with Rumeh's as to the magnitude of corporate effects, hi
sum, the findings made a significant contribution to the field in tiiat tiie autiiors
demonstrated that all three effects, especially corporate effects, were important
determinants of firm performance. Furtiiermore, scholars need to consider the size of
firms contained in their study for interpreting the importance of the various effects.
Although most of the studies on corporate effects have used ROA as the
dependent variable, Chang and Singh (2000) used market share as the performance
measure. Their study, using manufacturing data and SIC as defined by four digits, found
support for industry, business, and corporate effects with business effects being the most
explanatory and corporate effects the least. Additionally, the authors, using samples that
varied by the size of the business as measured by sales or market share, found that all
effects varied by firm size, with the medium sized firms exhibiting the largest corporate
effects, with a magnitude of 25.7%.
Usually, one of two methods is used in studies that investigate mdustry, firm, and
corporate effects on tiie variance of profitability: (1) analysis of variance and (2) variance
components. However, m a unique study, Bmsh, Bromilley, and Hendrickx (1999) used
a two stage least squares model and continuous variables to estimate the influences of the
individual effects. This alternative approach found tiiat botii corporate and industiy
effects influence business unit profitability, with corporate effects having the greater
33
mfluence. Furtiiermore, they suggest that corporate headquarters influence tiie
performance of business segments.
In summary, the preponderance of the evidence (refer to Table 2.3) suggests tiiat
corporate effects are an important determinant of firm performance. Bowman and Helfat
(2001) concur and suggest, that overall, "the studies do mform the field as to the
importance of corporate effects, which encompass a much larger range of estimated
corporate effects than is commonly thought. In short, corporate strategy matters" (p. 1).
Additionally, the support for the robustness and importance of corporate effects has been
achieved through a variety of approaches. For example, scholars have used different
dependent variables, different data bases, and the use of different methodologies.
However, two studies, McGahan (1998) and McGahan and Porter (1997a),
derived conclusions that are at variance with all the other studies in the effects research.
In an attempt to reconcile findings with the rest of the field, several important differences
were discovered concemmg the research design. First, in operationalizing corporate
effects m the 1998 study, McGahan used a corporate focus measure that is, in actuality,
only a measure of the relatedness of diversification. This measure captures just one of the
set of factors that comprise corporate effects. Second, in both tiie 1997(a) and 1998
studies, the manufacturing sector was operationalized as SIC 3000, whereas the
customary practice of field uses tiie SIC 2000-3000 classification. Finally, in tiieir
1997(a) study, single business firms were included in the analysis. As Bowman and
34
Helfat (2001) argue, "m studies tiiat include single-business firms, a negHgible corporate
effect may simply reflect the proportion of single-business firms in the sample" (p. 14).
In concludmg the review of the research and empirical studies on corporate
effects, there appears to be a convergence in censuses on the importance of corporate
effects as well as industry and business effects as important determinants of firm
performance. As Bowman and Helfat (2001) suggest, the field is now ready to move
forward.
35
Table 2.1. Theoretical Perspectives and Suggested Level of Diversification to Exploit Their Respective Theoretical Premise.
Theoretical
Perspectives
Operational Synergies
lO: Sharing of activities
RBV: Leveraging of core competencies
Resource Advantage Theory
Financial Economies
Intemal Capital Market
Risk reduction
Market Power Economies
Multi-point competition (and mutual forbearance)
Cross-subsidization
Behavioral Motives
Power Perspective
Reduce unemployment risk
Maximizing compensation
New Institutionalism
Mimetic Isomorphism
Low to moderate level
of diversification Related
Diversification
X Profit maximization
X Profit maximization
X Profit maximization
Silent
Moderately high to high
level of diversification Um-elated
Diversification
X Profit maximization
X Profit maximization
X Profit maximization
X Profit maximization
X Silent on profit maximization
X Silent on profit maximization
Silent
Adapted fi-om Bamey, 1997.
36
la. Operational Synergy Theories.
N o n - m o D o t o n i c M o d e l
S t a n d - a l o o a
U n r a l a t a d
L a v a l o f O l v a r t l d c a t l o n
lb. Financial and Market Power Theories.
( a ) L I D c a r M o d e )
( b ) D e c r e a s i n g F a n c t i o o M o d e l
(a)
(b)
L a v a l o r D I v a r a l f l c a t l o n
Ic. Behavioral (Power Perspective and Institutional) Theories. H o r i z o o t a l M o d c l
S ta n d -a lo n e U n re la l e d
C o r r e l a t i o n B e t w e e n D e g r e e o f D iv e r s i f i c a t i o n a n d P e r f o r m a n c e A p p r o a c h e s 0
L a v a l o f O l v a r t l f l c a t l o n
Figure 2.1. Graphical Representations of Diversification Theories.
37
Table 2.2. Empirical Findings on tiie Diversification-Performance Linkage.
Diversification versus No Diversification
Weston and Mansinghka, 1971
Lang and Stulz, 1994
Levit, 1975
Jose, Nichols, and Stevens, 1986
Stetz and Phillips, 2000
Performance, measured by ration of net income to net worth, is somewhat higher for conglomerate firms, but the difference is not statistically significant.
Strong evidence that highly diversified firms are consistently valued less than specialized firms. ''Evidence supports that diversification is not a successful path to higher performance.''''
Diversification outperforms no diversification.
Diversification has a statistically significant and positive influence on the value of the firm.
Diversification outperforms no diversification. Differences are highly significant. Controlled for industry, corporate, and business effects.
Related versus Unrelated Diversification Low to Moderate and Moderate to High Diversification
Grant and Jammie, 1988
Grant, Jammine, and Thomas, 1988
Galbraitii et al., 1986
Michel and Shaked, 1984
Amihud and Lev, 1981
Related diversification does not outperform unrelated. Controlled for industry effects.
Among large British manufacturing firms, profitability is positively related to both product diversification and multinational diversification. The principle direction of causation runs from profitability to diversification. No significant differences exist between related and unrelated diversification strategies.
Unrelated diversification most valuable m uncertain settings.
Firms diversifying uito unrelated areas are able to generate statistically superior performance over those with businesses that are predominately related.
Managers engage m conglomerate mergers in order to reduce their unemployment risk.
Adapted from Bamey, 1997; Ramanujam and Varadarajan, 1989.
38
Table 2.2. Continued.
Related versus Unrelated Diversification (Continued)
Chenhall, 1984
Simmods, 1990
Hoskisson, 1987
For Australian manufacturing enterprises, a multivariate relationship is uncovered between the extent of a firm's diversification and a host of environmental, market structure, organizational, and managerial variables.
Related diversification does not outperform unrelated.
The implementation of the M-form stmcture increases the rate of retum of firms that diversify through an unrelated business strategy, but decreases the rate of retum of firms that adopt vertically integrated and related busmess approaches to diversification. Risk or variability of firm rate of retum generally decreases after the M-form restmcturing regardless of the diversification strategy a firm has implemented.
Stand-Alone versus Related versus Unrelated Diversification
Palich, Cardinal, and Miller, 2000
Rumelt, 1974
Bettis and Hall, 1982
Rumelt, 1982
Meta-analysis that found inverted U-shaped relationship with related diversifiers outperforming both single and unrelated diversified corporations. Most of the studies did not control for firm size nor industry effects of other determinants of firm performance.
Performance differences between single, dominant, related, and unrelated product firms, with dominate and related strategies particularly profitable.
Investigated performance differences m Rumelt's study. Upon accounting for the influence of industry (pharmaceutical), they found no statistical significant difference in profitability.
Even after adjusting for industry effects, a declming profitability premium is associated with increasing diversity.
Adapted from Bamey, 1997; Ramanujam and Varadarajan, 1989.
39
Table 2.3. Review of Corporate, hidustry, and Business Effects Sttidies.
Study Database Years
Industry Definition Types of Industries Definition of a Business Firm Size Number of Firms Number of Industries Number of Businesses Number of Businesses /Firm Dependent Variable Statistical Technique
Corporate Effect Business Effect Industry Effect Year Effect Industry x Year Otiier Effects
Unexplained Variance
Schmalensee (1985) FTC Line of Business (LOB) 1975
LOB 3'/2 digit SIC
Manufacturing only
All Co. business in each LOB category
Market share > 1% 456
242
1,775
Avg. = 3.89
ROA per business
i) OLS - hierarchical regression (ANOVA) ii) variance components i) negligible ii) not included
Market share effect: i) 0.2% to 0.6%; ii) 0.6%
0 18.8% to 19.3% ii) 19.5%
Not included
Not included
Interactions i) negative covariance - business and industry suggested ii) covariance business and industry-0.6% Unexplained Variance: 80%
Wemerfelt and Montgomery (1988) Trinet/EIS; FTC; other sources 1976
2 digit SIC
Industrial and utility Cos. American manufacturers
All Co. business in each LOB category
Not given 247- overall sample fi-om which data was drawn
Not reported
Not reported
Not reported
Tobin's q per company
OLS - hierarchical regression (ANOVA) Recorded incremental contributions to R-
Corporate focus (relatedness): 0.2% to 3.7%
Market share effect: 0% to 2.3%
10.9% to 20.1%
Not included
Not included
Not included
Unexplained variance: 77% (estimated) unadjusted for intangible assets
Adapted from Bowman and Helfat, 2001.
40
Table 2.3. Continued.
Study
Database Years Industry Definition Types of Industries Definition of a Business Firm Size
Number of Firms Number of Industries Number of Businesses Number of Businesses /Firm Dependent Variable Statistical Technique Corporate Effect
Business Effect
Industry Effect
Year Effect Industry x Year Otiier Effects
Unexplained Variance |
Rumelt (1991)
FTC LOB 1974-1977 LOB 3'/2 digit SIC
Manufacturing only
All Co. business in each LOB category
Sample A: market share > 1% Sample B: market share > 0 A: 457 B:463
A: 242 B:242
A: 1,774 B: 2,810
Minimum = 1 A: Avg. = 3.88 B: Avg. = 6.07
ROA per business
i) sequential analysis of variance ii) variance components
0 A: 14.8% to 17.6% B: 10.9% to 11.6% ii)A:«0% B: 1.6%
i) A: 33.9% to 34.0% B: 41.3% to 41.4%
ii) A: 47.2% B: 44.2% i) A: 15.3% to 17.9%
B: 9.8% to 10.3% ii) A: 7.3% B: 4.0% i) A: 0.0% B:0.1% ii) A: 0.0% B: 0.0%
i) A: 9.6% to 9.8% B: 6.8% to 7.1% iO A: 8.9% B: 5.3%
Interactions: i) Not included ii) covariance of industry and corporate A: 0.76% B: 0.0%
Roquebert, Phillips, and Westfall (1996) COMPUSTAT® 1985-1991 4-digit SIC (broadly defined)
Manufacturing only SIC 2000-3000
All Co. business in each SIC code
+/- 3 Std. Deviations of Mean ROA
94-114 in each sample (10 samples)
223 - 266 in each sample (10 samples)
387-451 in each sample (10 samples)
Minimum = 2 Avg. = 4.01
Ratio of operating returns to tangible assets
Variance components
17.9% (avg. across samples)
37.1% (avg. across samples)
10.1% (avg. across samples)
0.4% (avg. across samples)
2.3% (avg. across samples)
None
Unexplained variance: 32.2% (avg.)
Adapted from Bowman and Helfat, 2001.
41
Table 2.3. Continued.
Sttidy Database Years Industry Definition Types of Industries Definition of a Business Firm Size Number of Firms Number of Industries Number of Businesses Number of Businesses /Firm Dependent Variable Statistical Technique Corporate Effect Business Effect Industry Effect Year Effect Industry x Year Other Effects Unexplained Variance Adapted from
McGahan and Porter (1997b) COMPUSTAT® 1981-1994 4-digit SIC (broadly defmed)
Non-financial
All Co. business in each SIC code
Sales and Assets > $10 million 7,793
668
13,660
Minimum = 1 Avg. = 1.75
Operating income/tangible assets
OLS - hierarchical regression (ANOVA)
8.8% to 23.7%
32.5% to 59.1%
6.9% to 16.3%
0.2% to 1.1%
Not included
None
McGahan and Porter (1997a)
COMPUSTAT® 1981-1994 4-digit SIC (broadly defined)
Non-financial; Results are reported only for manufacturing SIC 3000
All Co. business in each SIC code
Sales and Assets > $10 miUion 2,432 Diversified: 836
219
4,068
Not broke out for manufacturing
Operating income/tangible assets
Variance components (COV)
Manufacturing sector: N/A Across all sectors: 4.33
35.45%
10.81%
2.34%
Not included
Interactions: i) covariance of industry and corporate:-2.27%
Unexplained variance: 53.67%
Bowman and Helfat, 2001.
42
Table 2.3. Continued.
Sttidy Database Years Industry Definition. Types of Industries
Definition of a Business Firm Size
Number of Firms Number of Industries Number of Businesses Number of Businesses/Firm Dependent Variable Statistical Technique
Corporate Effect
Business Effect
Industry Effect
Year Effect
Industry' x Year Otiier Effects Unexplained Variance
McGahan (1998)
COMPUSTAT® 1981-1994
4-digit SIC (broadly defmed) Non-financial and manufacturing defined as SIC 3000 Business Segment and Corporation
Sales and assets > $10 million Financial-market > $50M [Corporate] 4,947 648
Minimum = 1 Avg. 1.6
Tobin's q and ratio of operating retums to tangible assets
OLS hierarchical regression (ANOVA) Permanent and transient effects Estimates are increases in Adj. R (Corporate focus) Permanent Transient Accounting profit .00 .00 Tobin's q .00 .00 Replacement value .00 .00 Business effects Permanent Accounting profit .319 Tobin's q .408 Replacement value .340
Permanent Transient Accounting profit .125 .189 Tobin's q .293 .091 Replacement value .157 .145 Accounting profit .017 Tobin's q .021 Replacement value .033 Not mcluded None
Kessides (1990) FTC LOB 1975 LOB 3'/2 digit SIC Manufacturing only
All Co. business in each SIC code
Market share > 1%
456 242
1,775
Avg. = 3.89
In(l-ROS) per business
Weighted least squares with a mix of fixed and random effects - hierarchical regression (modified ANOVA) 5.1% to 9.8%
Market share effect: 6.6% to 27.5%
4.7% to 25.2%
Not included
Not included None
Adapted from Bowman and Helfat, 2001.
43
Table 2.3. Continued.
Sttidy Database Years Industry Definition
Types of Industries Definition of a Business
Firm Size
Number of Firms Number of Industries. Number of Businesses Number of Businesses/Firm Dependent Variable Statistical Technique
Corporate Effect
Business Effect
Industry Effect
Year Effect
Industry X Year
Otiier Effects Unexplained Variance
Bercen-a (1997)
COMPUSTAT® 1991-1994
4-digit SIC (broadly defined) + classified by broad world geographic area None excluded All Co. business in broad world geographic area (manufacturing Cos. only) Within largest 100 US Cos. in 1994
41 11 industries, 5 geographic areas 134 Minimum = 3, maximum = 5 Avg. = 3.27
ROA per business i) hierarchical regression (ANOVA) ii) variance components iii) repeated measures random factors (ANOVA) i) 12% ii) 4.71% iii) 3.05% to 10.95% a) not reported b) 27.2% c) not included
Industry Geographic Area ii) 30.4% 6.9% iii) 41.9% to 46.8% 0% to 1.1% i) not reported ii) not included iii) not significant i) not included ii) not included iii) significant i) none ii) none iii) year x corporate significant
Chang and Singh (2000) Trinet/EIS 1981, 1983, 1985, 1987. 1989 4-digit SIC (narrowly defined) (reporting only 4-digit results)
Manufacturing only All Co. business in each 4-digit SIC code Small = 2 to 170 M, Medium = 171 to 893 M, Large = 893 M to 121 B Sample A: mkt share > 1% Sample B: $2 miUion to $2 billion sales Sample B includes Businesses < 1% MS A: 475 (4-digit) B: 693 (4-digit) A: 374 (4-digit) B: 390 (4-digit) A: 1,531 (4-digit) B: 3,070 (4-digiO Minimum = 1 A: Avg. = 3.22 (4-digit)
B: Avg. =4.43 (4-digit)
Market share per business Variance components
Sample A: 4.3% B: 8.5% Sample B: Firm Size Large: 10.9%, Medium: 25.7%. Small: 6.3% SIC: (4-digit) A: 52.7% B: 46.8% Sample B: Firm Size Large: 44.4%, Medium: 15.8%. Small: 15.6% SIC: 4-digit A: 19.4% B: 25.4% Sample B (4-digit SIC) Firm Size Large: 24.1%, Medium: 40.6%, Small: 59.4% SIC: 4-digit A: 0.9% B: 0.3% Sample B (4-digit SIC) Firm Size Large: 0.7%. Medium: 0%, Small: 0% SIC: 4-digit A: 0.9% B: 1.8% Sample B (4-digit SIC) Firm Size Large: 1.3%. Medium: 6.9%, Small: 12.5% None
A: 21.8% B: 17.2%
Adapted from Bowman and Helfat. 2001.
44
Table 2.3. Continued.
Sttidy Database Years Industry Definition Types of Industries Definition of a Business Firm Size
Number of Firms Number of Industries Number of Businesses # of Business /Firm Dependent Variable Statistical Technique Corporate Effect
Business Effect
Industry Effect
Year Effect Ind. X Yr. Otiier Effects
Unexplained Variance
Brush, Bromiley, & Hendrickx (1999) COMPUSTAT® 1986-1995
4-digit SIC (broadly defined)
Non-financial
All Co. business in each 4-digit SIC code
Multibusiness with 3 segments ($1.05 B) and 4 business segments ($1.96 B), on average. No specific size limit was reported 3 segments: 535 4 segments: 173
Not identified (Used industry ROA rather than SIC code)
3 segment: 1,605 4 segment: 692
Exactly 3 and 4, respectively
Business segment ROA
Two staged least squares
Ratio of corporate effect to industry effect is 1.7 for standardized coefficients
Incremental R of business effects dominate corporate and industry effects
Ratio of corporate effects to industry effects in terms of R and R is greater than 1 not included not included
Stetz and Phillips (2000)
COMPUSTAT® 1991-1997 4-digit SIC (broadly defined)
Manufacturing only SIC: 2000 - 3999
All Co. business in each SIC code
+/- 4 std. deviations of mean ROA
2,342
557
3,849
Minimum = 1 Average = 1.64
Ratio of operating profit to identifiable assets
Linear mixed model with fixed and random effects
Null Model: 34.83 or 7.66% Full Model: 32.05 or 7.12% Change in parameter estimate: 2.78 Null Model: 225.000 or 49.51% Full Model: 225.001 or 49.98% Change in parameter estimate: .001 Null Model: 26.720 or 5.88% Full Model: 25.213 or 5.60% Change in parameter estimate: 1.507 not included not included Residual: Null: 167.88 or 36.94%
Full: 167.61 or 37.31% Change in parameter estimate: 0.27 Unexplained variance: 37.31%
Adapted from Bowman and Helfat, 2001.
45
CHAPTER m
HYPOTHESES
The lack of consensus conceming tiie linkage between diversification and
performance, as noted by Palich, Cardinal, and Miller (2000), may be best summarized by
comments from three leadmg scholars. Markides and Williamson (1994) suggest little
direction may be gleaned as how and when diversification can be used to build long-run
competitive advantage and Grant (1995) concludes tiiat tiie inconsistency of the empirical
evidence on diversification points to the impossibility of generalizuig about the
performance outcomes of diversification. This inconsistency has dfrect implications for
this study's research design and formulation of hypotheses.
In the classical approach of hypotheses construction and verification (Boal &
Willis, 1983) theoretical conceptualizations and empirical testing are considered to be at
different levels and are linked through three stages of development. In Stage 1, concepts
are defined and propositions offered; m Stage 2, measurement of concepts are devised
and testable hypotheses are suggested (theory of measurement); and finally, in Stage 3,
data are gathered and analyzed and inferences drawn (theory of testing). It may be argued
that the lack of consensus conceming diversification and performance has origins m all
three stages. For example, Palich, Cardinal, and Miller (2000) suggest that the threat of
fragmentation of findings on the relationship between diversification and performance is
great owing to the myriad approaches and frameworks from which this research has been
46
generated. Given the depth and breath of tiieoretical perspectives that have genesis in
multiple business disciplines, it also could be argued tiiat tiie measurement of tiieoretical
constructs are laden by each perspective's theoretical lens (Stage 2). Hoskisson and Hitt
(1990) suggest as much when they concluded tiiat the confusion regarding the
diversification-performance relationship is partially theoretical and partially
methodological, although both are mextricably woven because the methods employed to
measure diversification often are associated with a specific theoretical perspective.
Conceming Stage 3, scholars have voiced their concem, not so much in the testmg of the
relationships between diversification and performance, but in controlling for variables
that have a demonstrated determinant on business unit performance. Furthermore, as
Palich et al. (2000) suggest, accountmg for these variables in future research will further
aid in the understanding of the linkage between diversification and performance.
To address the concems of Stage 3, this study draws from the corporate effects
literature (Bowman & Helfat, 2001) and the tiieory of linear models (Littell, Milliken,
Sfroup, & Wolfinger, 1996). The empirical research on corporate effects has
demonsfrated that not only is industry a determinant of firm performance, but also
corporate and business effects are as equally or more important. In sum, this line of
research has empirically demonstrated that m the investigation of business unit
performance, researchers need to control not only for mdustry, but also for corporate and
business effects as well (theoty of testing).
47
The theory of Imear models suggests a modeling techruque, general linear mixed
models, tiirough which a researcher may model tiie phenomena of interest while
accounting for various effects m tiie estimation of means (or relationships) and standard
errors. This technique can represent very complex, muhilevel phenomenon
parsimoniously, with only a few variance components.
Other sources of inconsistency in empuical research at Stage 3 have been the use
of small samples or the inadvertent selection of firms m superior industries in terms of
higher ROAs (Christensen & Montgomery, 1981). To this end, tiiis study utilizes a
sample of over 19,500 observations that spans seven years and includes an entire sector of
tiie economy (Manufacturing: SIC 2000-3999).
To address the epistemic relationships of Stage 2, this study identifies and utilizes
measures of diversification that have established psychological properties of objectivity,
reliability, and validity. An in-depth discussion of the data, measurement of concepts,
confrol variables, and the modeling technique is presented in the research design section.
Chapter W.
From a philosophy of science pouit of view (Boal & Willis, 1983), where there is
mconsistency between theoty and data, addressing the empirical limitations of past
research is of primary concem before one begins to mvestigate the pattems within the
data (Johnson, 1981), or theory testing. Having presented the research design, in brief, on
how these limitations are addressed and prior to presenting the formal hypotheses, the
48
constiiicts of diversification and performance are defined and tiie level of analysis is
explicated.
Most empirical research examinmg tiie value of diversification explores tiie
linkage between economic performance and the level of diversification at tiie corporate
level of analysis. However, witiiout comparing tiie rettims to diversification to business
units operating witiiin a corporation's govemance system to the retums of stand-alone
businesses or to other business units embedded m other diversified corporations (Bamey,
1997), it could be argued tiie analysis can not directly address the most fundamental
question underpinnmg the research on diversification, "Do corporations improve business
performance?" (Rumeh, Schendel, & Teece, 1994; Porter, 1987; Bowman & Helfat,
2001). This question echoes Bamey's observation and implies that a more appropriate
level of analysis may be to focus on the business unit (BU). Therefore, to answer their
call, the level of analysis for this study is the business unit.
A business unit may be defined as company operations contained within an
industry, whether m a single-business or a multiple-business firm (McGahan & Porter,
1997a) and a firm may have more than one business unit in the same industry. This
definition, is in effect, the same as that for a business segment, a reporting criteria
required by the FASB; however, Grant (1995) notes that segment usually refers to
product markets within an industry rather than company operations in product markets.
Therefore, to avoid confusion in using the term segment, I used the term business unit.
49
Performance is defined as the level of profitability of a business unit and is
measured by operating profits divided by identifiable assets. Diversification may be
defined as the entry of a firm or business unit into new lines of activity, either by
processes of intemal business development or acquisition, which entails changes in its
administrative structure, systems, and other management processes (Ramanujam &
Varadarajan, 1989). For this study, diversification is defined as the level of activity of
business units embedded within an enterprise engaged in providing a product or service
or a group of related products and services primarily to unaffiliated customers (i.e.,
customers outside the enterprise). Diversification is measured through a combination of
Rumelt's typology and an entropy measure which delineates corporations as to their level
of activities in different products or product groups that are provided to unaffiliated
customers for profit. Data, for operationalizing the above measures, is provided in the
COMPUSTAT® Segment File (COMPUSTAT II®, section 2, p. 2) which records
segment data m accordance with the reportmg criteria of the FASB 14, paragraph 10a
(Davis & Duhaime, 1989).
Having addressed the operational level of hypotheses formulation and
verification, I now tum to the conceptual level. The theoretical perspectives on
diversification may be broadly classified into rational and behavioral theories. Rational
theories suggest diversification is a means through which a firm can gam economies of
scope through either operational synergies, tiie leveraging of market power, or through
financial synergies. Implementing a diversification strategy, based on achieving
50
economies of scope, enables a firm to either reduce costs or increase revenues which
resuhs in the respective business units eaming higher rates of retums over that of single
stand-alone businesses. Conversely, the behavioral theories suggest diversification is a
means through which a firm may gain legitimacy or, for those in power, to protect or
enhance their own position. The resuh in performance is undetermined, with the
likelihood that performance will remain the same, or possibly, will slightly decline. In
sum, these broad categories of theories may be distinguished as to their suggested impact
of diversification on firm performance. A rational perspective argues that the
implementation of a diversification strategy will enable a firm to achieve higher
profitability over that that may be attamed by a single stand-alone business, ceritas
paribus, while the behavioral perspective suggests performance would remain
approximately the same as a firm diversifies, ceritas paribus. Based on this basic
demarcation, the first hypothesis investigates the question, "Does diversification improve
performance?" by comparing the retums of non-diversified corporations (single stand
alone businesses) to that of business units that are embedded withm diversified firms.
Hoi: While controlling for industry, corporate, and business effects, there is no difference between business unit profitability of diversified and non-diversified firms within the manufacturing sector.
Hoi: M'Non-diversified ~ M-Diversified ^ h c r e :
H = the mean ROA of a business unit
ff a statistically significant difference in performance between diversified and
non-diversified corporations is substantiated, two hypotheses are developed to further
explore the rational theories (which suggest that performance benefits accrue to
51
diversified firms) at a more fine grained level of specification. The broad classification of
rational theories can be further subdivided into two subcategories. Each
subclassification, given that diversification is implemented in accordance with its
premises, such as planning and control and organizational stmcture (Hitt, Hoskisson, &
freland, 1990), suggest that a given level of diversification may be as optimal as another,
that is, low levels of diversification may be an equally viable path to increased
performance as high levels of diversification, albeit for different reasons. For example, if
one takes a synergistic perspective, theory would suggest firms pursue a low level of
diversification to achieve performance benefits while the financial perspective would
suggest that a high level of diversification is needed to maximize retums. Thus, although
the different theoretical perspectives differ as to the level of diversification through which
to achieve high performance, it could be argued there is no definitive conclusion
regarding the performance superiority of one diversification strategy over another (Seth,
1990). Therefore, the second hypothesis investigates the question, "Do performance
differences accrue to business units that are embedded within multibusiness corporations
of varying levels of diversification?" This general hypothesis generates ten specific
hypotheses and statistical tests.
Ho2: While controlling for industry, corporate, and business effects, no difference exists among business unit profitability affirms across the spectrum of diversified corporations within the manufacturing sector.
Ho2: ^1 = ^2 = M3 = M4... = mc; where:
^ = the mean ROA of a business unit within a MBC, and k = 1 to 5 and equals the number of groups of corporations
that are categorized according to their level of diversification.
52
The above hypothesis investigates business unit performance among diversified
corporations; the next step is to investigate the degree to which business units within
diversified firms are eaming a retum over and above that which may be attained by non-
diversified corporations. Therefore, the third hypothesis investigates the question, "ff, at
any level of diversification, business units within diversified firms eam a high ROA over
that attained by nondiversified corporations?" This specifically addresses the question if
corporate strategy creates value by making business units, that are embedded within
multibusiness corporations, better off. This general hypothesis generates five specific
hypotheses and statistical tests.
Ho3: While controlling for industry, corporate, and business effects, there is no difference in profitability between business units embedded within diversified firms and nondiversified firms within the manufacturing sector.
Ho3: | Non-divers,f.ed = I D 5 where:
|i = the mean ROA of a business unit, and D = levels of diversified corporations in which business
units are embedded, ranging fi-om dominate to highly diversified corporations (1 through 5).
The hypotiieses, developed m tiiis chapter, first investigate the phenomenon of
diversification from tiie broadest of perspectives and then narrow the scope to mvestigate
the linkages between performance and diversification among diversified firms, only, and
finally, between diversified firms and single stand-alone business while controllmg for
industry, corporate, and business effects within tiie manufactiiring sector. I now develop
and explicate tiie research design of tiiis study in the following section. Chapter IV.
53
CHAPTER rV
RESEARCH DESIGN
The linkage between diversification and performance has received considerable
attention within tiie strategic management literature over tiie last three decades
(Chatterjee 8c Wemerfelt, 1991); however, scholars have not reached an empirically
informed consensus (Palich, Cardinal, & Miller, 2000) as to tiie performance benefits that
may be derived from types and/or levels of diversification. This assessment is not new to
tiie field as evidenced by Reed and Luf&nan's (1986, cited in Hoskisson et al., 1993)
comment that "although explanations abound, confusion has grown concemmg the nature
of the diversification- performance relationship" (p. 215).
In the examination of the efficacy of diversification with respect to profitability,
Hoskisson and Hitt (1990) concluded that the confusion regarding the diversification-
performance relationship is partially theoretical and partially methodological, although
both are inextricably woven because the methods employed to measure diversification
often are associated with a specific theoretical perspective. Acknowledging the
importance of operationalizing constructs by measures that have been assessed as to their
objectivity, reliability, and validity, diversification is operationalized through the use of
Rumelt's typology and a refined entropy measure, an objective measure with established
psychometric properties.
54
Palich et al. (2000) examined over thirty years of research on diversification and
in their meta-analysis of over fifty studies commented that "very few of tiie studies
accounted for the impact of industry; firm size; firm leverage; and advertismg, capital,
and R&D intensities; each of which have demonstrated effects on performance in prior
research" (p. 169). The authors conclude "that adjusting or accounting for these variables
in fiiture research may aid the clarification of diversification-performance relationships"
(p. 169). To address these criticisms of past research efforts, this study, drawing from the
research stream on corporate effects, both past (Rumelt, 1991; Roquebert, Phillips, &
Westfall, 1996; Chang & Singh, 2000) and emergent (Bowman & Helfat, 2001)
demonsfrates how many of these variables may be controlled for in a parsimonious
model.
In addition to measurement and control issues, past studies also have been
plagued by small sample sizes as well as the inadvertent selection of firms in superior
industries in terms of higher ROAs (Christensen & Montgomery, 1981). Therefore, tiiis
study uses a very large sample (over 19,599 observations) that encompasses over three
thousand corporations from the manufacturing sector with operations in over 550 unique
industries. Additionally, tiie research design addresses a call for future research by
conducting the study at the business unit level of analysis.
55
Level of Analysis
Most empirical research examining the value of diversification explores the
economic performance attributed to diversification at the corporate level of analysis
without comparing the retums of diversification to business units operating within a
corporation's govemance system to tiie retums of stand-alone businesses (Bamey, 1997).
It could be argued that the most fundamental question underpmning the research
on diversification is, "Do corporations improve business performance?" (Bowman &
Helfat, 2001; Rumelt, Schendel, & Teece, 1994; Porter, 1987). This question echoes
Bamey's observation and implies that a more appropriate level of analysis may be to
focus on the business unit (BU). Therefore, to directly mvestigate the question if
corporations make busmesses better off (Porter, 1987), this study's research design
focuses on the business unit level of analysis. This focus has two major research
advantages in that it allows for the assessment of the effects on business unit performance
of (I) corporate sfrategy; i.e., sfrategic choices conceming the domain and scope of the
business unit, and (2) a business unit competing with other business units across the
spectrum of diversified corporations.
Investigating the effects of corporate sfrategy and competition on individual
business performance (business unit level of analysis) requfres a more disaggregated data
set than consolidated financial statements; therefore, I used the COMPUSTAT® Industry
Segment File data set covering the years from 1991 to 1997.
56
Data
The Business hiformation COMPUSTAT® hidustry Segment breaks out
individual corporate activities by business segment — if a segment contributes greater
tiian 10 percent or more to consolidated sales, operating profits, or assets. Each
individual corporation within the database may contam from 1 to 10 individual business
segments and is identified by an overall corporate SIC code. Additionally, an SIC code is
also assigned to each busmess segment that may exist within the corporate umbrella. The
Financial Accounting Standards Board (FASB) issued Statement of Financial Accounting
Standards No. 14, Financial Reporting for Segments of a Business Enterprise (SFAS No.
14) (COMPUSTAT®, 1997), which requires this disclosure of uiformation.
In accordance with my definition, a business unit (BU) is a unique business
segment that is identified both by an SIC code and a title (other than tiie name of tiie
corporation). The BU may be a stand-alone business (a corporation with only one
business imit and is, in essence, the corporation even though separate names may exist for
each) or may be part of a set of business imits aligned under a corporate umbrella. This
definition is, in effect, the same as that for a business segment, a reporting criteria
requfred by the FASB; however. Grant (1995) notes that "segment" usually refers to
product markets within an industty. Therefore, to avoid confusion in using the term
segment, I used the term business imit.
The sample was restricted to corporations identified withm the manufacturing
sector (SIC codes 2000-3999), and after unusable observations were deleted, I screened
57
tiie sample in tiie following ways. First, if sales or assets were equal to zero or less than
zero, tiie observation was deleted. Additionally, if absolute sales or identifiable assets
were equal to 0.001, these observations were also deleted. Second, to eliminate any
undue influence from outliers m the analysis, business segment ROA (calculated by
dividing operating profit by identifiable assets) was standardized and any observations
greater than +/- 4 standard deviations from the mean were deleted. Third, if the average
of the assets or sales over time of the business segment equals less than ten million, these
observations were also deleted. Fourth, if a segment did not contain a primary SIC code,
or if corporate headquarters was used as a segment name, these observations were also
eliminated. Finally, if a particular business segment appeared for only one year in the
data set, these observations were also eliminated due to the common practice that a
segment may be newly formed m a particular year solely for divestiture. In sum, the
sample resulted in 19,725 usable observations that contained 2,341 unique corporations,
3,838 imique busmess imits operating in 589 unique industry sectors. The mean ROA of
all business segments was 9.67%.
Diversification Measures
The body of research on diversification contains muhiple measures for
diversification, such as the diversity measure (Varadarajan & Ramanujam, 1987) and the
Herfindahle measure; however, it could be argued tiiat the most often used means to
58
classify firms conceming tiieir level of diversification is tiirough the use of either tiie
entropy measure or Rumeh's typology.
Rumelt's Typology
hi 1974, Rumeh expanded upon a typology developed by Wrigley (1969) that
delineates corporations as to their level or type of diversification (Figure 3.1). Limited
diversification encompasses single businesses (95% or more of revenues are generated
from a single business) and dommate firms (10% to 95% of revenues generated from a
single business) while moderate to very high levels of diversification include firms that
may be classified as related-constrained, related-linked, and unrelated diversified
corporations, respectively. Levels and types of diversification are interrelated, and the
rationale is that as the type of diversification changes, the level of diversification changes
(Hoskisson et al., 1993). For example, dominant firms are less diversified than related
firms and related firms are less diversified than unrelated firms.
Being that the operationalization of the Rumelt measure is relatively subjective,
especially for the related-link and related-constrained, and in response to other subjective
measures used to operationalize diversification, the field has sought other alternatives that
are more objective m nature, such as the entropy measure.
59
Enfropy Measure
The entropy measure was originally developed by Jacquemin and Berry (1979)
and has been used by strategy researchers m response to the need for an objective
measure tiiat addresses strategic differences. The number of scholars using this measure
for strategy research has grown significantly (Amit & Livnat, 1988a, 1988b; Baysinger &
Hoskisson, 1989; Palepu, 1985), and was the most operationalized measure in studies
investigated in Palich et al.'s (2000) meta-analysis. Additionally, the entropy measure is
approached from tiie business level of analysis, which lessens the potential aggregation
problem that may arise at the corporate level of analysis and is thus, aligned with this
study's level of analysis.
In 1995, Raghuanathan refined the entropy measure to improve its precision. The
measure was modified to reflect its strategic dimensions — the extent of diversification
across segments (distribution) and the number of segments in which a firm operates. The
author suggests the refinement helps to delineate equivalence among firms with different
diversification profiles. Level of diversification is defined as a two dimensional
construct, the two dimensions being the number of businesses and the distribution among
those busmesses. Whichever level of diversification a firm may adopt, Raghunathan
(1995) states "what matters is how the spread of the business base is managed in terms of
the number of segments and the distribution of the resources across those segments" (p.
990). Thus, the refined measure helps to distmguish firms in a study when they are
diversified across and within mdustries. The refined entropy measure, used in this study.
60
is tiie total diversification score (TDS) which is an integrated fimction ratiier than just the
additive effects of related and unrelated scores. The equation for TDS is as follows:
Total Diversification score = { [ I S P. * hi(l / P ) ] / [hi(M) + Z P * hi(N.)]} *(N* M)
where: Py = proportion of firm's total operations within tiie ith business of jth industry;
P,j ;t 0; Pj = proportion of firm's operations within jth industry;
M = total number of industries;
N = total number of busmess;
Nj = total number of business within jth industry;
M
N = average of businesses within industries = ( 1 ; N ) / M; and
N *M = number of segments.
The TDS measure operationalizes diversification as a continuous variable. I
acknowledge that level and type of diversification are conceptually distinct; however, h is
common for researchers to convert measures of type of diversification into continuous
data representing levels of diversification (Denis, Denis, & Atulya, 1997; Hoskisson et
al., 1993; Lubatkin, Merchant, & Srinivasan, 1993).
Although empirically research consistently indicates that type of diversification is
sfrongly associated with continuous data representing levels of diversification, Galuiuc
and Eisenhardt (1994), in a review addressing the strengths and shortcomings of the
sfrategy-structure-performance paradigm and drawing from research at the intracorporate
level of analysis, suggest that corporate strategy, in their assessment of Gupta and
61
Govindarajan's (1986) work, is a ''portfolio of separate strategic business unit strategies''
(emphasis added) ratiier tiian "an overall and simple diversification strategy" (p. 227). hi
otiier words, a corporation may have a mix of strategy types, such as prospectors,
defenders, and tiie like, which suggests tiiat one overall pattern of complimentary or
interdependence (Bariet & Ghoshal, 1991), or independence among the business units
embedded within a corporation may not exist. To put this in terms of type, some busmess
units within the corporation may be related-consfrained, while others may be totally
unrelated to any of the other business units that are embedded within the same corporate
umbrellas. Therefore, because of the possible mix of strategies contained within a
corporate stmcture, I suggest it is more meaningful and parsimonious to measure
diversification by level rather than by type. In this vein, the entropy measure was found
to be the most effective in the identification of diversification levels, which is the primary
focus of this study.
In comparing the degree of association between Rumelt's measure and the entropy
measure, Baysinger and Hoskisson (1989) found a correlation of 0.58 between Rumeh's
categories and a categorical measure created from the continuous related and unrelated
components of the entropy measure. Hoskisson, Hitt, Johnson, and Moesel (1993) found
strong support for the joint criterion-related validity of the Rumelt and enfropy
diversification measures with respect to accounting performance. Their study also
provided strong support for the reliability and validity of the entropy measure in relation
to the Rumeh approach, as a proxy for diversification strategy, findmg a correlation of
62
0.82 between tiie subjective and objective measure (latent constmcts). Finally, tiieir study
found negligible path coefficient changes between diversification and accounting
performance, when eitiier the subjective or entropy measures were substituted.
Classification Metiiodology for Level of Diversification
Because of tiie empirically demonstrated similarity between Rumeh's typology
and tiie enfropy measure, I used the two measures in combination (Figure 3.3), utilizing
the strengths of each measure. Business Units (BUs) were first classified as being single
stand-alone or nested within a dominant corporation as defined by RumeU (1974).
Although Rumeh's typology suggests that firms with sales of 95%) or greater be classified
as single firms, this study is limited to using the 90% demarcation because of the
reporting requirements for segment data. Of the 2,341 corporations in the sample, 1,824
were classified as single firms and included corporations such as Nike Inc., Oakley Inc.,
Wrigley Jr. Co., Gateway 2000 Inc., and Rubbermaid Inc. One hundred and sixty-six
firms where classified as dominate corporations and included corporations such as
Colgate-Palmolive Co., Harley-Davidson Inc., Baldwin Piano & Organ Co., and Seagram
Co. Ltd.
For the remaining corporations, a total diversification score (TDS) was calculated
for each diversified multibusiness corporation, using only primary SIC codes because
secondary SIC codes may not effectively represent strategic intent of corporate HQ
(Davis & Duhaime, 1992). A histogram reflectmg tiie frequency of scores is presented in
63
Figure 3.4. The total diversification scores were then standardized and screened for
outliers, with any score greater than +/- 5 standard deviations being elimmated. One
corporation (General Electric) with a TDS of 9.2 was deleted. As a resuU, 351 unique
corporations were identified witii TDSs ranging from 1.5 to 7.94, with an overall mean of
2.88 and a standard deviation of 1.014. Hierarchical cluster analysis of the TDSs, using
squared Euclidean distance within groups (SPSS software, release 6.1), was then
performed to group diversified muhibusiness corporations (MBC) as to their level of
diversification (see Table 3.1). The clustering of diversified corporations followed, fairly
closely, the peaks and valleys of the histogram of the total diversification score. The
analysis resulted in four clusters. A conclusive summary of the classification
methodology using Rumelt's typology, the entropy measure (Total Diversification Score),
and cluster analysis is presented in Figure 3.5.
Cluster one included TDSs ranging from 1.5 to 2.25 and was termed Least
Diversified MBC. This cluster contained 114 unique corporations with an average
number of 2.45 BUs and included corporations such as Abbott Laboratories, Sherwin-
Williams Co., Baxter Intemational Inc., Bell & Howell Company, and Burlington
Industries Inc.
Cluster two, termed Low/Moderate Diversified MBC, with total diversification
scores ranging from 2.26 to 3.15, contained 133 unique corporations with an average of
3.23 BUs per corporation. This group mcludes such corporations as Avery Dermison,
64
Johnson & Johnson, Wamer-Lambert Co., Bristol Myers Squibb, and Toshiba
Corporation.
Cluster tiiree, termed Moderate/High Diversified MBC and witii TDSs ranging
from 3.19 to 3.87, contained 57 unique corporations with an average of 4.27 BUs per
corporation. Some of the corporations included in this cluster are Monsanto Co., Procter
& Gamble Co., Tyco Intemational Ltd., and Weyerhaeuser Co.
The final cluster was termed Highly Diversified MBC and contained corporations
with TDSs greater than 3.9. Included in this group were 47 unique corporations with an
average of 5.71 BUs per corporation. Examples of corporations included in this category
are Siemens, Du Pont De Nemours, Gillette Co., and Hanson PLC. A comprehensive
recap of the summary statistics of corporations and corporate profiles as classified by
level of diversification are represented in Table 3.2 and Table 3.3, respectively.
Performance Metric
Assessing organizational effectiveness is complex, and contemporary approaches
consider multiple criteria simuhaneously, such as the Stakeholder (Connolly, Cordon, &
Deutsch, 1980) or Competing Values frameworks (Qumn & Rohrbaugh, 1983). Owing to
its roots as a more applied area, strategy has traditionally focused on business concepts
tiiat affect performance (Hoskisson, Hitt, Wan, & Yiu, 1999) and has led various autiiors
to suggest tiiat strategy's raison d'etre is tiie ongoing search for and sustamability of
economic rents (Amit & Schoemaker, 1993; Bowman, 1974; Penrose, 1959).
65
Common measures of performance are market share, revenue growth, and tiie like
(McGahan, 1999), with the two most commonly used in the corporate Hterature being
accounting measures and measures of financial market premiums. One measure for
operationalizing financial market premiums is Tobin's q, which reflects a firm's
prospects for profitability. However, tiiis measure is available only at the corporate level
and not the business uiut level. Additionally, Hoskisson et al. did not find a statistically
significant path between tiie entropy measure and Tobm's q. Furthermore, this measure
can fluctuate with shifts in investor expectations that are not fundamentally related to the
operations of the business (McGahan, 1999).
Market premiums are based on an efficient market hypothesis (Rumelt, Teece, &
Schendel, 1994). In the 1960s, the stock market responded favorably to conglomerate
acquisitions that led many researchers to conclude that these firms created value.
However, Shleifer and Vishny (1994) argue that the stock market was merely reflecting
the mistaken beliefs of a majority of investors. Drawing from their research on arbitrage
and market fads, the authors suggest that fads persist because it is too costly for the best
informed investors to bet against them. In sum, what the authors suggest is that using
stock market residuals, which is a standard way of investigating value creation, is not
really measuring value, but only what investors think value is.
Holzman, Copeland, and Hayya (1975) argued that the use of market measures
was problematic because managers relied more heavily on accoimting-based performance
in formulating diversification sfrategy. Accounting measures have also been found to be
66
a good predictor of fiittire expected performance (Keats & Hitt, 1988), an argument
supported by Jacobson (1987).
The accounting measure used in this study is commonly referred to as ROA
(Retum on Assets = Operating Profits/Total Assets); however, because of data
limitations, I used identifiable assets rather tiian total assets. ROA is very similar to
anotiier measure termed OROI (Keown, Martin, Petty, & Scott, 2000).
OROI is a combination of two ratios, as shown in Figure 3.4, consisting of an
operating profit margin ratio (operating profits/sales) and a total asset tumover ratio
(sales/total assets). The significance of the first ratio is tiiat it captures the five main
driving forces from the income statement while tiie total asset tumover ratio is a fimction
of how efficiently management is using the firm's assets to generate sales.
Therefore, in usmg ROA as a measure of performance, I suggest this ratio is
capturing a broad spectmm of operatmg qualities of the business unit and reflects much
more than a firm's historical advantage arising from management's ability to obtain assets
at less than full value in use (McGahan, 1998).
Although Bromiley (1986) and Jacobson (1987) strongly support accounting
measures, other scholars have argued that accounting conventions may generate specific
effects and that accounting rates of retum are distorted by a failure to consider differences
in systematic risk, temporary disequilibrium effects, tax laws, and accounting conventions
regarding research and development (R & D) and advertising (Wemerfelt & Montgomery,
1988). An important caveat to their argument is that the authors suggest these properties
67
are likely to vary more across industries (my emphasis) tiian across firms, hi
consideration of tiiese objections to the use of accounting measures, this study focuses on
just tiie manufacturing sector, thereby reducmg the possible bias that may arise from a
cross-industry study.
In sum, tiiis study uses an accounting based measure, ROA, because it does
capture a broad array of operatmg qualities of a busmess. Furthermore, inherent
differences in accounting practices are kept to a minimum by focusing on just the
manufacturing sector. Fmally, accounting measures are one of the few measures that are
available at the business unit level of analysis, which is the focus of this study.
Controls
Authors, such as Dess, freland, and Hitt (1990) and Hoskisson, Hitt, Johnson, and
Moesel (1993), have urged scholars to control for industry effects when uivestigating firm
performance. Palich, Miller, and Cardinal (2000), in their meta-analysis of diversification
studies that spanned over three decades, noted that a majority of studies failed to control
for a number of variables that have demonstrated significant effects on firm performance
independent of diversification. For example, few of the studies accounted for the unpact
of industry effects. Additionally, the research grounded in corporate effects has
demonstrated that not only are industry effects unportant determinants of firm
performance, but that corporate and business effects are equally if not more important
determmants. Therefore, drawing from the corporate effects literature, I not only
68
controlled for industry effects, but also corporate and business effects witiun tiie analysis.
Central to the modeling of these effects is tiie use of a general linear mixed model and is
discussed in the next subsection.
The importance of modeling entire classes of effects was demonstrated by Scott
and Pascoe (1986) by showing tiiat a class, representing muhiple factors, accounted for
tiie majority of tiie variance in profitability in their model over that explained by the
operationalization of specific constmcts. hi tiiis study, I suggest tiiat tiiese classes of
effects may be efficient proxies for more specific constmcts; i.e., firms effects for R & D
and capital mtensity and corporate effects for plaiming and control, organizational
stmcture, and scope (diversification) of the firm. Support for this argument may be
drawn from a previous variance decomposition study (Stetz & Phillips, 2000).
The authors demonstrated, through the use of a full and reduced model, that when
diversification was operationalized as a fixed effect m the full model, the parameter
estimates of the variance components of industry and corporate effects (random effects)
were smaller in comparison to the estimates in the reduced model. Additionally, the
parameter estimates for the business unit and the residual remained virtually unchanged.
Two conclusions may be ascertained: (1) Variance attributable to diversification was
being captured by the industry and corporate effects (reduced model), and (2) business
effects and the residual were not capturing any of the variance attributable to
diversification. Thus, these results lend support to the argument that industry, corporate.
69
and firm random effects may be adequate proxies for factors that are determinant of
business unit performance.
Another confrol that may be considered in the analysis of BU performance,
especially when the data from which tiie sample is drawn covers muhiple years, is tiie use
of tune. Time, usually operationalized as a year effect, was not included in tiie model
because multiple studies have found small or ttivial effects associated with this factor
(Schmalensee, 1985; Wemfeh & Montgomery, 1988; Rumelt, 1991; Roquebert, Phillips,
& Westfall, 1996; McGahan & Porter, 1997a, 1997b).
Model
From the review of the literature on corporate/industry effects, most studies have
used some means of variance decomposition such as OLS, ANOVA, or Variance
Components that are variations of a fixed or a random effects model. However, in this
study, I utilized a mixed modeling technique (Searle, Casella, & McCuUoch, 1991) which
allows the flexibility in modeling not only random factors (i.e., industry, corporate, and
business segments) but also fixed factors (spectrum of diversified firms, single, related,
and unrelated), as well as estimating parameters, means, and standard errors
simultaneously.
In a mixed model, there are two types of effects, random and fixed. An effect is
fixed if the levels in the study represent all possible levels of the factor, or at least all
levels about which inference is to be made. In this study, levels of diversification is
70
considered a fixed effect. Factor effects are random if the levels of the factor that are
used in the study represent only a random sample of a larger set of potential levels. The
factor effects correspondmg to the larger set of levels constitute a population of effects
with a probability distribution (Littell, Milliken, Stroup, & Wolfinger, 1996). Therefore,
modeling industry, corporate, and business effects as random effects make good sense m
that the data set contamed approximately 6,000 manufacturing firms (publicly traded) out
of a total population of 220,000 firms (includmg many small and private firms). To my
knowledge, this is the first study that has integrated industry, corporate, and business
effects into a model to test for differences in profitability of business units embedded
within corporations with varying levels of diversification. By so doing, this study has
attempted to integrate these two streams of research.
To test our hypotheses of differences in profitability among business units across
the spectrum of diversified corporations while simultaneously accounting for population
level factors (industry, corporate, and business effects), I used a general linear mixed
model (Searle, Casella, & McCuUoch, 1991; Littell et al., 1996) tiiat may be expressed in
the general form: Yijkim = |J.i + Ii + Cj + Bk + Sijkim
where: Dependent variable
Y = ROA (operating profit/identifiable assets) of uidividual business unit.
Fixed and Random Effects are the following:
m = Diversification'' as fixed effects, all else random;
I, = hidustiy effects, N(0,a;);
71
Cj = Corporate effects, N(0, CT^ );
Bk = Business effects, N(0, a]);
eijkim = Error, N(0, a ]); with tiie
assumptions of tiie model: All Ii, Cj, Bk, and Eijkim are independent of each other.
Subscripts are the following:
1 = Level of diversification, 1 to 6 possible states of a BU according to classification methodology;
i = Level of industry (590 levels);
j = Level of corporations (2,342 levels);
k = Level of unique business units (3,849 levels); and
m = Year (1991 tiurough 1997).
Other advantages to using tiie mixed model (SAS/STAT® software; SAS, 2000)
include the ability to specify the use of maximum likelihood estimation procedures,
which is the preferred method in unbalanced panel designs since the estimators are
consistent and asymptotically normal (Searle et al., 1991). Furthermore, through the use
of generalized least square estimates (GLS), the model takes into account that
observations may be correlated over time, and thus, it is more efficient with higher
reliability in the estimation of means and standard errors.
Additionally, this technique enables a researcher to integrate research across two
or more levels of sfrategy (Dess, Gupta, Hennart, & Hill, 1995). For example, in our
model, random effects represent three levels of sfrategy — mdustry, corporate, and
72
business level. By including multiple levels in tiie model, a researcher is also integrating
multiple tiieoretical frameworks, such as industrial organization economics (mdustry
effects) and sfrategic management (corporate effects) (Hitt, Hoskisson, & Kim, 1997).
Finally, mixed models can represent very complex, multilevel phenomenon
parsimoniously, witii only a few variance components, rather than hundreds or tiiousands
of regression coefficients (Littel et al., 1996). For example, I would have had to use six
tiiousand seven hundred and sixty-eight (6,768) dummy variables.
Summary
The research design for this study features not only some unique aspects but also
answers multiple calls of researchers within the strategic management literature. The
dependent variable, ROA, is operationalized at the business unit level of analysis which
allows for the assessment of the effects on business unit performance of (1) corporate
strategy and (2) rivalry among and between business units. This focus answers multiple
calls from within the sfrategic management literature.
In the investigation of retums to business units and by the inclusion of single
businesses as well as business units embedded in multibusiness corporations, the study is
also able to investigate the degree to which corporations make business units, to use
Porter's (1987) words, "better off." Focusing on the business unit of analysis and
uivestigating if corporations make businesses better off answers the calls for future
research by Bamey (1997), Porter (1987), and Bowman and Helfat (2001), respectively.
73
As noted in tiie review of empirical studies on diversification, criticisms have
been voiced conceming tiie lack of controlling for variables tiiat have an hnpact on
business unit performance. Drawing from tiie corporate effects research, tiiis study
controls for industry, firm, and corporate effects. The latter effect has been shown to
have as important an impact on performance and thus, it is unportant for researchers to
also account for this effect in future studies.
Through the use of a general linear mixed model, this study is able to model the
effects of industry, corporate, and business effects while investigatmg the linkages
between diversification and performance across the spectmm of diversified corporations.
To my knowledge, this is the first study that has mtegrated these two streams of research
to investigate the linkages between diversification and performance, especially at the
business unit level of analysis.
In sum, the results and conclusions drawn from this study rest on the sttength of
three elements: (1) the data, (2) the methodology, and (3) the measures. Value as to the
use of the COMPUSTAT® segment database for investigating diversification has been
substantiated by various in-depth reviews, such as that by Davis and Duhaime (1992).
The methodology employed (GLMM) is widely used in genetic research and is
more efficient, with higher reliability, in the estimation of means and standard errors.
Additionally, mixed models can represent very complex, multilevel phenomenon
parsimoruously, with only a few variance components, rather than hundreds or thousands
74
of regression coefficients. Fuially, the technique encompasses the use of maximum
likelihood estimates, which is the preferred method for unbalanced panel designs.
Finally, the entropy measure of diversification has been substantiated for its
reliability and validity not only as an objective measure of levels of diversification but
also for its criterion-related validity with accounting measures, such as ROA.
75
Levels and Types of Diversification
Low levels of diversification
Single business
Dominant business
Over 95% of revenues come from a single business
Between 70% and 95%) of revenues come from a single business
Moderate to high levels of diversification
Related-constrained
Mixed related and unrelated (related-linked)
Less than 10% of revenues come from the dominant busmess, and all businesses share product, technological, and distribution linkages
Less than 10% of revenues come from the dominant business, and there are only limited links between businesses
Very high levels of diversification
Unrelated diversification Less than 70% of revenues come from the dominant business, and there are no common Imks between businesses
Figure 4.1. Rumeh's Typology. Adapted from Rumeh, 1974.
76
77
Histogram of Total Diversification Score^ (on avg.) of Multibusmess Corporations
80
60
40
ti
3^^_ std Da/ = 1.07 tVban = 290 N=35200
1.50 250 3.50 4.50 5.50 6.50 7.50 a50 200 3.00 4.00 5.00 600 7.00 BOO 9.00
Figure 4.3. Total Diversification Scores: Histogram. ^ Actual calculation of total diversification scores reported within article.
78
Table 4.1. Total Diversification Scores: Cluster Analysis.
Cluster Analysis of Total Diversification Scores
Clusters
Cluster One
Cluster Two
Cluster Three
Cluster Four
TDS^ range
1.50 to 2.25
2.26 to 3.15
3.19 to 3.87
> than 3.90
Level of diversification
Least Diversified Multibusiness corporations
Low/Moderate Diversified Multibusiness corporations
Moderate to High Diversified Multibusiness corporations
Highly Diversified Multibusiness corporations
Named
LDMBC
L/MDMBC
M/HDMBC
HDMBC
The above clusters may be presented as to their level of diversification
Range of Total Diversification Scores
1.5 -•7.9
Level of Diversification
Least Diversified
MBC TDS = 1.5 to 2.25
Low/Moderate Diversified
MBC TDS = 2.26 to 3.15
Moderate/High Diversified
MBC TDS = 3.19 to 3.87
Highly Diversified
MBC TDS = > 3.9
* Multibusiness corporations (MBC) other than single and dominant firms. ^ Hierarchical Cluster analysis, using squared Euclidean distances within groups. '^ TDS: Total Diversification Score, continuous measure of diversification.
79
tn
u
a
B ii ^ - »
C o
- :s (2 c ^
•s|l.i
^ I S i I -.C „ S M E c 05 g -S « 5 o
o <c ffl w .^ Z
U I
80
Table 4.2. Descriptive Statistics of Sample by Level of Diversification.
Statistics for Corporations by Level of Diversification
Level of Diversification
Single^ Corporations
Dominant^ Corporations
Least Diversified Multibusiness Corporations''
Low/Moderate Diversified Multibusiness Corporations''
Moderate/High Diversified Multibusiness Corporations'"
Highly Diversified Multibusiness Corporations"'
Total Sample
OBS
9,888
2,476
1,677
2,673
1,475
1,536
19,725
Number of Unique
Corps.
1,824
166
114
133
57
44
2,341
Avg. # of BUs per
Corp.
1
2.71
2.45
3.23
4.27
5.71
3.84
Unique SICs
367
233
170
253
174
190
589
ROA
7.32
11.06
13.21
13.32
11.81
10.25
9.67 %
Std. Error
0.195
0.458
0.622
0.385
0.478
0.303
^ Corporations are classified as Single or Dominant according to Rumeh's Diversification Typology.
"" Groups of corporations classified through cluster analysis as to their level of diversification as measured by the Total Diversification Scale.
81
Table 4.3. Examples of Corporations as Classified by Level of Diversification.
Level of Diversification Names of Corporations^
Single business WD-40 CO
OAKLEY INC
WRIGLEY (WM) JR CO
NIKE INC -CLB
GATEWAY 2000 INC
RUBBERMAID INC
Dominant-bus uiess SCHERING-PLOUGH
COLGATE-PALMOLIVE CO
HARLEY-DAVIDSON INC
BALDWIN PL^LNO & ORGAN CO
SEAGRAM CO LTD
Least Diversified MBC
TDS =1.5 to 2.25
ABBOTT LABORATORIES
SHERWIN-WILLIAMS CO
BAXTER INTERNATIONAL INC
BELL & HOWELL COMPANY
BURLINGTON INDS INC
Low/Moderate Diversified MBC
TDS = 2.26 to 3.15
Moderate/High Diversified MBC
TDS = 3.19 to 3.87
Highly Diversified MBC
TDS = > 3.9
AVERY DENNISON CORP
JOHNSON & JOHNSON
WARNER-LAMBERT CO
BRISTOL MYERS SQUIBB
TOSHIBA CORP
MONSANTO CO
PROCTER & GAMBLE CO
TYCO INTERNATIONAL LTD
WEYERHAEUSER CO
SIEMENS AG -ADR
DU PONT (E I) DE NEMOURS
GILLETTE CO
HANSON PLC -ADR
' Corporate names as listed within COMPUSTAT® data set MBC: Multibusiness Corporation (other than dominant or single firms).
82
Decomposition of the ROA and OROI Ratio
j^Q^ = /^Operating Profits^ ^ f Operating Profits V Total Assets y V Identifiable Assets
OROI may also be defined as operating profits/total assets and can be decomposed into two other ratios:
/ n K . « r o + ; « r r \
OROI = Operating
profit margin^
I Total asset^ V tumover J
•c 11 r^^^T Operating profits ,^ Sales or more specifically, OROI gales ^ Total assets
Therefore, this ratio captures two important dunensions pertaining to firm operations, which are the following:
1)
Operating profits _ ratio captures five factors or "driving forces" from the income Sales statement:
a) The number of imits of product sold. b) The average selling price for each product unit. c) The cost of manufacturing or acquiring the firm's product. d) The ability to control general and administrative expenses. e) The ability to control the expenses in marketing and distributing
the firm's product.
2) Sales ratio is a function of how eflficiently management is using the
Total assets ~ firm's assets to generate sales and is a major determinant ui the retum on investment.
Figure 4.5. Performance Metric: Accounting Based. Adapted from Keown, Martm, Petty, and Scott, 2000.
83
CHAPTER V
ANALYSIS
Results
This diversification study is a cross sectional analysis of firms witiun tiie
manufacturing sector (SIC 2000-3999) and consists of a large sample, 19,725
observations, that spans seven years, and includes 2,341 corporations with 3,838 unique
business units operatmg in 589 unique industries. Level of diversification of all
corporations was determined through the use of Rumeh's typology and an entropy
measure. Rumeh's typology was first used to identify suigle and dominant businesses.
For the remaining muhibusiness corporations (MBC), an entropy measure (total
diversification score) and cluster analysis of the scores was utilized to categorize MBC
into groups based on their level of diversification. Four groups of MBC resulted and
were coded as to their level of diversification — least diversified MBC (LDMBC),
low/moderate (L/MDMBC), moderate/high (M/HDMBC) and highly (HDMBC). hi sum,
six unique groups of corporations were identified that varied as to then level of
diversification — smgle through highly diversified multibusiness corporations. Each
business unit within each corporation was coded in accordance to the overall level of
diversification of the corporation, respectively.
To begin to uncover the pattems in the data, least squares means (LSMEAN) and
standard errors of business unit ROAs (fixed effects) were estimated usmg a general
84
Imear mixed model (GLMM) while controlling for population level factors (random
effects), i.e., industry, corporate, and business effects (see Table 5.1). A GLMM model
was used in this analysis because of its flexibility in using both the general least squares
and maxunum likelihood estimates and the incorporation of random effects into tiie
estimation procedure. This flexibility is advantageous because the means and standard
error estimates are corrected for autocorrelation and repeated measures (dependence of
observations) as well as takmg into account the unbalanced panel design of the data.
Because of these corrections, the LSMEAN estimates of ROAs are slightly different from
a purely mathematical derivation of the mean and much different in the estimation of
standard errors (review Table 3.2). The estimates and standard errors of business units
across the spectrum of diversified corporations, as categorized by level of diversification,
are as follows: single firms, p. = 7.8, std. error = 0.0519; dominant firms, p. = 10.76, std.
error = 0.932; LDMBC, ^ = 12.24, std. error =1.121; L/MDMBC, ^ = 12.51, std. error =
0.912; M/HDMBC, ^ = 11.14, std. error = 1.273; and HDMBC, |ii = 9.91, witii std. error
= 1.256, respectively.
The plot of the estimated means of all groups suggests a non-monotonic
relationship (inverted U-shape) among business imit performance, on average, and the
level of diversification (see Figure 5.1). To my knowledge, only two other empirical
studies have suggested a similar shape and form (Grant, Jamie, & Thomas, 1988; Palich,
Cardmal, & Miller, 2000); however, Grant et al.'s study used a concentric measure of
diversification based on SIC codes while Palich et al.'s study was a meta-analysis of
85
diversification stiidies, many of which had not controlled for business effects or industry
effects, let alone for corporate effects.
The plot fiirther suggests that business units embedded in low to moderate
diversified corporations eam, on average, a much higher retum tiian tiiat of single stand
alone busmess, 60%) greater as measured by ROA. In comparison to corporations at the
other end of the diversification spectrum, i.e., highly diversified corporations, low to
moderate diversified corporations eamed a 26% greater retum, on average.
Within Figure 5.1 is also a plot of the number of business units per corporation,
on average, accordmg to the level of diversification of the corporation. In this study, the
number of business units initially increases for dominate corporations to 2.71, then
decreases to 2.45 for least diversified corporations and thereafter, steadily increases as the
level of diversification increases, with highly diversified corporations, on average, having
5.71 business units per corporation. The increase in the number of business units is, in
actuality, a reflection of an increase in activities of corporations into additional product
markets and tracks in parallel to the level of diversification. In other words, as firms
expand their operations into additional product markets, the level of diversification
increases accordingly.
Hypothesis 1
Although the plot of business unit ROA versus tiie level of diversification
suggests a curvilinear-shaped relationship and is uiformative, more robust conclusions
86
can be made by determining if the various mean ROAs are significantly different from
one anotiier across the spectrum of diversified corporations. The first hypotiiesis
mvestigated tiie question if diversification, witiim the manufactiirmg sector, unproves
busmess unit performance by comparing the ROAs of non-diversified corporations
(smgle stand-alone firms) to that of business units that are embedded witiiin diversified
corporations (while confroUmg for mdustry, corporate, and business effects). Since, a
priori, eitiier of tiie two populations may have a higher mean, I performed a two-tailed
test at tiie 0.05%) confidence level. The difference in ROA means of tiie two broad
groups was 3.659 witii a standard error of 0.625 (see Table 5.2) and was statistically
different witii a t-value of 5.62. The p-value was 0.0001 and may be interpreted as tiie
probability of getting a test statistic as extreme as the observed test statistic, given that the
null hypotiiesis is tme (Berger & Sellke, 1987). Based on tiiese resuhs, I reject tiie null
hypotiiesis (Hoi) that there is no difference between business unit profitability of
diversified and non-diversified corporations withui the manufacturing sector.
Hypothesis 2
By delineating the broad category of diversified corporations (one of the two
categories investigated in the first hypothesis) into five categories based on their level of
diversification — dominate through highly diversified corporations (see classification
scheme Figure 3.4) the second hypothesis investigates if performance differences, as
measured by ROA of business units that are embedded within diversified corporations,
87
accrue to firms with respect to tiiefr level of diversification (while controlling for
hidustry, corporate, and business effects). In other words, do busmess units embedded
with corporations at a specific level of diversification eam a higher retum than any of the
other BUs at other levels of diversification. Again, smce, a priori, any of the populations
may have a higher mean than any of the others, I performed a two-tailed test at the 0.05%
confidence level. The difference in mean ROAs of busmess units, across all levels of
diversification, was non-significant at the 0.05 significance level with all t-values below
1.96 (see Table 5.3). Based on these results, I fail to reject the null hypothesis (Ho2) that
no differences in profitability (ROA) exist among business units of diversified
corporations (within the manufacturing sector). However, the difference in mean ROA
between low to moderate diversified corporations and highly diversified corporations is
2.59 with a p-value of 0.089 and approaches tiie alpha level that would resuh m a
rejection of the null hypotheses.
Hypothesis 3
The last hypothesis mvestigates if performance differences exist between BUs of
diversified corporations and smgle stand-alone firms. This inquiry specifically addresses
the question if a corporate diversification strategy creates value by makmg business units,
that are embedded withm muhibusmess corporations, "better off." "Better off' is defmed
as the capacity of BUs within muhibusmess corporations to eam a higher ROA tiian
single-stand alone firms, over time. Smce, a priori, any of tiie populations may have a
88
higher mean than any of the others, I performed a two-tailed test. The results of tiie
LSMEAN pair-wise tests between smgle stand-alone firms and BUs within the five levels
of diversification suggests that, at tiie 0.05%o confidence level, significant differences
exist between the mean ROA of all BUs except for BUs that are embedded within highly
diversified corporations (see Table 5.4). The difference m means between single and
HDMBC was 2.11 with a standard error of 1.334 and was not statistically significant with
a t-value of 1.58. The p-value for this test statistic was O.I 138 and thus, I fail to reject the
null hypotheses that no performance differences exist between BU profitability of
HDMBC and nondiversified firms (within the manufacturing sector). However, for all
other levels of diversification — dominate through moderate/high diversified
corporations, I reject the null hypotheses.
The most significant differences in means were between the least and low to
moderate diversified corporations and single stand-alone firms (p = 0.0002 and p =
0.0001, respectively), with dominant (p = 0.0034) and then moderate to high diversified
corporations declming in significance to p = 0.0108. A summary plot of the test statistics
for differences m means (ROA) between single firms and BUs embedded m corporations
of varymg levels of diversification along (includmg the respective p-values) are shown in
Figure 5.2. As may be noted, this graphical representation also depicts an mverted U-
shaped relationship not only for the differences m means but also for tiie levels of
significance.
89
Additional Confirmation of Results
Several tests and additional analyses were performed to check the reliability of the
findmgs tiius presented. One area deemed necessary to mvestigate was tiie degree to
which assets may have a direct effect on performance of business units. Specifically, it
may be argued, firms tiiat use historical costs rather than replacement value of assets will
show a higher ROA, not because of efficiencies or economies, but merely because tiie
retum is an artifact of tiie accounting method. Therefore, to specifically mvestigate tiie
degree to which identifiable assets (lA) are a determinant of firm performance, L\ were
operationalized as an additional fixed effect in the model (with diversification categories
retamed m the model as fixed effects) while controlling for industry, corporate, and
busmess effects (random effects), h may be noted tiiat tiie modelmg of assets as a fixed
effect is, m effect, the partitionmg out of a specific factor from the multiple factors that
comprise random business effects (similar to diversification and corporate effects as
discussed ui Chapter IV). The results of this test suggest, that the relationship between
identifiable assets and ROA is highly insignificant (see Table 5.5). The parameter
estimate for lA was 0.00000972 (millions of dollars) with a correspondmg p-value of
0.8529. This test suggests, albeit indirectly, that assets, however valued, are not
significantly influencing the results of the analysis.
A second concem of this study was the degree to which the classification system
of multibusiness corporations (categorization of multibusiness corporations into specific
groups corresponding to a given level of diversification) was accurately reflecting the
90
relationship between ROA and the level of diversification of all uidividual multibusiness
corporations. This concem lies specifically with the aggregation of corporations through
cluster analysis of the total diversification scores (TDSs) and tiie eventiially
categorization of MBC mto four unique groups denoted by tiieir level of diversification.
To address tins concem, I operationalized tiie TDSs as a contmuous variable in a
regression equation (fixed effects) ratiier tiian operationalizmg MBC as discrete
categories based on tiieir level of diversification, hi tiiis ahemative model, mdustry,
corporate, and busmess effects were operationalized as random effects to be consistent
with previous models used in this analysis.
A contmuous variable may take on essentially any real value in some interval,
with tiie bounds of tiie mterval defined by the endpoints of the data. In this study, the
bounds of the total diversification score are 1.5 and 7.94. The estimates of the parameters
for Po and pi were 13.918 and -0.0716 respectively, and the plot of the total
diversification scores, ranging m value from 1.5 to 7.5, are shown in Figure 5.3. As may
be noted, the plot of the TDSs as a continuous variable is linear with a continuously
decreasing slope and tracks very closely to the plot of the corporations categorized into
groups as to thefr respective level of diversification. (Polynomials of the TDSs were also
tested, such as a quadratic, but the parameter estimates were nonsignificant). In sum, the
similarity of the plots of the entropy measure (TDSs), one as a continuous variable and
the other as categories of corporations with varying levels of diversification, suggest that
91
tiie clustermg technique used m this stiidy is accurately reflectmg the level of
diversification of multibusmess corporations.
A final concem of this study was the commensurability of the resuhs to other
studies origmatmg m tiie corporate effects literature concemmg the parameter estimates
of the random effects. The estimates for all three random effects — mdustry, business,
and corporate — were found to be significant (see Tables 5.1, 5.2, 5.3, and 5.4).
Furthermore, the magnitude of the parameter estimate for corporate effects is similar to
other studies (see Table 2.1) and adds additional empirical evidence to the proposition
that corporate strategy matters (in contrast to tiie revisionists' view) (Bowman & Helfat,
2001, p. 1).
Estimates for industry effects were somewhat less than in other studies and may
be a result of the coding scheme to identify business units and industries ui this study.
Rather than use SIC codes as a proxy for unique busmess units, I used the actual segment
name assigned by the corporation to identify the respective business units of corporations
in an attempt to derive more specificity to the origins of industry and corporate effects.
For example, it is possible that two business segments may have the same SIC code even
though they have different names and are serving different markets. If prior studies use
only the SIC code to identify business units, the possibility exists that the analysis would
assume that there is only one busmess unit rather than two because of one SIC code
demarcating the business segment. By the coding scheme used in this study, I believe this
underaccounting was eliminated, in part, by the use of segment names and helped to
92
reduce tiie amount of variance that would have been incorrectly attributed to industry
effects in previous studies.
As a final check, tiie mean ROA of all busmess unhs from tiie entire
manufacturmg sector was compared to otiier studies tiiat have used the COMPUSTAT®
segment-file database, hi a study of performance of busmess units from 1981 through
1994, McGahan (1998) reported an overall ROA for tiie manufacturing sector, defmed as
SIC code 30, of 8.59%). In this study, the manufacturing sector was defined ui accordance
with the SIC codes 20 through 39, and found an overall ROA for the manufacturing
sector of 9.67%), arguably, a very similar finding for the entire sector.
Discussion
The persistence in performance of corporations, whether high, low, or moderate,
is a phenomena that has been commented on by many authors, including Rumelt,
Schendel, and Teece (1994) and is documented m a recent study by McGahan (1999) in
which 77.6%) of the firms, within the entire US economy, sustamed high performance for
over fourteen years, while 81.4%) of the moderate and 78.4%) of the low performers
remained relatively consistent in their performance levels. This evidence of persistence is
remarkable m the light of other studies that have shown how quickly above-average
performance can collapse toward tiie averages within approxunately six years (Ghenawat,
1991). The above demonstrated persistence m performance, over substantial periods of
time, serves to underpin the credibility of the findmgs of this study conceming the
93
performance of manufactiiring firms across tiie spectrum of diversified corporations as
delmeated by tiieir level of diversification — from smgle stand-alone through highly
diversified corporations. Additional support for the credibility of the findings of tins
analysis lies m tiie similarity in tiie parameter estunates for tiie random effects as
compared to tiiose derived m tiie effects literatiire. hi sum, I argue the resuhs obtamed in
tills cross-sectional analysis, tiiat covers a time span of seven years, is representative of
ongomg performance similarities and differentials that accme to business units of
diversified corporations.
The question if diversification is a means through which a corporation can create
value was addressed by Chandler (1962, 1977) and Williamson (1975) who emphasized
the advantage of the muhidivisional form over the fimctional organization of muhi
business operations and argued tiie M-form is a means to ensure the efficient employment
of resources to overcome the increasmg complexity and uncertainty within the firm.
However, the widely held assumptions that bigger is better, that all the advantages are on
the side of bringing more and more activities and resources under the control of a single
hierarchy, are being supplemented with the notion that important strengths may be
associated with alliances or loose confederations of smaller and more flexible forms
(Scott, 1995). From an equifinality perspective, it has been suggested that being
organized as a single stand-alone firm may be as optimal as a firm that is diversified. The
sustained performance of Microsoft (prior to the anti-tmst litigation) and Nike would
appear to support this viewpoint.
94
McGahan (1998) found negligible support for the benefits of diversification
(measured by corporate focus) and concluded that the low significance is consistent with
the hypothesis of a regime change from a period in which moderate diversification was
typically optimal to a period in which diversification was no longer as necessary to
achieve the benefits of relatedness. Development of new types of contracts of arm's
length relationships may have contributed to widespread divestment of loosely related
business. Furthermore, the proliferation of alliances suggests that companies use
arrangements other than full-scale diversification to achieve the benefits of coordination.
Researchers have acknowledged that in an uicreasmgly complex and turbulent
environment, firms can enhance then performances through strategic collaboration
(Confractor & Lorange, 1988). Dyer and Smgh (1998) argue tiiat tiie relationships
between firms are uicreasmgly unportant in explammg super normal profits with the
primary sources of high retums bemg relation-specific investments, interfirm knowledge-
sharing routmes, complementary resource endowments, and effective govemance. Other
authors (Barrmger & Harrison, 2000 (review); Hanssen-Bauer & Snow, 1996; Jarillo,
1988; Powell, Koput, &. Smitii-Doerr, 1996) suggest that cooperative strategies, such as
alliances and networks, allow small and medium-size firms to compete agamst large
companies by allowmg tiiem to leverage tiiefr resources tiu-ough idea and information
sharmg, and tiu-ough joint business activities (e.g., marketmg, production, and product
design), while remammg mdependent and unburdened by intemal costs.
95
Altiiough tiie above arguments are sound and very plausible, tiie results of tius
stiidy suggest tiiat, on average, busmess units embedded within diversified firms are able
to attam a higher retum, as measured by ROA, than single stand-alone businesses. At the
most optimal level of diversification, the retum on assets was 60%) greater. Given tiie
likelihood of tiie persistence in retums over time and tiie magnitude of the difference in
ROA (statistically significant), tiie difference in accumulated value would be substantial.
Furtiiermore, m contrast to the findmgs of negligible corporate effects in the McGahan
study, I found corporate effects (random effect) explained approximately 7% of the
variance m ROA and explamed more variance than industry effects. This findmg of a
non-trivial corporate effect is in parallel to other empirical studies in the effects literature
and further supplements the argument that corporate level strategy significantly
confributes to business unit performance. In sum, this study suggests that corporate level
factors are a significant determinant of business unit performance and second, on average,
diversified firms performed significantly better than single stand-alone firms, with one
exception, which will be discussed later.
The findings that diversified firms perform, on average, better than single stand
alone firms also fiimishes evidence, albeit in an mdirect way, that the motive for
diversification is based more on profit maximization rather than as a means through
which employment risk may be reduced or compensation maximized. Although, by
chance, either motive may have resuhed in high performance, the likelihood is minimal
that diversification, pursued from a purely behavioral perspective, would resuh ui
96
sustamed perfonnance. hi a similar vein, it would also be unlikely for firms, tiiat
diversify m an attempt to gam legitimacy from thefr instittitional environment, to achieve
and sustam above average retums as reflected m busmess unit ROA. Furthermore, as
Porter (2001) suggests, "it is more important than ever for companies to distinguish
tiiemselves tiirough strategy" (p. 63). One may conclude, tiierefore, tiiat if diversification
is not based on an integrated and coordinated set of commitments and actions designed to
gam a competitive advantage (strategy) not only will performance outcomes most likely
be pedestrian but also the firm may not survive.
Arguable, therefore, diversification is a means through which performance
advantages may accme to business units that are embedded within a corporate stmcture.
But the question remains as to what level of diversification is the most optimal for a
corporation to achieve high performance. As noted in the literature review, the various
theoretical perspectives suggest different mechanisms and accordmgly, different levels of
diversification, through which a firm may attain above average retums. Broadly
speaking, the various mechanisms may be summarized into three categories: (1)
operational synergies through low to moderate levels of diversification, (2) financial
economies through moderate to high levels of diversification, and (3) market power
economies through moderate to high levels of diversification. Each of these perspectives
will be discussed in tum, respectively.
Theories that suggest low to moderate levels of diversification as a means to
achieve higher performance are based on the notion of economies of scope underpinned
97
by some form of synergy. The mechanisms through which synergies may be achieved
depends on which theoretical lens one wishes to employ, with one perspective arguing
that the genesis of synergy is through the sharing of activities while the other argues that
it is the leveraging of capabilities and core competencies, while another argues it is the
efficiency and effectiveness of the firm. Furthermore, as each perspective has contrastmg
underlying assumptions conceming the mobility of resources and firm homogeneity and
heterogeneity, each evaluates the sustainability of performance achieved through
synergies with differing criteria.
The mdustrial stmcture perspective (Conner, 1994) suggests strategy is the
creation of a fit among a company's activities and competitive advantage depends on how
these activities fit and remforce one anotiier. If tiiere is no fit among activities, tiiere is no
distmctive strategy and little sustamability. There are three types of fits: (1) sunple
consistency, (2) activities tiiat are remforcmg, and (3) optimization of effort (Porter,
1996). Additionally, if tiie system of activities is based on second and third order fit, the
more sustamable tiie firm's competitive advantage.
Altematively, tiie resource-based view of tiie firm asks how rare and inunitable is
the synergy (economies of scope) tiiat a low to moderate level of diversification seeks to
create. AdditionaUy, for it to be a source of sustamed competitive advantage, synergy
must also create value and the finn needs an appropriate organizational stmctiire tiiat will
enable tiie hnplementation of tiie specific sfrategy mtended to captiire synergy among its
busmess units (Bamey, 1997). As suggested above, it has been argued tiiat tiurough tiie
98
use of strategic alliances, a firm may gain tiie economies of scope that could be obtained
if tiiey had carefully developed relations across businesses they owned (Bamey, 1997).
Given this scenario and m accordance to tiie VIRO framework, it may be deducted that
the synergy of multibusiness firms would be dissipated, over time, by firms usmg
interorgaruzational relationships as a substitute for gaining economies of scope.
With the findings that low to moderate levels of diversification was the optimum
level through which firms achieved high performance, one may conclude, indirectly, that
the underlying mechanisms through which these diversified corporations attained a 60%
higher retum on assets over that of single stand-alone corporations are based on some
form of operational synergies. As the 60%) higher retum is an average over seven years
and drawing from the notion of persistence in retums, one may further speculate that this
performance is sustained either through the sharing of activhies with second or thfrd order
fit or the imitability and rareness in the leveragmg of capabilities and/or competencies
across busmess units through which these synergies are achieved.
An additional insight that may be gleaned from the correspondence between levels
of diversification and performance (see Figure 5.1) is that tiie relationship peaks at a low
to moderate level of diversification and tiien subsides as the level of diversification
mcreases. This non-monotonic relationship suggests, tiierefore, tiiere are limits to tiie
degree to which operational synergies may be leveraged across muUiple busmess units, in
tiiat tiie retums to increased diversification begin to taper off after some optmial point.
The sources of this declme may rest on tiie added costs of coordmation witiiout parallel
99
mcreases in retiuns or tiie lack of alignment between appropriate control and plannmg
mechanisms and tiie firm's organizational stmctiire as tiie level of diversification
mcreases. Nevertheless, the respective theoretical frameworks are silent as to tiie limhs
to which synergies may be achieved as well as tiie causal mechanisms that may moderate
or dissipate the benefits of synergy. This is an area for ftitiire tiieoretical development.
Given the positive resuhs indicatmg the benefits of operational synergy, in
general, and the arguments for the benefits to the sharing of production activities,
specifically, it is interesting to note that St. John and Harrison (1999) suggest financial
benefits do not accme from shared resources ui manufacturing. Furthermore, Davis and
Thomas (1993) argue that production relatedness between dmgs and chemicals showed
no evidence of synergy. Thus, a question emerges as how does one reconcile the above
findings to the results of this study.
To regress back to the early 1980s, Porter's (1980) generic strategies were
originally conceived m light of the constraints of tradhional manufacturing technology.
Low-cost leadership strategies were achieved by low variety, standardized products and
long production runs and were necessary for buildmg large market shares. Because high
market share reinforces benefits of scale, managers delayed mvestments into developing
new products and processes in order to amortize currently high fixed costs surrounding
dedicated, mflexible production. Differentiation sfrategies were based on the notions tiiat
small batches, high-quality products and a premium image can come only from a smaller
100
(and often higher cost) production process that is more flexible, largely because of greater
labor intensity.
Today, flexible manufacttiring technologies are able to elunmate tiie tradeoffs
between cost versus variety and volume versus flexibility in ways tiiat render obsolete the
manufacturing constraints of an earlier time. The essence of flexible manufacturing
systems (FMS) is to erase the cost (productivity) versus variety (innovation) tradeoff
mherent in traditional manufacturing technology. When properly implemented, FMS
enables greater product breadth/variety, flexibility in response to shiftmg customer
demand, and quicker adaptation to new design configurations, while also retaining low-
cost advantages and consistent high quality.
Another technology advance that has allowed greater strategic flexibility and
responsiveness is new information networks that link suppliers and customers with
manufacturers. A specific example of how information technologies have created
opportunities for efficient networking between firms involves CAD/CAM tie-ins that
enable multiple designs from a broader range of extemal suppliers, rather than solely
relying on m-house efforts that often take much longer.
With the advent of flexible manufacturing (Goldhar & Jelinek, 1983; Nemetz &
Fry, 1988; Stalk, 1988) and CAD/CAM systems (Lei, Hitt, & Goldhar, 1994), contract
manufacturing has become a muhibillion dollar mdustry with an annual growth rate of
25%) to 30%) annually (Tanzer, 1999). The Melboume Manufacturing Division is a good
example of how manufacturing is changing. Approximately 40% of the output of the
101
division goes mto Dictaphone Corporation's (parent company) dictation-based systems,
communication recordmg systems, and loggers while the remaining 60% of tiie output is
attributed to confract manufacturing, hiterestingly, tiie busuiess challenge for tiie division
is to contmue to leverage its strengths to expand the contract manufacturing side of the
business (Maim, 1999).
Thus, underpinned by significant changes in technology and advancements in
software development, contract manufacturing may be a viable, ahemative means through
which a firm may outsource its production needs. Or, to put this in another way, contract
firms, utilizing cutting edge technology, may be able to manufacture the same products of
the focal firm more effectively and efficiently. The pouit being, if a production process is
pedestrian and can be easily imitated or duplicated by other firms, havmg economies of
scope through the sharing of production facilities is a necessary but not a sufficient
condition for sustained performance. Therefore, a possible reconciliation to the disparity
in findmgs (mentioned before) may lie m the degree to which the sharing of activhies are
of second and thfrd order fit or the leveraging of resources are inimitable and rare within
the manufacturing muhibusiness corporations of the respective studies. To validate or
disconfirm this notion m future research through a more fine-grained analysis would be
extremely uiformative not only to academics but also to practitioners.
Those who view synergy as tiie essence of corporate level sfrategy, acknowledge
tiiat companies often find it difficuh to gain synergy benefits (Porter, 1985), and in some
mstances, the potential for synergy may simply not exist in some muhibusuiess
102
corporations (Goold & Luchs, 1993). Otiiers have argued that tiie managing of complex
mterrelationships to create synergy across businesses is not the onfy means of creating
value. For some companies, the advantages of managing stand-alone business may
outweigh the long-term mvestment required to create Imkages among tiiose businesses.
For example, Goold and Campbell (1987) suggest tiiat companies, such as Hansen, which
places little emphasis on synergy as a source of corporate value added, performed at least
as well as companies that placed more emphasis on linkages across businesses.
Hansen PLC is a manufacturing company that was identified in this study as being
a highly diversified muhibusiness corporation (see Table 4.3). In contrast to the findmgs
of the above authors, the performance of business units embedded within highly
diversified corporate stmctures was not, on average, statistically different than the
performance of single stand-alone firms at the 0.05 level of significance. Conversely,
however, this study did not find a statistically significant difference between performance
of highly diversified corporations and low and low to moderately diversified companies,
suggesting some support for the equivalence of performance among corporations at
varying levels of diversification.
Nevertheless, the essence of corporate strategy, and diversification m particular, is
that the businesses in the corporate portfolio must be worth more under the management
of the company in question than they would be under any other form of ownership.
Therefore, the question arises as to why a firm diversifies to such levels and yet, eams
average retums. Jensen (1986) suggests and as Fligstem (1985) confirms, firms will
103
mvest m diversification projects whose net present value is less tiian zero when tiiefr
managers pursue maxmiization of tiieir own interests rather tiian shareholder-value
maximization. An ahemative explanation for firms being at high levels of diversification
rests witiim changes m tiie mstitutional environment, such as product or capital markets.
For example, Bhide (1989, cited in Markides, 1992) argues tiiat tiie rismg sophistication
of the capital markets resuhuig from deregulation (and increased power of mstitutional
investors) and increased competition has "eroded one of the major advantages of a
diversified firm, tiiat of actuig as an intemal capital market to its divisions" (p. 401).
Conversely, in parallel to changes in the extemal envfronment, increased globalization
has increased some costs intemally to the organization, such as costs due to information
and control loss problems associated with steep hierarchies (Williams, Paez, & Sanders,
1988).
As Williamson (1995) acknowledges, from a transaction cost perspective, the
institutional environment is treated as the locus of shift parameters, changes in which
shift the comparative costs of govemance. Thus, the benefits of an intemal capital market
is relative to the costs of market fransactions as well as to the costs of intemal
govemance. As is evidenced from the above authors, not only are the costs of the market
decreasing but also the cost of govemance is increasing, resuhing m two fundamental
drivers reducing the efficacy of a hierarchical stmcture based on the premise of an
mtemal capital market. However, the question tiien becomes, why hasn't tiie firm taken
steps to adapt to these changes. As Williamson (1995) furtiier suggests, each form of
104
govemance differs systematically in its capacity to adapt, and as was evidenced by IBM
for many years. It took an environmental jolt to finally change tiieir dominant logic
(Bettis & Prahalad, 1995) concemmg tiie viability and dommance of tiie main frame
computer. Thus, tiiere are multiple reasons why firms have become and remam highly
diversified and likewise, multiple reasons for corporations to resist change by not
refocusmg and realignmg their stiiicture to an optimal level of diversification. Even
though evidence is begmning to mount that high levels of diversification is not a means
through which business units may attam higher performance, h is important to note that
some highly diversified companies, during the time period of this study, did refocus
rather than remain at a high level of diversification or even increase thefr level of
diversification, e.g., Tyco Intemational. Monsanto Corporation, a muhibillion dollar
multinational, is a case in point. At the beginning of the time period, 1991, Monsanto
had six business units operating in various sectors of the economy but by 1997, Monsanto
had refocused to just three segments, each operating within the SIC 28 sector.
The final broad category of theories, market power economies, also suggests that
moderate to highly levels diversification is a means through which busuiess units may
obtain higher performance. However, the mechanisms through which performance
advantages are achieved is through mutual forbearance or cross subsidization among hs
busmess units. As articulated in the resuhs section, business units within moderate to
highly diversified corporations did not eam, on average, above average retums m
comparison to smgle stand-alone firms. As only general pattems witiim tiie data can be
105
discussed, I was not able to mvestigate if the suggested mechanisms of market power
were operating.
However, drawing from a cun-ent event broadcast on CBS News on March 7,
2001, it is mterestmg to note tiiat pharmaceutical companies, such as Merck & Company,
have drastically cut prices for HIV dmgs sold in Africa to about one-tenth of tiiefr US
price. As noted on their web page, this move was in part a response to increasing
competitive pressure from hidian generic dmg manufactures. Interestmgly, altiiough the
prices were drastically reduced in Africa, it is unportant to note the prices witiiin the US
will stay at the former, higher prices because of patent protection. Without making too
much of a leap, I believe this could be considered a form of cross subsidization between
business units (although this is intemational rather than domestic) and the implications
for performance as well as institutional responsibility will be profound. Nevertheless,
this type of cross subsidization in response to competition, assuming domestic operations
only, would have been picked up in the analysis. However, what the analysis could not
pick up would be the long-term effects (decade) of this cross subsidization to the long
range profit of the business unit. If the move by Merck is successful, the busuiess unit
would be able to sell the dmgs at tiie former high price, or possibly, at an even higher
price with a resulting increase in performance attributed to the busmess unit. Therefore,
within the time frame of this analysis and given the above caveat, the resuhs do not
suggest that a market power perspective confributed to the performance of business units
embedded withm highly diversified corporations.
106
A final pattern in the data, not previously discussed, is the relationship of the
average number of busmess units per corporation to a given level of diversification. As
noted m Figure 5.1, as a firm begms to diversify from a smgle stand-alone busmess, tiie
average number of BUs associated with dominate firms is higher than for muhibusiness
corporations at tiie next, higher level of diversification. Witiiout data to confirm my
speculation, it appears tiiese firms may be using BUs as trial balloons to test tiie
diversification waters, and then, once an industry or customer group has been chosen, the
firm refocuses on a smaller set of BUs or possibly, retuming to a smgle organizational
stmcture but in a new market. This pattem was effectively followed by a Swedish fum,
Stora, a 700 year old company (de Geus, 1997). Because of the projected loss of a key
resource, copper, Stora began to experiment in new product markets. Once the benefits
of a new market were determined, the company then shifted business into this new area,
in this case, it was forestry. Eventually, this same scenario was repeated with the firm
moving into iron smelting and eventually into its current markets, wood pulp and
chemicals. With the possible scenarios and directions a firm may take, once h becomes a
dominate firm, I amusingly refer to this level of diversification as the meltmg pot of
diversified corporations and is one of the reasons this category of firms was treated as a
separate and unique category among all corporations.
Havmg discussed the implications of the resuhs and suggest, mdirectly, how tiiese
findings have and have not supported the basic theoretical perspectives conceming
diversification, I now discuss some of tiie implications of my findings.
107
Implications
The overarchmg conclusion that may be drawn from this study is that business
units, embedded withm corporations operatmg m muhiple product markets, do eam, on
average, a statistically significant higher retiim over that attamed by smgle stand-alone
businesses. Furthermore, this performance premium holds across the spectrum of
dominate through moderate/high diversified firms. However, business units embedded
within firms at high levels of diversification do not, on average, eam significantly
superior retums over single stand-alone firms. In short, these findings begin to answer a
most fimdamental question underpinning the research on diversification, "Do
corporations improve business performance?" with a resounding, but qualified, yes.
Although a well defined, non-monotonic relationship is observed between
diversification and performance (Figure 5.1), the differences in performance among
business units of multibusiness corporations (excludes single stand-alone firms) were
non-significant. Therefore, no definitive empirical conclusion could be drawn conceming
the performance superiority of one theoretical perspective over the other, e.g., operational
synergies versus financial market theories.
The findings conceming the linkage between diversification and performance
suggest that, on average, as the level of diversification increases, performance mcreases
to a point and then decreases producing an mverted U-shaped relationship. From this
demonstrated Imkage, albeit mdirectly, several implications may be drawn. First, high
modes of diversification may not be a viable path through which a busmess unit may
108
obtam above average retums over that of smgle stand-alone firms. With shifts in the
institutional environment occurring m the new competitive landscape and the mcreasmg
costs of intemal govemance, the premiums that once accmed to highly diversified
corporations are, quite possibly, beuig dissipated. Second, smgle stand-alone firms may
be at a competitive disadvantage due to the absence of other business units upon which to
leverage resources or activities or implement an intemal capital market. Furthermore,
sfrategic alliances between independent firms seems not to be a viable ahemative through
which to build economies of scope and thus, attain performance equivalence with hs
multibusiness manufacturing rivals. Thfrd, in conjunction with the second implication,
although the sharing of activities or the leveraging of resources may be necessary to attain
high performance, it may not be a sufficient condition to sustain high performance
(persistence of retums). Fourth, even though synergies might enable a firm to eam higher
retums, as demonstrated by the highest performance accramg to low to moderately
diversified firms, product innovation may be an unpairment to the sharing and leveraging
of resources and capabilities (synergy creation). In a study of 412 high-iimovative
projects and 375 low-mnovative projects. Song and Parry (1999) found that product
mnovativeness weakens tiiree key relationships that determine new product success, one
of which was the impact of technical synergy on technical proficiency. The unportance
of technological synergy to product development (one of several critical success factors)
was also supported by Cooper (1990) in a sttidy of 203 new product projects m 125
mdustrial product firms. Thus, witii tiie unperatives of technological synergy for new
109
product development and product mnovativeness to compete m a global economy (new
competitive landscape), a diversified firm may be faced with tradeoffs and tensions in
trying to create synergy (to improve performance) and pursue product innovation (to
remain competitive). Discovering and possibly resolving these tensions would be an
important area for future research m the new millennium.
110
O **- CD ^ tn —
< > < i > > - n i a ) ( u 2 CO l O T T
— ! • 1 ^
I <U U) U>
3 £ J3 4) i_
m
S 5
X Si S o
•a s o
1 Si 0) > ^
"8 b
3
C
m
£ Z T3 C n
•8 •S) s o « i« CD
Q
1 0)
Q
S (M
>• If)
m m
3 c — — w £ t O < t 0 ' ~ 0 0 3 " ' — C ( U < f l < »
c o
o u
3
u
en C
o & o U U
0)
>
o
3
"o u
1)
c/1
O o
C
D en (/} u
_C 3
CQ
4J
B 3
z •T3
C CQ
< o c CQ U
o O
u l - l 3
111
Table 5.1. Mean and Standard Error Estunates of Busmess Umts ROAs Across the Spectrum of Diversified Corporations.
Least Squares Means
Level of Diversification
Single Stand-AIone Corporations Dominant Corporations Least Diversified Corporations Low to Moderate Diversified Corporations Moderate to High Diversified Corporations Highly Diversified Corporations
LS MEAN
7.80
10.76
12.24
12.51
11.14
9.91
Std. Error
0.519
0.932
1.121
0.912
1.273
1.256
DF
16E3
16E3
16E3
16E3
16E3
16E3
t
15.02
11.54
10.92
13.71
8.89
7.79
Pr>|t|
0.0001
0.0001
0.0001
0.0001
0.0001
0.0001
DF: Degrees of Freedom. Std. Error: Standard error of the mean estimate. t: t - value Pr>|t|: p-value.
Random Effects
Covariance Parameter Estimates (Maximum Likelihood Estimates)
Covariance Parameter Industry effects Corporate effects Business effects Residual
Estimate
20.825 31.373
221.840 170.123
% of Total Variance
4.69% 7.06%
49.95% 38.30%
Standard Error 3.981 5.890 8.132 1.923
Z
5.23 5.33
27.28 88.46
Pr> |Z|
0.0001 0.0001 0.0001 0.0001
Sample size: 19,725. Number of Corporations: 2,341. Number of Business Units: 3,838. Number of Industry Sectors: 589. Level of Diversification: 6 levels.
112
Table 5.2. Test of Differences in Means Between Diversified and Non-Diversified Corporations.
Fixed Effects
Least Squares Means
Diversified Corporations
Non-diversified Corporations
LSMEAN
11.448
7.789
Std. Error
0.529
0.520
DF
16E3
16E3
t
21.65
14.97
Pr> | t |
0.0001
0.0001
Differences of Least Squares Means Level of Diversification
of Corporations
Diversified Non-diversified
Difference in LSMEAN
3.659
Standard Error
0.652
DF
16E3
t
5.62
Pr>|t|
0.0001
DF: Degrees of Freedom.
Random Effects
Covariance Parameter Estimates (Maximum Likelihood Estimates)
Covariance Parameter Industry effects Corporate effects Business effects Residual
Estimate
21.071 31.850
221.665 170.117
% of Total Variance
4.74%
7.16% 49.85% 38.25%
Standard Error 4.008 5.908 8.132 1.923
Z
5.26 5.39
27.26 88.47
Pr>|Z|
0.0001 0.0001 0.0001 0.0001
Observations: 19,725. Number of Corporations: 2,341. Business Units: 3,838. Industry Sectors: 589. Level of Diversification: 2 levels.
113
Table 5.3. Test of Differences m Means Among Busmess Units Embedded Within Corporations of Varying Levels of Diversification.
Fixed Effects
Differences of Least Sq Levels of Diversification
of Corporations Least DMBC
Least DMBC
Least DMBC
Least DMBC
Low/Moderate DMBC Low/Moderate DMBC Low/Moderate DMBC Moderate/High DMBC Moderate/ffigh DMBC ffighly DMBC
Dominant
Low/Moderate DMBC Moderate/High DMBC Highly DMBC Dominant
Moderate/High DMBC ffighly DMBC Dominant
ffighly DMBC Dominant
Difference in LSMEAN
-1.480
-0.269
1.096
2.330
-1.749
1.366
2.599
-0.384
1.233
0.849
uares Means Standard
Error
1.415
1.396
1.641
1.661
1.256
1.504
1.526
1.516
1.746
1.537
DF
16E3
16E3
16E3
16E3
16E3
16E3
16E3
16E3
16E3
16E3
t
-1.05
-0.19
0.67
1.40
-1.39
0.91
1.70
-0.25
0.71
0.55
Pr > |t|
0.2956
0.8472
0.5040
0.1608
0.1638
0.3638
0.0886
0.8001
0.4801
0.5806
DMBC: Diversified Multibusiness Corporations. DF: Degrees of Freedom.
Random Effects
Covariance Parameter Estimates (Maximum Likelihood Estimates)
Covariance Parameter Industry effects Corporate effects
Business effects Residual
Estimate
20.825 31.373
221.840 170.123
% of Total Variance
4.69% 7.06%
49.95% 38.30%
Standard Error 3.981 5.890 8.132 1.923
Z
5.23 5.33
27.28 88.46
Pr> |Z|
0.0001 0.0001 0.0001 0.0001
Observations: 19,725. Number of Corporations: 2,341. Number of Business Units: 3,838. Number of Industry Sectors: 589. Level of Diversification: 5 levels.
114
Table 5.4. Tests of Differences ui Means Between Busmess Units Embedded Within Diversified Corporations and Single Stand-AIone Businesses.
Fixed Effects
Differences of Least Squares Means
Levels of Diversification of Corporations
Dominant
Least DMBC
Low/Moderate DMBC
Moderate/ffigh DMBC
Highly DMBC
Single
Single
Single
Single
Single
Difference in LSMEAN
2.959
4.440
4.709
3.343
2.110
Standard Error
1.008
1.179
0.986
1.311
1.334
DF
16E3
16E3
16E3
16E3
16E3
t
2.93
3.77
4.77
2.55
1.58
Pr> | t |
0.0034
0.0002
0.0001
0.0108
0.1138
DMBC: Diversified Multibusiness Corporations. DF: Degrees of Freedom.
Random Effects
Covariance Parameter Estimates (Maximum Likelihood Estimates)
Covariance Parameter Industry effects Corporate effects Business effects Residual
Estunate
20.825 31.373
221.840 170.123
Sample size: 19,725. Number of Corporations: 2,341. Number of Business Units: 3,838. Number of Industry Sectors: 589. Level of Diversification: 6 levels.
% of Total Variance
4.69% 7.06%
49.95% 38.30%
Standard Error 3.981 5.890 8.132 1.923
Z
5.23 5.33
27.28 88.46
Pr>|Z|
0.0001 0.0001 0.0001 0.0001
115
.a>i=
a> CM
/ ;
^ Q CD
i n CO CO
O ffi
•a c a
E o
O • a c a 01
c Rl
s
C IS
O
c CD
X 0) ^ S o •S S IO
S
0)
II" S !! ffi
i f l
ix:
1 « .S g
CO 0>
C I O
8 c
^ 2 CO O) CM
« .£!
CO « S v \
M
« fe o
c 10 CA II
Q
o 3 C ~ • - M
c T ) c 1 ^ CO £ «
S o m e a: O < o <•- m 3 «
116
Table 5.5. Test of Significance of Identifiable Assets, as a Fixed Effect, and ROA.
Fixed Effects
Effect
Identifiable Assets
Estimate
-0.00000972
Std. Error
0.00005241
DF
16E3 0.19
P r >
0.8529
Identifiable assets: Millions of dollars.
Random Effects
Covariance Parameter Estimates (Maximum Likelihood Estunates)
Covariance Parameter Industry Corporate Firm Effects Residual
Estimate
20.825 31.373
221.840 170.123
% of Total Variance
4.69% 7.06% 49.95% 38.30%
Standard Error 3.981 5.890 8.132 1.923
Z
5.23 5.33
27.28 88.46
Pr> |Z|
0.0001 0.0001 0.0001 0.0001
Sample size: 19,725. Number of Corporations: 2,341. Number of Business Units: 3,838. Number of Industry Sectors: 589. Level of Diversification: 6.
117
-- N.
o
/ / / / '/
- - CD
O a. o o 10 CO 0) c
'lO 3 .a
o c .g (0 u
§
>
. 7
_£ J3 (0 •c > tn 3 O 3 C
o
• / 1 /
'* /' X
o «) U) ni 0) .N
m c tic
(S
a. o
a. 3 o O) o
nt
73 <D N • c o a> 0) (D O
£ 8 en c o
o
\
-- CM
<0 c
*- o CO 2
.£ o 00 E 9-o o o ° Q O
0) c
« -a (o c n E
N- (0 « i l 00
s
CO
Q: o <
c»
O — CD D
(O c 3
09
D OQ
118
CHAPTER VI
CONCLUDING COMMENTARY
Limitations of Studv
The mam Ifrnitations of this study are tiie time period over which tiie data are
drawn, tiie mability to make a definitive determination of risk-adjusted retiims, and use of
one performance measure.
The data was drawn from a business cycle that has been unprecedented witii
regards to growtii witiim tiie United States. However, on a global perspective, researchers
have argued that the world economy has entered mto a new competitive landscape (NCL)
(Hitt, Keats, & DeMarie, 1998) with the degree of mtemational competition increasing,
product life-cycles bemg dramatically shortened, and the critical need for contmuous
organizational change to navigate effectively within the mcreasing turbulence ui the
competitive landscape (Bettis & Hitt, 1995; Hitt, Keats, & DeMarie, 1998). This stiidy is
not able to determine the degree to which performance was affected by domestic versus
intemational economic trends and competition.
A second limitation to this study is a result of a tradeoff in research design. To
address the effect of corporate strategy on business units, it was necessary to conduct the
study at the business unit level of analysis; however, this negates the ability to determine
a risk-adjusted retum. Information on firm performance relative to the stock market may
be obtained at the corporate level, but this information is not disaggregated to the
119
busmess unit. A sharp measure could be used to estimate total variability in busuiess unit
ROA; however, the primary mterest is the degree of systematic risk that is associated with
the business urut. At best, risk-adjusted retums at the BU level of analysis could be
derived from mdustry estunates. Because of the subjectivity and lack of precedence, this
approach was not investigated fiirther.
Another limitation to the study is the use of one measure for performance, ROA.
However, the use of an accounting measure is consistent with a preponderance of past
research on corporate effects and diversification and thus, aids in the commensurability of
this study to the research in both domains. Furthermore, Hoskisson et al. (1993) found no
statistically significant, direct relationship between the entropy measure and market value
premiums; e.g., Tobin's q. The lack of criterion-related validity between these two
measures was an important determinant in not using Tobin's q as a second performance
measure.
A final limitation of this study is its inability to determine the causality between
the constmcts. In surfacmg the implicit assumptions concemmg the causal relationship
between diversification and performance, underpiiming many of the diversification
theories articulated m this study I would argue most of the theories, grounded in the
premise of economies of scope, imply that causality flows from diversification to
performance (diversification is exogenous and performance is endogenous) as shown m
the simple diagram. For example, diversification is implemented to maximize
shareholder value.
Diversification > Performance
120
However, rather than a firm implementing a diversification strategy to increase
shareholder wealth, diversification may be implemented because o/below average
performance. In this scenario, the geneity of the constmcts are just reversed with
diversification becoming — endogenous and performance becoming — exogenous. Low
performance may be due to a variety of factors. Drawmg from a earlier discussion, Stora,
a 700-hundred-year-old company, began to diversify because of the depletion of a key
resource maturity. Smith and Cooper (1988) suggest firms in mature or maturing
industries sometimes find it necessary to diversify to survive over the long run. These
examples help to demonstrate that the causality is reversed, in the above examples, low
perfonnance is the underlying incentive to diversify rather than profit maximization. This
causal relationship may be depicted in the foUowmg diagram.
Diversification < Performance
An alternative view of the relationship between diversification and performance
may be drawn from Peru-ose (1959) when she posited the idea of a 'virtuous circle,'
whereby a firm grows m order to take advantage of surplus resources, and, m so domg,
acquires additional surplus resources that encourage yet more growth. Based on her
premise, it may argued that the relationship between diversification and performance is
reciprocal, as depicted in the foUowmg diagram.
Diversification ^ Performance
<
The implications of the various patiis of causahties have direct bearmg on the
modelmg technique that may be used by a researcher. For instance, it may be possible to
121
model the above relationships usmg a stmctural equation model using a recursive path
between the two constmcts and then, if the model runs, determine if one or both paths are
significant. However, with any modeling technique, causality is very difficult, if not
impossible, to accurately determine. Therefore, it is precisely this point that I choose to
investigate the pattems in the data using mean ROA estimates of business units
embedded in multibusiness corporations with varying levels of diversification rather than
making an explicit assumption (and possible erroneous) conceming the causal dfrection
between diversification and performance.
Caveat
In 1997, the SIC system was changed to tiie North American Industrial
Classification System (NAICS) to address tiie rapid changes occurring in botii tiie United
States and world economies as noted m Table 6.1. In this fransition, several changes have
occurred m how corporations are classified. The manufactiiring sector is now denoted by
NAICS 31-33, whereas under tiie former SIC system, the manufactiiring sector was
classified as SIC 20-30. Otiier changes tiiat have occurred withm tiie manufactiiring
sector classification system are as follows: (1) manufacttiring mdustries included m
NAICS mcreased from 459 to 474, with tiie most significant change bemg the creation of
tiie computer and elecfronic product manufactiirmg subsector, (2) publishmg was moved
out of tiie manufacttiring sector, (3) loggmg was moved to tiie agricultiire sector, and (4)
122
bakeries tiiat bake on tiie premises and custom manufactiiring were moved into tiie
manufacturing sector.
In sum, tiie largest changes m classification occurred witiun tiie manufacttiring
sector ratiier tiian movmg mdustiies m or out of tiie sector. The changes tiiat did occur
witiun tiie sector seem to have resuhed m a more fimed gramed specification of tiie
different processes m manufacturing and tiius, do not undermme tiie fimdamental value of
tiie information contamed withm tiie data set. Additionally, tables are readily available so
that SIC and NAICS codes may be cross-referenced.
Contributions
This study contributes to the sfrategic management literature m several ways.
More explicitly, the first contribution of this study answers the calls within strategic
management to conduct research that controls for industry effects. However, research on
corporate effects (Bowman & Helfat, 2001) has shown that, in addition to industry
effects, corporate and business effects are also important determinants of business unit
profitability. In this study, the analysis of the data not only controls for industry effects,
but also for corporate and business effects in the determination of the linkages between
levels of diversification and business unit performance.
The second contribution is the rigorous determination of the shape and form of the
linkage between levels of diversification and busuiess unit performance across the entfre
spectmm of diversified corporations withm the manufacturing sector (mcluding single
123
stand-alone businesses), hi addition to graphically demonsfrating tiie relationship, tiie
sttidy established if statistically significant differences existed not onfy among busmess
units of diversified corporations but also among business units of diversified firms and
single stand-alone businesses while confrollmg for all effects.
The final contribution answers tiie call to use tiie business unit as tiie level of
analysis m detemuning tiie effect of strategy and BU rivafry on business unit
performance, hi sum, tiie stiidy deterauned if corporations make businesses better off and
answers one of the most fimdamental questions underpmning the research on corporate
strategy.
In sum, the test of time will be the final arbitrator of the significance of this study
and the degree to which tiie study has contributed to botii the literature on corporate
effects and sfrategic management. Humbly, I do believe this study has made inroads
towards a reconciliation of the diversification-performance paradox.
Future Research
This study has laid the foundation to pursue multiple avenues for fiiture research.
With the form and shape of the linkage between diversification and performance across
the diversification spectrum statistically determined, it would be informative to
investigate the robustness of this relationship across the suggested areas m future
research.
124
lormance Ghemawat (1991) suggests tiiat "sticky factors" underpm sustained perfo
differences and vary m tiiefr unportance from industiy to industiy. The autiior suggests
tiiese factors may be grouped mto three broad classes of capacity, customer base, and
knowledge, and as tiieory suggests, competition over tiiese classes of factors take
somewhat different forms. Additionally, tiie autiior argues, while some mdustries cannot
be classified as such, tiie empirical evidence indicates tiiat, at least in tiie manufactiiring
sector, the industry can be sorted relatively cleanly into three categories, with capital
mtensity, advertismg-intensity, and R&D mtensity serving as proxies for the unportance
of the above three factors.
Drawing from this msight, one suggestion for futiire research would be to cluster
the manufacturing sector into these three dimensions and then determine the linkage
between diversification and performance for each set of firms withm the cluster. This
fine grained analysis could then be compared and confrasted to the findings derived at the
sector level of analysis.
A second area for future research would be to extend the analysis to include the
degree to which intemationalization affects the diversification-performance relationship
within the manufacturing sector. Some scholars have suggested that an inverted U-
shaped relationship exists between diversification and performance (Geringer, Beamish,
& da Costa, 1989), but few studies have investigated both product and global
diversification (Hitt, Hoskisson, & Kim, 1997; Tallman & Li, 1966).
125
A thfrd area for fiiture research would be to extend this original study of the
manufacturing sector to include all mdustry sectors, comparing and contrasting each
sector. This investigation would mform the research on diversification as to the
generalizability or not, across all industry sectors, of the non-monotonic relationship
between diversification and performance.
126
Standard Industrial Classification/North American Industrial Classification System
The New Hierarchical Stmcture: NAICS'
XX Industry Sector (20 broad sectors up from 10 SIC) XXX Industry Subsector XXXX Industry Group XXXXX Industry XXXXXX US, Canadian, or Mexican National specific
*NAICS industries are identified by a 6-digit code, in contrast to the 4-digit SIC code. The longer code accommodates the larger number of sectors and allows more flexibility in designating subsectors. It also provides for additional detail not necessarily appropriate for all three NAICS coimtries. The intemational NAICS agreement fixes only the first five digits of the code. The sixth digit, where used, identifies subdivisions of NAICS industries that accommodate user needs in individual countries. Thus, 6-digit US codes may differ fi-om coimterparts in Canada or Mexico, but at the 5-digit level they are standardized.
NAICS: 20 Sectors
Code 11 21 22 23 31-33 42 44-45 48-49 51 52 53 54 55 56 61 62 71 72 81 92
NAICS Sectors Agriculture, Forestry, Fishing and Hunting Mining Utilities Construction Manufacturing Wholesale Trade Retail Trade Transportation and Warehousing Information Finance and Insurance Real Estate and Rental and Leasing Professional, Scientific, and Technical Services Management of Companies and Enterprises Administrative and Support and Waste Management and Remediation Services Education Services Health Care and Social Assistance Arts, Entertainment, and Recreation Accommodation and Food Services Other Services (except Public Administration) Public Administration
* Adapted from NAICS Association, October 20, 1999.
Figure 6.1. SIC/NAICS Classification System.
127
REFERENCES
Amihud, Y., & Lev, B. (1981). Risk Reduction as a Managerial Motive for Conglomerate Mergers. BellJournal of Economics, 12, 605-617.
Amit, R. & Livnat, J. (1988a). Concept of Conglomerate Diversification. Strategic Management Journal, 14, 593-604.
Amit, R., & Livnat, J. (1988b). Diversification Strategies, Busmess Cycles, and Economic Performance. Strategic Management Journal, 9, 99-110.
Amit, R. & Schoemaker, P. J. H. (1993). Strategic Assets and Organizational Rent. Strategic Management Journal, 14, 33-45.
Andrews, K.R. (1971). The Concept of Corporate Strategy. Homewood, IL: Dow-Jones Irwin.
Ansoff, H. I. (1965). Corporate Strategy. New York: McGraw-Hill.
Argyres, N. (1996). Capabilities, Technological Diversification and Divisionalization. Strategic Management Journal, 17, 395-410.
Bamey, J. B. (1997). Gaining and Sustaining Competitive Advantage. Readmg, MA: Addison-Wesley Publishing Co.
Bamey, J. B. (1991). Firm Resources and Sustained Competitive Advantage. Journal of Management, 17,99-120.
Bamey, J. B., & Ouchi, W. G. (1986). Organizational Economics. San Francisco: Jossey-Bass.
Baninger, B. R., & Harrison, J. S. (2000). Walkmg a Tightrope: Creating Value Through Interorganizational Relationships. Journal of Management, 26, 367-403.
Bartlett, C. A., & Ghoshal, S. (1991). Managing Across Borders: The Transnational Solution. Cambridge: Harvard University Press.
Baysinger, B. D., & Hoskisson, R. E. (1989). Diversification Strategy and R&D hitensity m Large Muhiproduct Firms. Academy of Management Journal, 32, 310-332.
128
Berger, J. O., & Sellke, T. (1987). Testmg a Pomt Null Hypotiiesis: The frreconcilability of P values and Evidence. Journal of American Statistical Association, 82, 112-122.
Bergh, D. D. (1995). Size and Relatedness of Units Sold: An Agency Theory and Resource-based Perspective. Strategic Management Journal, 16, 221-239.
Bettis, R. A., & Hall, W. K. (1982). Diversification Strategy, Accountmg Determmed Risk, and Accountmg Determined Retum. Academy of Management Journal, 25, 254-264.
Bettis, R. A., & Hall, W. K. (1983). The Business Portfolio Approach — Where h Falls Down m Practice. Long Range Planning, 16 (2), 95-104.
Bettis, R. A. & Hitt, M. A. (1995). The New Competitive Landscape. Strategic Management Journal, 75 (special issue), 7-19.
Bettis, R. A., & Prahalad, C. K. (1995). The Dominant Logic: Retrospective and Extension. Strategic Management Journal, 16, 5-14.
Boal, K. B., & Willis, R. E. (1983). A Note on the Armstrong/Mitroff Debate. Journal of Management, 9 (2), 203-211.
Bowman, E. (1974). Epistemology, Corporate Strategy, and Academe. Sloan Management Review, 15, 35-50.
Bowman, E.H. (1982). Risk Seeking by Troubled Firms. Sloan Management Review, 23, 33-42.
Bowman, E. H., & Helfat, C. E. (2001). Does Corporate Strategy Matter? Strategic Management Journal, 22, 1-23.
Bromiley, P. (1986). Corporate Capital Investment: A Behavioral Approach. New York: Cambridge University Press.
Bmsch, T. H. (1996). Predicted Change m Operational Synergy and Post-acquisition Performance of Acquired Busmesses. Strategic Management Journal, 17, 1-24.
Brush, T. H., & Bromiley, P. (1997). What Does a Small Corporate Effect Mean? A Variance Components of Corporate and Busmess Effects. Strategic Management Journal, 18, 825-835.
129
Bmsh, T. H., Bromiley, P., & Hendrickx, M. (1999). The Relative hifluence of hidustry and Corporation on Busmess Segment Performance: An Ahemative Estimate. Strategic Management Journal, 20, 519-547.
Cannella, A. A. Jr., & Monroe, M. J. (1997). Contrastmg Perspectives on Strategic Leaders: Toward a More Realistic View of Top Managers, Journal of Management, 23, 213-237.
Carroll, G.R. (1993). A Sociological View on Why Firms Differ. Strategic Management Journal, 14, 237-249.
Caves, R. E., & Porter, M. E. (1977). From Entry Barriers to Mobility Barriers: Conjectiiral Decisions and Contrived Deterrence to New Competition. Quarterly Journal of Economics, 91, 241-261.
Caves, R. E., Porter, M. E., & Spence, M. (1980). Competition in the Open Economy. Cambridge: Harvard University Press.
CBS News (2001, March 7). Cheaper HIV Dmgs Offered to Africa. http://www.cbsnews.com/now/story/0,1597,277076-412,00.shtml
Chandler, A. (1962). Strategy and Structure: Chapters in the History of American Industrial Enterprise. Cambridge: The MIT Press.
Chandler, A. (1977). The Visible Hand: The Managerial Revolution in American Business. Cambridge: Harvard University Press.
Chandler, A. D., Jr. (1991). What Are tiie Functions of tiie HQ Unit m a Multibusmess Firm? Strategic Management Journal, 72,31-50.
Chang, S., & Singh, H. (2000). Corporate and Industry Effects on Business Unit Competitive Position. Strategic Management Journal, 21, 739-752.
Chatterjee. S., & Wemerfelt, B. (1991). The Link Between Resources and Type of Diversification: Theory and Evidence. Strategic Management Journal, 12, 33-48.
Chenhall, R. H. (1984). Diversification Within Australian Manufacturing Enterprise. Journal of Management Studies, 21 (I), 23-60.
Christensen, H. K., & Montgomery, C. A. (1981). Corporate Economic Performance: Diversification Strategy versus Market Stmcture. Strategic Management Journal, 2, 327-343.
130
CoUis, D. J., & Montgomery, C. A. (1998a). Corporate Strategy. Boston: Irwin/McGraw Hill.
Collis, D. J., & Montgomery, C. A. (1998b). Creatmg Corporate Advantage. Harvard Business Review, 76 (3), 71-83.
COMPUSTAT (1997). Standard & Poor's COMPUSTAT. Englewood, CO: The McGraw-Hill Companies, Inc.
Conner, K.R. (1991). A Historical Comparison of Resource-based Theory and Five Schools of Thought Within the Industrial Organization Economics: Do We Have a New Theory of the Firm. Journal of Management, 17, 121-154.
Conner, K.R. (1994). The Resource-Based Challenge to tiie Industry-Stmcture Perspective. Academy of Management Best Paper Proceedings, 17-21.
Connolly, T., Conlon, E. J., & Deutsch, S. J. (1980). Organizational Effectiveness: A Multiple-Constituency Approach. Academy of Management Review, 5, 211-217.
Confractor, F. J., & Lorange, P. (1988). Why Should Firms Cooperate: The Strategy and Economics Basis for Cooperative Ventures. In F. J. Contractor & P. Lorange (Eds.) Cooperative Strategies in International Business, 3-30. New York: Lexington Books.
Cooper, R. G. (1990). New Products: What Distinguishes the Wiimers? Research-Technology Management, 33 (6), 27-31.
Copeland, T. E., & Weston, J. F. (1983). Financial Theory and Corporate Policy. Reading, MA: Addison-Wesley.
Crock, S. (1997, October 27). Can This Farm Boy Keep Lockheed m Orbit? Business PFeeA:, 103-115.
Daft, R. L. (1998). Organization Theory and Design (6thEd.). Cmcmnati, OH: Soutii-Westem College Publishmg.
Davis, P., Robmson, R., Pearce, J., & Park, S. (1992). Busmess Unit Relatedness and Performance: A Look at the Pulp and Paper hidustry. Strategic Management Journal, 13,349-361.
Davis, R., & Duhaime, I. M. (1989). Busmess Level Data Disclosure Under the FASB No. 14: Effective use m Strategic Management Research. Academy of Management Best Paper Proceedings, 7-11.
131
Davis, R., & Duhaime, I. M. (1992). Diversification, Vertical Integration, and hidustry Analysis: New Perspectives and Measurement. Strategic Management Journal, 73,511-524.
Davis, R., & Thomas, L. G. (1993). Dfrect Estimation of Synergy: A New Approach to tiie Diversity-Performance Debate. Management Science, iP (11), 1334-1346.
de Geus, A. (1997). The Living Company: Habit for Survival in a Turbulent Environment. Boston: Harvard Business School Press.
Demsetz, H. (1973). Industry stmcture, market rivalry, and public policy. Journal of Law and Economics, 16, 1-9.
Denis, D., Denis, J., & Atulya, D. K. (1997). Agency Problems, Equity Ownership, and Corporate Diversification. Journal of Finance, 52 {\), 135-160.
Dess, G. G., Gupta, A., Hennart, J., & Hill, C. W. (1995). Conductmg and hitegrating Strategy Research at the Intemational, Corporate, and Business Levels: Issues and Directions. Journal of Management, 21, 357-393.
Dess, G. G., freland, R. D., & Hitt, M. A. (1990). hidustry Effects and Strategic Management Research. Journal of Management, 16, 7-27.
Dial, J., & Murphy, K. J. (1995). Incentives, Downsizmg, and Value Creation at General Dynamics. Journal of Financial Economics, 37, 261-314.
Dierickx, L. & Cool, K. (1989). Asset Stock Accumulation and Sustamability of Competitive Advantage. Management Science, 35, 1504-1511.
DiMaggio, P. J., & Powell, W. W. (1983). The fron Cage Revisited: histitutional Isomorphism and Collective Rationality m Organizational Fields. American Sociological Review, 48, 147-160.
Dyer, J. H., & Singh, H. (1998). The Relational View: Cooperative Strategy and Sources of Interorganizational Compethive Advantage. Academy of Management Review, 23, 660-679.
Fmkelstem, S., & D'Aveni, R. A. (1994). CEO Duality as a Double-Edged Sword: How Boards of Directors Balance Entrenchment Avoidance and Unity of Command. Academy of Management Journal, 37, 1070-1108.
132
Fmkelstem, S., & Hambrick, D. C. (1989). Chief Executive Compensation: A Sttidy of tiie hitersection of Markets and Political Processes. Strategic Management Journal, 10, 121-134.
Fmkelstem, S., & Hambrick, D. C. (1996). Strategic Leadership: Top Executives and Their Effects on Organizations. St. Paul, MN: West Publishmg Company.
Fligstem, N. (1985). The Spread of tiie Muhidivisional Form Among Large Firms, 1919-1979. American Sociological Review, 50, 377-391.
Fligstem, N. (1990). The Transformation of Corporate Control. Cambridge: Harvard University Press.
Galbraith, C , Samuelson, B., Stiles, C, & Merril, G. (1986). Diversification, hidustry Research and Development and Performance. Academy of Management Proceedmgs, 46th annual meetmg. Academy of Management, 17-20.
Gallop, F. M., & Monahan, J. (1991). A Generalized hidex of Diversification: Trends in US Manufacturing. Review of Economics and Statistics, 73 (2), 318-330.
Galunic, D. C , & Eisenhardt, K. M. (1994). Reviewmg the Strategy-Stmcture-Performance Paradigm. In B. M. Staw and L. L. Cummmgs (Eds.) Research in Organizational Behavior, 215-255. London: JAI Press, Inc.
Geringer, L. M., Beamish, P. W., & da Costa, R. C. (1989). Diversification Strategy and Intemationalization: Implications for MNE Performance. Strategic Management Journal, 10, 109-119.
Ghemawat, P., & Ricart I Costa, J. E. (1993). The Organizational Tension Between Static and Dynamic Efficiency. Strategic Management Journal, 14, 59-73.
Ghemawat, P. (1991). Commitment: The Dynamic of Strategy. New York: The Free Press.
Gimeno, J., & Woo, C. Y. (1999). Multunarket Contact, Economies of Scope, and Firm Performance. Academy of Management Journal, 42, 239-259.
Goldhar, J. D., & Jelmek, M. (1983). Planfor Economies of Scope. Harvard Business Review, 61 (6), 141-148.
Goold, M., & Campbell, A. (1987). Strategies and Styles. Oxford: Basil Blackwell, Ltd.
133
Goold, M., & Luchs, K. (1993). Why Diversify? Four Decades of Management Thinkmg. Academy of Management Executive, 7,7-25.
Grant, R.M. (1988). On Donunant Logic Relatedness and the Lmk Between Diversity and Performance. Strategic Management Journal, 9, 639-642.
Grant, R. M. (1995). Contemporary Strategy Analysis: Concepts, Techniques, Applications (2nd ed.). Cambridge: Blackwell Publishers.
Grant, R. M., & Jammme, A. (1988). Performance Differences Between tiie Wrigley/Rumek Strategic Categories. Strategic Management Journal, 9, 333-346.
Grant, R. M. A., Jammine, A., & Thomas, H. (1988). Diversity, Diversification, and Profitability Among British Manufacturing Companies, 1974-1984. Academy of Management Journal, 31, 771-801.
Gray, S. R., & Cannella, A. A. Jr. (1997). The Role of Risk in Executive Compensation. Journal of Management, 23, 517-540.
Gupta, A. K., & Govindarajan, V. (1986). Resource Sharing Among SBUs: Strategic Antecedents and Administtative Implications. Academy of Management Journal, 29, 695-714.
Hansen, G. S., & Wemerfelt, B. (1989). Determmants of firm performance: The relative importance of economic and organizational factors. Strategic Management Journal, 10,399-411.
Hanssen-Bauer, J., & Snow, C. (1996). Responding to Hypercompetition: The Stmcture and Processes of a Regional Leammg Network Organization. Organization Science, 7, 413-427.
Hitt, M. A., Hoskisson, R. E., & freland, R. D. (1990). Mergers and Acquisitions and Managerial Commitment to Innovation m M-Form Firms. Strategic Management Journal, 11, 29-47.
Hitt, M. A., Hoskisson, R. E., & Kfrn, H. (1997). hitemational Diversification: Effects on Innovation and Firm Performance in Product-Diversified Firms. Academy of Management Journal, 40, 767-798.
Hitt, M. A., freland, R. D., & Hoskisson, R. E. (2001). Strategic Management: Competitiveness and Globalization. Boston: Soutii-Westem College Publishmg.
134
Hitt, M. A., & freland, R. D. (1987). Peters and Waterman Revisited: The Unended Quest for Excellence. Academy of Management Executive, 1, 91-98.
Hitt, M. A., Keats, B. W., & DeMarie, S. (1998). Navigating m tiie New Competitive Landscape: Buildmg Sfrategic Flexibility and Compethive Advantage in the 21st Century. Academy of Management Executive, 13, 22-42.
Holzman, O. J., Copeland, R. M., & Hayya, J. (1975). hicome Measures of Conglomerate Performance. Quarterly Review of Economics and Business, 15 (3), 67-78.
Hoskisson, R. E. (1987). Muhidivisional Stmcture and Performance: The Contingency of Diversification Sfrategy. Academy of Management Journal, 30, 625-644.
Hoskisson, R. E., Hill, C. W. L., & Kim, H. (1993). The Muhidivisional Stmctiire: Orgaruzational Fossil or Source of Value? Journal of Management, 19, 269-298.
Hoskisson, R. E., & Hitt, M. A. (1990). Antecedents and Performance Outcomes of Diversification: Review and Critique of Theoretical Perspectives. Journal of Management, 16, 461-509.
Hoskisson, R. E., Hitt, M. A., Johnson, R. A., & Moesel, D. D. (1993). Constmct Validity of an Objective (Entropy) Categorical Measure of Diversification Strategy. Strategic Management Journal, 14, 215-236.
Hoskisson, R. E., Hitt, M. A., Wan, W., & Yiu, D. (1999). Theory and Research m Strategic Management: Swings of a Pendulum. Journal of Management, 25 (Yearly Review of Management Supplement), 417-456.
Hoskisson, R. E., & Turk, T. (1990). Corporate Restmcturing: Govemance and Control Limits of the Intemal Market. Academy of Management Review, 15, 459-477.
Hunt, S. D. (1994). A Realist Theory of Empirical Testmg: Resolvmg The Theory Ladenness/Objectivity Debate. Philosophy of the Social Sciences, 24 (2), 133-158.
Hunt, S. D. (1995). The Resource-Advantage Theory of Competition: Toward Explammg Productivity and Economic Growth. Journal of Management Inquiry, 4,311-332.
Hunt, S. D. (1997). Resource Advantage Theoty: An Evolutionary Theory of Competitive Firm Behavior? The Journal of Economic Issues, 31, 59-77.
135
Jacobson, R. (1987). The Validity of ROI as a Measure of Busmess Performance. American Economic Review, 77, 470-478.
Jacquemm, A. P., & Berry, C. H. (1979). Enfropy Measure of Diversification and Corporate Growtii. Journal of Industrial Economics, 27, 359-369.
Jarillo, J. C. (1988). On Sfrategic Networks. Strategic Management Journal, 9,31-41.
Jensen, M.C. (1986). Agency Costs of Free Cash Flow, Corporate Fmance, and Takeovers. American Economic Review, Papers and Proceedmgs of the 98th Annual Meeting of the American Economic Association, 76, 323-329.
Johnson, S. G. (1981). Unlucky 13. Collegiate News and Views, 35 (1), 11-13.
Jose, M. L., Nichols, L. M., & Stevens, J. L. (1986). Contributions of Diversification, Promotion, and R&D to the Value of Muhiproduct Firms: A Tobm's q Approach. Financial Management, 15 (4), 33-81.
Kaback, H. (2000). Complacency is not an Option. Directors & Boards, 24 (3), 14-22.
Kaplan, A. (1963). The Conduct of Inquiry: Methodology for Behavioral Science. New York: Harper and Row.
Keats, B. W., & Hitt, M. A. (1988). A Causal Model of Linkages Among Environmental Dimensions, Macro Organizational Characteristics, and Performance. Academy of Management Journal, 31, 570-598.
Keown, A. J., Martm, J. D., Petty, J. W., & Scott, Jr., D. F. (2000). Foundations of Finance (3rd Ed.). Upper Saddle River, NJ: Prentice Hall.
Kessides, LN. (1990). Intemal versus Extemal Market Conditions and Ffrm Profitability: An Exploratory Model. The Economic Journal, 100, 773-792.
Kochhar, R., & Hitt, M. A. (1998). Research Notes and Communications Lmking Corporate Strategy to Capital Stmcture: Diversification Strategy, Type and Source of Financing. Strategic Management Journal, 79,601-610.
Lang, L. H., & Stulz, R. M. (1994). Tobm's q, Corporate Diversification, and Firm Performance. Journal of Political Economy, 102, 1248-1280.
Lei, D., Hitt, M. A., & Goldhar, J. D. (1994). Generic Strategies, Complementarities and Organization Design in Global Manufacturmg Firms. Strategic Management Departmental Workuig Paper, Texas A&M University.
136
Levintiial, D. A. (1995). Strategic Management and tiie Exploration of Diversity. hiC. A. Montgomery, (Ed.) Resource-Based and Evolutionary Theories of the Firm. Norwell, MA: Kluwer Academic Publishers.
Levit, T. (1975). Dmosaurs among the Bears and Bulls. Harvard Business Review, 53 (I), 41-53.
Littell, R. C , Milliken, G. A., Sfroup, W. W., & Wolfinger, R. D. (1996). SAS System for Mixed Models. Gary, NC: SAS Institute, hic.
Lubatkin, M. (1983). Mergers and the Performance of Acquiring Firms. Academy of Management Review, 8, 218-225.
Lubatkfri, M., & Chetterjee, S. (1994). Extendmg Modem Portfolio Theory mto tiie Domain of Corporate Strategy: Does It Apply? Academy of Management Journal, 37, 109-136.
Lubatkin, M. H., Merchant, M. H., & Srinivasan, N. (1993). Constmct Validity of Some Unweighted Product Count Diversification Measures. Strategic Management Journal, 14, 433-449.
Mann, P. (1999, October). A second chance. Manufacturing Systems, A14-A15.
Markides, C. C. (1992). Consequences of Corporate Refocusmg: Ex-ante Evidence. Academy of Management Journal, 35, 398-412.
Markides, C. C. (1997). To Diversify or not to Diversify. Harvard Business Review, 75 (6), 93-99.
Markides, C. C , & Wilhamson, P. J. (1994). Related Diversification, Core Competencies, and Corporate Performance. Strategic Management Journal, 15 (special issue), 149-165.
Markides, C. C , & Williamson, P. J. (1996). Corporate Diversification and Orgaruzational Stiiicture: A Resource-Based View. Academy of Management Journal, 39, 340-367.
Mason, E. S. (1939). Price and Production Policies of Large Scale Enterprises. American Economic Review, 29, 61-74.
McGahan, A. M. (1999). Compethion, Strategy, and Busmess Performance. California Management Review, 41 (3), 74-101.
137
McGahan, A.M. (1998). The Performance of US Corporations: 1981-1994. Workmg paper. Harvard Busmess School.
McGahan, A. M., & Porter, M. E. (1997a). How Much Does hidustty Matter Really? Strategic Management Journal, 75 (special issue), 15-30.
McGahan, A. M., & Porter, M. E. (1997b). What Do We Know About Variance in Accountmg Profitability? Working paper, Harvard Busmess School.
Michel, A., & Shaked, I. (1984). Does Business Diversification Affect Performance? Financial Management, 13 (4), 18-25.
Miller, C. C , & Cardmal, L. B. (1994). Sfrategic plannmg and firm performance: A synthesis of more than two decades of research. Academy of Management Journal, 37, 1649-1665.
Montgomery, C. A. (1982). The Measurement of Firm Diversification: Some New Empirical Evidence. Academy of Management Journal, 25, 299-307.
Montgomety, C. A. (1994). Corporate diversification. Journal of Economic Perspectives, 8, 163-178.
Myers, R. (1996, October). Metric Wars. CFO, 41-50.
NAICS Association, October 20, 1999. http://www.naics.coni/info.htm#Principles
Nemetz, P., & Fry, L. (1988). Flexible Manufacturing Organizations: Implications for
Strategy Formulation and Organization Design. Academy of Management Review, 13 (14), 627-638.
Nourse, E. G., & Drury, H. B. (1938). Industrial Price Policies and Economic Progress. Washington, DC: The Brookings Institution.
Palepu, K. (1985). Diversification Strategy, Profit, Performance, and tiie Entropy Measure. Strategic Management Journal, 6, 239-255.
Palich, L. E., Cardmal, L. B., & Miller, C. C. (2000). Curvilmerarity m tiie Diversification-Performance Linkage: An Examination of Over Three Decades of Research. Strategic Management Journal, 21, 155-174.
Penrose, E.T. (1959). The Theory of the Growth of the Firm. New York: Wiley.
138
Perrow, C. (1961). The Analysis of Goals m Complex Organizations. American Sociological Review, 26, 854-866.
Perrow, C. (1970). Organizational Analysis: A Sociological View. Behnont, CA: Wadsworth.
Peteraf, M. A. (1993). The Comerstones of Competitive Advantage: A Resource-Based View. Strategic Management Journal, 14,119-191.
Peters, T. J., & Waterman, R. H. (1982). In Search of Excellence. New York: Free Press.
Pfeffer, J. (1981). Power in Organizations. Marshfield, MA: Pitman.
Pfeffer, J., & Salancik, G. R. (1974). Organizational Decision Making as a Political Process: The Case of a University Budget. Administrative Science Quarterly, 19, 135-151.
Pitts, R. A., & Hopkms, H. D. (1982). Ffrm Diversity: Conceptualization and Measurement. Academy of Management Review, 7, 620-629.
Porter, M.E. (1980). Competitive Strategy. New York: The Free Press.
Porter, M.E. (1985). Competitive Advantage. New York: The Free Press.
Porter, M. E. (1987). From Competitive Advantage to Corporate Strategy. Harvard Business Review, 65 (3), 43-59.
Porter, M.E. (1996). What is Strategy? Harvard Business Review, 7^(6), 61-78.
Porter, M. E. (2001). Sfrategy and the Intemet. Harvard Business Review, 79 (2), 63-78.
Powell, T. C. (1996). How much does industry matter? An ahemative empirical test. Strategic Management Journal, 77, 323-334.
Powell, W. W., Koput, K. W., & Smith-Doerr, L. (1996). hiterorganizational Collaboration and the Locus of hmovation: Networks of Leaming m Biotechnology. Administrative Science Quarterly, 41, 116-145.
Prahalad, C. K., & Hamel, G. (1990). The Core Competence of the Corporation. Harvard Business Review, 68 (3), 79-91.
139
Qumn, R. E., & Rohrbaugh, J. (1983). A Spatial Model of Effectiveness Criteria: Toward a Competing Values Approach to Organizational Analysis. Management of Science, 29, 363-377.
Quote.com (2001). http://www.quote.com/quoteconi/stocks/quotes.asp?symbols=TYC
Raghunatiian, S. P. (1995). A Refinement of tiie Entropy Measure of Firm Diversification: Toward Definitional and Computational Accuracy. Academy of Management, 21, 989-1002.
Ramanujan, V., & Varadarajan, P. (1989). Research on Corporate Diversification: A Synthesis. Strategic Management Journal, 10, 523-551.
Reed, R., & Luf&nan, G. A. (1986). Diversification: The Growing Confusion. Strategic Management Journal, 7, 29-35.
Reichers, A. E., & Schneider, B. (1990). Clunate and Culture: An Evolution of Constmcts. In B. Schneider (Ed.) Organizational Climate/Culture, 5-39. San Francisco: Jossey-Bass.
Riordan, J. (1986). Six Blind Men and An Elephant. An Illustrated Treasury of Fairy and Folk Tales, 30-33. Twickenham: The Hamlyn Publishmg Group, Ltd.
Roquebert, J. A., Phillips, R. L., & Westfall, P. A. (1996). Markets vs. Management: What Drives Profitability? Strategic Management Journal, 17, 653-664.
Rumelt, R.R. (1974). Strategy, Structure, and Economic Performance. Boston: Harvard Busmess School Press.
Rumelt, R.R. (1982). Diversification Sfrategy and Profitability. Strategic Management Journal, 3, 359-369.
Rumelt, R.R. (1986). Strategy, Structure, and Economic Performance. Boston: Harvard Business School Press.
Rumelt, R.R. (1991). How Much Does hidustry Matter? Strategic Management Journal, 12, 167-185.
Rumelt, R. R., Schendel, D. E., & Teece, D. J. (1994). Fundamental Issues in Strategy: A Research Agenda. Boston: Harvard Busmess School Press.
SAS Software, Release 8.1. (2000). Gary, NC: SAS histittite hic.
140
Schmalensee, R. (1985). Do Markets Differ Much? American Economic Review, 75, 341-351.
Scott, J. T., & Pascoe, G. (1986). Beyond Ffrm and hidustry Effects on Profitability m Imperfect Markets. Review of Economics and Statistics, 68, 284-292.
Scott, R. W. (1995). Organizations: Rational, Natural, and Open Systems (3rd Ed.). Englewood Cliffs: Prentice Hall.
Searle, S. R., Casella, G., & McCuUoch, C. E. (1991). Variance Components. New York: John Wiley & Sons, Inc.
Seth, A. (1990). Value Creation ui Acquisitions: A Re-exammation of Performance Issues. Strategic Management Journal, 11, 99-115.
Shepherd, W. G. (1986). On tiie Core Concepts of hidustrial Economics. hiH. W. de Jong & E. G. Shepard (Eds.) Mainstreams in Industrial Organization. Dordrecht: Kluwer Publications.
Shleifer, A. & Vishny, R. W. (1994). Takeovers m the 1960s and tiie 1980s: Evidence and Implications. In R. P. Rumelt, D. E. Schendel, & D. J. Teece (Eds.) Fundamental Issues in Strategy, 403-418. Boston: Harvard Business School Press.
Simmods, P. (1990). The Combined Diversification Breadth and Mode Dimensions and the Performance of Large Diversified Firms. Strategic Management Journal, 11, 399-410.
Smith, C. G., & Cooper, A. C. (1988). Established Companies Diversifying mto Young Indusfries: A Comparison of Firms with Different Levels of Performance. Strategic Management Journal, 9, 111-121.
Song, X. M., & Parry, M. E. (1999). Challenges ofManaging the Development of Breakthrough Products in Japan. Journal of Operations Management, 17(6), 665-688.
St. John, C. H., & Harrison, J. S. (1999). Manufacturing-Based Relatedness, Synergy, and Coordmation. Strategic Management Journal, 20, 129-145.
Stalk, G. (1988). Time — the Next Source of Competitive Advantage. Harvard Business Review, 66 (4), 41-51.
141
Staw, B. M. (1981). The Escalation of Commitment to a Course of Action. Academy of Management Review, 6, 557-587.
Stetz, P. E., & Phillips, R. L. (in press). Does Corporate Membership Improve or Impede Busmess Unit Performance? Southern Management Association, 2000 Proceedings.
Suchman, M. C. (1995). Managing Legitimacy: Strategic and histitutional Approaches. Academy of Management Review, 20, 571-610.
Talhnan, S., & Li, J. (1996). Effects of Intemational Diversity and Product Diversity on the Performance of Muhmational Ffrms. Academy of Management Journal, 39, 179-196.
Tanzer, A. (1999, June 14). Smgapore Flmg. Forfte.?, 755 (12), 148-150.
Teece, D. J. (1980). Economies of Scope and the Scope of the Enterprise. Journal of Economic Behavior and Organization, 1, 223-245.
Teece, D. J. (1981). Intemal organization and economic performance. Journal of Industrial Economics, 30, 173-199.
Tfrole, J. (1988). The Theory of Industrial Organization. Cambridge: The MIT Press.
Tosi, H., & Gomez-Mejia, L. (1989). The Decoupling of CEO Pay and Performance: An Agency Theoty Perspective. Administrative Science Quarterly, 34, 169-189.
Varadarajan, P., & Ramanujam, V. (1987). Diversification and Performance: A Reexamination Using a New Two-Dimensional Conceptualization of Diversity in Firms. Academy of Management Journal, 30, 380-393.
Verespej, M. A. (2000). Unheralded Success. Industry Week, 249 (18), 46-52.
Wemerfelt, B. (1984). A Resource-Based View of the Firm. Strategic Management Journal, 5, Ill-ISO.
Wemerfelt, B., & Montgomery, C. A. (1988). Tobm's q and the hnportance of Focus in Firm Performance. American Economic Review, 78, 246-250.
Weston, J. F., & Mansmghka, S. K. (1971). Tests of the Efficiency Performance of Conglomerate Firms. Journal of Finance, 26, 919-936.
142
Williams, J. R., Paez, B. L., & Sanders, L. (1988). Conglomerates Revisited. Strategic Management Journal, 9, 403-414.
WiUiamson, O. E. (1975). Markets and Hierarchies: Analysis and Antitrust Implications. Boston: Harvard Business School Press.
Williamson, O. E. (1979). Transaction-Cost Economics: The Govemance of Contractual Relations. Journal of Law and Economics, 22, 233-261.
Williamson, O. E. (1985). The Economic Institutions of Capitalism. New York: Free Press.
Williamson, O. E. (1995). Transaction Cost Economics and Organizational Theory. Organizational Theory. New York: Oxford Uruversity Press.
Wrigley, L. (1969). Divisional Autonomy and Diversification. Boston: Division of Research, Graduate School of Business Administration, Harvard University.
Zhang, H. (1995). Wealth Effects of US Bank Takeovers. Applied Financial Economics, 5, 329-336.
143
APPENDDC
HIERARCHIAL CLUSTER ANALYSIS OF
TOTAL DIVERSIFICATION SCORES:
DENDROGRAM
144
Figure A.l. Hierarchical Cluster Analysis of Total Diversification Scores: Dendrogram Overview.
145
Table A.l. Hierarchical Cluster Analysis of Total Diversification Scores: Dendrogram.
Dendrogram using Average Linkage (Within Group) : Sc[uared Euclidean Distance
Rescaled Distance Cluster Combine
5 10 15 C A S E Label
Case 23
Case 24
Case 25
Case 26
Case 27
Case 29
Case 3 0
Case 28
Case 21
Case 22
Case 19
Case 20
Case 18
Case 13
Case 14
Case 15
Case IS
Case 17
Case 31
Case 32
Case 33
Case 34
Case 35
Case 36
Case 3 7
Case 3 8
Case 39
Case 4 0
Case 41
Case 45
Case 46
Case 43
Case 44
Case 42
Case 11
Case 12
Case 3
Case 4
Case 6
Case 7
Case 5
Case 8
0
23 - 1 24 - 1 25 - 1 26 - 1 27 - 1 29 - 1 30 1 28 - 1 21 - 1 22 - 1 19 - 1 20 - 1 18 - 1 13 - 1 14 - 1 15 - 1 16 - 1 17 - 1 31 - 1 32 - 1 33 - 1 34 - 1 35 - 1 3G - 1 37 - 1 38 - 1 39 1 40 - 1 41 - 1 45 - 1 46 - 1 43 1 44 - 1 42 - 1 11 - 1 12 - 1 3 - 1
4 - 1 6 - 1 7 - 1 5 - 1 8 - 1
20 -- + -
25 -- +
146
Table A.l. Contmued.
C A S E 0
Label Num +
10 15 20 25
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
9
10
2
1
82
83
84
81
85
86
89
90
87
88
91
78
79
77
75
76
73
74
72
64
65
70
71
68
69
67
66
80
48
49
47
50
51
52
59
60
61
62
63
56
57
9 - 1 10 -1
2 - 1 1 - 1
82 - 1 83 - 1 84 - 1 81 - 1 85 - 1 86 - 1 89 - 1 90 - 1 87 - 1 88 - 1 91 - 1 78 - 1 79 - 1 77 - 1 75 - 1 76 - 1 73 - 1 74 - 1 72 - 1 64 - 1 65 - 1 70 - 1 71 - 1 68 - 1 69 - 1 67 - 1 66 - 1 80 - 1 48 - 1 49 -1 + 4 7 - 1 1 5 0 - 1 1 51 1 1 52 1 1 59 1 1 60 1 1 6 1 - 1 1 6 2 - 1 1 6 3 - 1 1 5 6 - 1 1 5 7 - 1 1
147
Table A.l. Contmued.
C A S E 0
L a b e l Num +
10 15 20 25
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
54
55
53
58
111
112
113
114
109
110
104
105
106
107
108
92
93
95
96
94
97
98
101
102
103
99
100
115
116
121
122
118
119
117
120
141
142
143
140
144
145
146
147
156
157
5 4 - 1 1 5 5 - 1 1 5 3 - 1 1 5 8 - 1 1
111 - 1 1 112 - 1 1 1 1 3 - 1 1 114 - 1 1 109 - 1 1 110 -1 1 104 - 1 1 105 - 1 1 106 - 1 1 107 - 1 1 108 - 1 1 9 2 - 1 1 9 3 - 1 1 9 5 - 1 1 9 6 - 1 1 9 4 - 1 1 9 7 - 1 1 9 8 - 1 1
101 -1 1 102 - 1 1 103 - 1 1 9 9 - 1 1
100 -J 1
115 -1 1 116 - 1 1 121 - 1 1 122 - 1 + 118 - 1 1 119 - 1 1 117 - 1 1
120 - 1 1 141 - 1 1 142 - 1 1 143 - 1 1 140 - 1 1 144 - 1 1 145 -+-+ 1 146 1 1 1 147 1 1 1 156 - 1 1 1 157 - 1 1 1
148
Table A.l. Contmued.
C A S E
Label
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
148
149
150
154
155
151
152
153
136
137
135
138
139
133
134
132
131
129
130
127
128
126
124
125
123
224
225
226
217
218
221
222
223
219
220
233
234
235
229
230
231
228
232
227
236
0
Num +
148
149
150
154
155
151
152
153
136
137
135
138
139
133
134
132
131
129
130
127
128
126
124
125
123
224
225
226
217
218
221
222
223
219
220
233
234
235
229
230
231
228
232
227
236
10 15 20 25
149
Table A.l. Contmued.
C A S E 0
L a b e l Num +
10 15 20 25
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
237
238
239
246
247
242
243
241
240
244
245
159
160
161
158 180
181
173
174
178
179
176
177
175
164
165
163
166
167
162
171
172
168
169
170
182
183
184
185
186
187
188
189
190
191
237 1 1
238 -1 1
239 - 1 1
246 - 1 1
247 - 1 1
242 - 1 {
243 - 1 1
241 - 1 1
240 - 1 1
244 - 1 1
245 - 1 1
159 - 1 1
160 - 1 1
161 - 1 1
158 - 1 1 180 -+-+
181 - 1
173 - 1
174 - 1
178 - 1
179 - 1
176 - 1
177 - 1
175 - 1
164 - 1
165 - 1
163 - 1
lee - 1 167 - 1
162 - 1
171 - 1
172 - 1
168 - 1
169 - 1
170 - 1
182 - 1
183 - 1
184 - 1
185 - 1
186 - 1
187 - 1
188 - 1
189 - 1
190 - 1
191 - 1
150
Table A.l. Contmued.
C A S E 0
L a b e l Num +
10 15 20 25
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
192
193
194
195
196
202
203
204
201
198
199
200
197
213
214
215
212
216
206
207
205
208
209
210
211
192 - 1 193 - 1 194 - 1 195 - 1 196 - 1 202 - 1 203 - 1 204 - 1 201 - 1 198 - 1 199 - 1 200 - 1 197 - 1 213 - i
214 - i
215 - 1 212 - 1 216 - 1 206 - 1 207 - 1 205 - 1 208 - 1 209 - 1 210 - 1 211 -J
Case 348 348 Case 349 349 Case 347 347 Case 350 350 Case 351 351
- + - +
-+ +-
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
345
346
332
333
334
335
339
340
341
337
338
336
342
343
344
345
346
332
333
334
335
339
340
341
337
338
336
342
343
344
151
Table A.l. Contmued. C A S E
Label
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
248
249
250
251
252
253
255
256
254
257
258
259
260
273
274
270
271
272
269
263
264
261
262
267
268
265 266
286
287
288
289
291
292
290
293
294
275
276
277
278
279
283
284
282
285
0 5
248 -1
249 - 1
250 - 1
251 1
252 1
253 - 1
255 1
256 1
254 1
257 1
258 - 1
259 - 1
260 1
273 - 1
274 1
270 - 1
271 - 1
272 - 1
269 - 1
263 - 1
264 - 1
261 - 1
262 - 1
267 - 1
268 - 1
265 - 1 266 - + -t
286 - 1
287 - 1
288 - 1
289 - 1
291 - 1
292 - 1
290 - 1
293 - 1
294 - 1
275 - 1
276 - 1
277 - 1
278 - 1
279 - 1
283 - 1
284 - 1
282 - 1
285 - 1
10
1
+
15 20 25
152
Table A.l. Continued.
C A S E 0
Label Num +
10 15 20 25
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case
Case Case
Case
Case
Case
Case
Case
Case
281
280
301
302
303
300
299
304
298
295
296
297
315
316
317
318
309 310
312
313
311
314
305
306
307
308
320
321
319
323
324
322 325
330
331
328
329
327
326
281 - 1 280 - 1 301 - 1 302 - 1 303 - 1 300 - 1 299 - 1 304 - 1 298 - 1 295 - 1 296 - 1 297 -J
315 -1 316 - 1 317 - 1 318 - 1 309 - 1 310 -+-+
312 -1 1 313 - 1 1 311 -1 1 314 - 1 1 305 - 1 1 306 - 1 1 307 -1 +--
308 - 1 1 320 - 1 1 321 - 1 1 319 - 1 1 323 - 1 1 324 - 1 1 322 - 1 1 325 -+-+
330 1 331 - 1 328 - 1 329 - 1 327 - 1 326 -J
+
153
Table A.2. Key for Correspondence Between Case Number and TDS Scores.
Case Number
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
TDS Case Score Number
1.500538 1.744503 1.790489 1.792616 1.801067 1.802756 1.803203 1.804959 1.807112 1.811580 1.818110 1.821255 1.835640 1.836905 1.845365 1.847462 1.850269 1.854305 1.857220 1.857642 1.862013 1.864371 1.869342 1.869344 1.871280 1.872755 1.874054 1.877106 1.880194 1.881915 1.891985 1.892971 1.894728 1.895805 1.903376
36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70
TDS Score ]
1.91019 1.911486 1.914068 1.915373 1.922141 1.924653 1.927296 1.928604 1.929738 1.931779 1.933195 1.941351 1.942981 1.943405 1.946181 1.946863 1.949966 1.957013 1.959192 1.960163 1.961553 1.961997 1.964527 1.970600 1.970842 1.971500 1.972256 1.975013 1.978426 1.978687 1.981837 1.982601 1.983186 1.983436 1.984382
Case Mumber
71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105
TDS Case Score Number
1.984796 1.986012 1.986838 1.987287 1.989315 1.989721 1.994155 1.994366 1.994521 2.014632 2.027541 2.031218 2.031564 2.032365 2.03762 2.044821 2.050931 2.054423 2.060299 2.060906 2.079048 2.095632 2.098074 2.100547 2.103892 2.104209 2.112584 2.122863 2.144163 2.152405 2.161076 2.164171 2.167838 2.186416 2.188031
106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140
TDS Case Score Number
2.19267 2.193278 2.195685 2.209322 2.21463 2.228056 2.228622 2.238366 2.245252 2.297485 2.302281 2.313887 2.316138 2.318346 2.320645 2.327692 2.328538 2.348844 2.360536 2.371768 2.394024 2.397557 2.398066 2.411085 2.420185 2.427919 2.433454 2.434894 2.435899 2.447418 2.449386 2.449863 2.45607 2.459324 2.473852
141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175
TDS Score
2.477565 2.477955 2.479360 2.483705 2.491947 2.493284 2.501127 2.509867 2.513595 2.517490 2.523557 2.523642 2.527907 2.531303 2.534414 2.545547 2.556698 2.580846 2.588965 2.589617 2.594689 2.633639 2.639158 2.640036 2.640319 2.643702 2.645570 2.650805 2.654760 2.659537 2.666654 2.667905 2.675555 2.678796 2.684382
154
Table A.2. Contmued.
Case Number
176 111 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210
TDS Case Score Number
2.685253 2.686088 2.690664 2.691247 2.698968 2.701139 2.718597 2.718633 2.722367 2.728560 2.730678 2.735782 2.746959 2.746990 2.750704 2.750874 2.751057 2.753356 2.755170 2.762159 2.772867 2.783870 2.791373 2.792400 2.794049 2.807787 2.814188 2.815381 2.817526 2.830107 2.835114 2.836371 2.842369 2.844973 2.849961
211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245
TDS Case Score Number
2.857925 2.867562 2.873901 2.875006 2.877203 2.883983 2.898682 2.899586 2.909477 2.912509 2.920635 2.921504 2.922789 2.933325 2.934224 2.939179 2.950736 2.962183 2.964227 2.964913 2.966075 2.969638 2.97456 2.975103 2.981860 3.004177 3.005954 3.008629 3.026031 3.039025 3.046026 3.052337 3.053311 3.066536 3.083676
246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280
TDS Case Score Number
3.132934 3.133171 3.201435 3.205682 3.213136 3.217405 3.248208 3.270948 3.292315 3.305593 3.306437 3.326473 3.334882 3.343916 3.352616 3.376840 3.379776 3.384018 3.384840 3.392671 3.398275 3.404052 3.408213 3.420741 3.427532 3.428319 3.432465 3.446497 3.448284 3.512188 3.516312 3.525604 3.529432 3.533324 3.551136
281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315
TDS Case Score Number
3.567592 3.573932 3.576315 3.578502 3.581435 3.598611 3.600870 3.608338 3.609096 3.612196 3.613866 3.614427 3.635683 3.654609 3.703701 3.713009 3.759137 3.797696 3.820143 3.837948 3.842911 3.843891 3.845021 3.864745 3.929145 3.938131 3.959864 3.999215 4.034479 4.040847 4.055075 4.069610 4.073005 4.102862 4.155257
316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351
TDS Score
4.162678 4.205500 4.243875 4.397034 4.418254 4.428218 4.464523 4.470048 4.473205 4.512696 4.559345 4.616848 4.631610 4.641641 4.700073 4.706761 4.811813 4.833366 4.870424 4.940522 5.009097 5.047187 5.059917 5.118065 5.122850 5.132268 5.216980 5.227096 5.340851 5.623644 5.776727 6.844279 6.977297 7.060179 7.355749 7.939686
155
top related