© copyright 2001, phil e. stetz

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

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

http://Quote.com

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



Silent

Moderately high to high

level of diversification Um-elated

Diversification





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


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


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


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)





None

Unexplained variance: 32.2% (avg.)


41


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


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


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


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


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


43


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


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


+/- 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%


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

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


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


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



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


-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 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


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



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

http://www.cbsnews.com/now/story/0,1597,277076-412,00.shtml

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

http://www.naics.coni/info.htm%23Principles

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

http://Quote.com

http://www.quote.com/quoteconi/stocks/quotes.asp?symbols=TYC

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


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


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


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


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


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


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


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


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


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


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

© copyright 2001, phil e. stetz

Documents