Corporate Elites and Corporate Strategy

7/29/2019 Corporate Elites and Corporate Strategy

1/18

Strategic Management JournalStrat. Mgmt. J., 25: 507 524 (2004)

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/smj.393

CORPORATE ELITES AND CORPORATE STRATEGY:

HOW DEMOGRAPHIC PREFERENCES AND

STRUCTURAL POSITION SHAPE THE SCOPE OF

THE FIRM

MICHAEL JENSEN1* and EDWARD J. ZAJAC2

1 University of Michigan Business School, Ann Arbor, Michigan, U.S.A.2 Kellogg School of Management, Northwestern University, Evanston, Illinois, U.S.A.

This study combines elements of the upper echelons and agency perspectives to resolve some ofthe ambiguity surrounding how corporate elites affect corporate strategy. We propose and test thenotion that while differences in individual characteristics of corporate elites may imply different

preferences for particular corporate strategies such as diversification and acquisitions, thesebasic preferences, when situated in different agency contexts (e.g., CEO, outsider director, non-CEO top management team member), generate very different strategic outcomes. Our detailedempirical findings, based on extensive longitudinal governance and corporate strategy data fromlarge U.S. corporations, also highlight the pitfalls of using aggregate units of analysis (e.g.,board of directors or top management team) when studying the influence of corporate elites oncorporate strategy. Copyright 2004 John Wiley & Sons, Ltd.

Over the last several decades strategy researchershave devoted considerable attention to the question

of how corporate elites (i.e., corporate executivesand directors) affect corporate strategy. Much ofthis research has been grounded in one of twodominant theoretical perspectives: upper echelonsand agency theory (Cannella and Monroe, 1997).While both perspectives agree that corporate elitespreferences and dispositions influence corporatestrategy, the upper-echelons perspective tends toemphasize the role of demography-based pref-erences and dispositions (Hambrick and Mason,1984), whereas the agency theory perspectivetends to emphasize the role of position-based pref-

erences and dispositions (Fama and Jensen, 1983).Specifically, upper-echelons theorists suggest thatthere is a close association between corporate

Key words: corporate elites; agency theory; upper-ech-elons theory; diversification; acquisitions*Correspondence to: Michael Jensen, University of MichiganBusiness School, 701 Tappan Street, Ann Arbor, MI 48109-1234,U.S.A. E-mail: [email protected]

elites demographic characteristics, such as age,

education, and functional background experiences,

and their cognitive bases and values, which in turndetermines their strategy preferences and disposi-

tions (Wiersema and Bantel, 1992). Agency theo-rists focus on structural analyses of corporate gov-

ernance arrangements and how boards of directors

align the interests of executives and shareholders

(Fama, 1980; Jensen and Meckling, 1976). Indeed,

agency theorists are less interested in the demo-

graphic characteristics of corporate elites, empha-

sizing instead the governance positions that corpo-rate elites occupy on boards, e.g., whether they

are CEOs, executive (inside) directors, or non-

executive (outside) directors (Jensen, 1986).While each perspective has inspired separate

streams of research on how corporate elites affect

corporate strategy (as reviewed by Finkelstein and

Hambrick, 1996; Johnson, Daily, and Ellstrand,

1996; and Cannella and Monroe, 1997), we sug-gest that a joint consideration of insights from

each stream can enhance our understanding of the

Copyright 2004 John Wiley & Sons, Ltd. Received 30 May 2001

Final revision received 13 October 2003


2/18

508 M. Jensen and E. J. Zajac

relationship between corporate elites and corpo-rate strategy. For example, while upper-echelons

research on elites demographic characteristicssheds important light on their effects on corpo-rate strategy, it has tended to neglect the gover-nance context in which corporate elites are situ-ated. Research building on the agency perspectiveis highly sensitive to the different governance posi-tions corporate elites occupy as executives andboard members, but neglects the demographicallybased preferences and dispositions that corporateelites may have (independent of their position onthe board).

Rather than arguing the relative merits of one

perspective vs. the other, we believe there is con-siderable value in accepting the relevance of bothdemography and position. In this study, for exam-ple, we will argue that the demographically basedpreferences and dispositions that corporate elitesbring to their governance position influence theirdecisions, but that these preferences and disposi-tions will also be affected by the role expectationsassociated with the corporate governance positionsthey occupy. By considering both demographiccharacteristics and governance positions, we hopeto explain the basis for the mixed empirical resultsthat have plagued research studies grounded solelyin either the upper-echelons or the agency per-spective (Finkelstein and Hambrick, 1996; Daltonet al., 1998). Specifically, in this study we examinehow demographically based preferences and dispo-sitions may have differential effects on corporatestrategy, depending on the governance position inwhich these preferences and dispositions are situ-ated.

A related issue involves the choice of unit ofanalysis. Agency theorists focus on the poten-tial conflicts of interests between CEOs, execu-tive, and non-executive directors typically leads

to discussions of how their orientations differ asa consequence of the different role they occupyin the agency relationship. Upper-echelons the-orists generally de-emphasize governance differ-ences and combine the CEO and other executivesinto the top management team unit of analysis(e.g., Hambrick and Mason, 1984), or even suggestcombining the top management team with the non-executive directors into a supra-TMT (Hambrickand Finkelstein, 1996). While both approacheshave contributed considerably to research on cor-porate elites, our study suggests that some ofthe empirical ambiguities associated with both

perspectives may in fact be caused by their choiceof unit of analysis.

Studies that have used only one or the otheraggregate unit of analysis and drawn conclusionsabout corporate elites have implicitly assumed thatcorporate elites effects on strategy and behaviorsdepend only on their board role or their demo-graphic characteristics. This study seeks to showthat corporate elites who are demographically sim-ilar but occupy different roles are not necessar-ily associated with similar strategic choices, norare elites who are demographically different butoccupy the same role.

We examine these issues by analyzing the effects

of elites demographic preferences and governancepositions on corporate diversification level andacquisition activity. While most research on thescope of the firm has focused on firms eco-nomic characteristics, such as performance andprior diversification levels (e.g., Markides, 1995),some research has focused on the importance ofcorporate elites and their demographic characteris-tics or their governance positions (e.g., Finkelstein,1992; Lane, Cannella, and Lubatkin, 1998). Byproviding a comprehensive framework for analyz-ing the role of corporate elites in diversificationand acquisitions, we hope also to contribute toa greater understanding of the important strategycontent question of what predicts firms decisionsin these areas.

We test our predictions using extensive longitu-dinal data on diversification level and acquisitionactivities in a stratified random sample of 200 For-tune 500 firms from 1985 to 1995. We find thatdemographic characteristics do predict corporatestrategy, but that both the strength and the directionof these predictions vary significantly, dependingon the governance position in which the demo-graphic characteristics are observed.

THE ROLE OF DEMOGRAPHYAND GOVERNANCE POSITIONIN DIVERSIFICATION ANDACQUISITIONS

To study the potential importance of combining afocus on demographically based and positionallybased preferences and dispositions, we examinehow the effects of corporate elites functionalbackground experiences differ depending ontheir governance positions. Our emphasis on

Copyright 2004 John Wiley & Sons, Ltd. Strat. Mgmt. J., 25: 507 524 (2004)


3/18

Corporate Elites and Corporate Strategy 509

functional background is based on the factthat it is the most widely cited demographic

characteristic thought to affect corporate strategy.Many prior upper echelon studies have focusedon corporate elites functional backgrounds andtheir relationships with either diversification oracquisitions (see Finkelstein and Hambrick, 1996).While diversification and acquisition decisions,strictly speaking, are orthogonal decisions (sincediversification could theoretically occur with noacquisition activity), these corporate strategydecisions are in fact often intertwined. Specifically,acquisitions are often viewed as an importantmechanism by which a preferred diversification

level is achieved (Salter and Weinhold, 1978).For example, Markides (1995: 6166) reportsthat Fortune 500 firms in the 1980s adjustedtheir diversification level through acquisitionsand divestitures rather than internal developmentsand divestitures. Since the theoretical argumentsrelating functional background experiences anddiversification and acquisitions are closely related,focusing on both provides a stronger test of thegeneral argument that it is important to considerdemography and governance position jointly.

Dearborn and Simon (1958) first examined the

relationship between functional background expe-riences and strategic decisions. They argued thatcorporate elites experiences bias their attentionand proposed solutions to complex business sit-uations, and showed that executives in an exper-imental setting gravitated towards interpretationsof a complex business situation that reflected theirown functional backgrounds. Finkelstein and Ham-brick (1996: 93) argue more specifically that acorrespondence between functional experiences,preferences and dispositions, and strategic choicesmay occur through self-selection and socializa-

tion: individuals may chose functional areas thatfit their cognitive models and values, but alsoover time become socialized and inculcated withthe areas dominant mode of thinking and acting.In either case, executives with similar functionalbackground experiences tend to converge on simi-lar understandings of business problems and appre-ciate similar solutions to these problems.

While recent studies in general have only foundweak empirical relationships between executivesfunctional background experiences and their per-ceptions and beliefs (Walsh, 1988; Waller, Huber,and Glick, 1995; Beyer et al., 1997; Chattopadhyay

et al., 1999), there is also strong empirical sup-port that corporate elites functional background

experiences predict diversification level and acqui-sition activities (Song, 1982; Finkelstein, 1992;Michel and Hambrick, 1992). Hayes and Aber-nathy (1980) and Fligstein (1990) both argue thatcorporate elites functional background experi-ences are reflected in their firms diversificationlevel and acquisition activities. They argue specif-ically that individuals with dominant functionalexperiences in finance (and accounting and law)typically perceive firms as a collection of return-generating assets that need not be associated with asingle line of business. The firm, from this perspec-

tive, can easily be viewed as a portfolio of mul-tiple businesses, and firms led by corporate eliteswith dominant functional background experiencesin finance are therefore more likely to emphasizegrowth through diversification and acquisitions.

In fact, several studies document the relation-ships between functional background experiencesin finance and firms diversification level andacquisition activity. Song (1982) found that financeCEOs, who are thought to typically view the firmas a bundle of financial assets, tend to preferto diversify through acquisitions, whereas pro-duction CEOs, who are thought to emphasize

more organic growth, tend to prefer diversificationthrough internal development. Palmer and Barber(2001) report similarly that finance CEOs weremore likely to complete diversifying acquisitionsbetween 1963 and 1968 than non-finance CEOs(see also Haunschild, Henderson, and Davis-Blake,1999). Finally, Finkelstein (1992) found that firmsdominated by finance executives were likely to bemore diversified and do more expensive acquisi-tions, and Michel and Hambrick (1992) found thatfirms with more executives with functional back-ground experiences in production (as opposed to

finance) diversified less (cf. Fligstein, 1987). Thesefindings suggest a close association between thefunctional background of a firms senior execu-tives and a firms subsequent diversification andacquisition strategies.1

1 Our study is consistent with prior research on executive demog-raphy in that our hypotheses to follow refer to central tendencies,based on ceteris paribus conditions. Thus, we are not suggest-ing that a production or marketing CEO would never prefer anacquisition vs. internal development; rather, we are saying thatwhen faced with the same logics for synergies (e.g., presumedeconomies of scale or scope), a finance CEO will on average pre-fer an acquisition and diversification than would a production ormarketing CEOfor the reasons noted above.



4/18


While most studies of corporate elites andcorporate strategy have tended to focus either

on the CEO or the executive directors, we fol-low Finkelstein and Hambricks (1996) sugges-tion to extend the demographic arguments toinclude non-executive directors. This seems par-ticularly appropriate when studying corporate-levelstrategies, since decisions such as acquisition anddiversification will require non-executive direc-tor involvement, review, and/or approval. In dis-cussing guidelines for board responsibilities, theBusiness Roundtable says that boards review and. . . approve the financial objectives, major strate-gies, and plans of the corporation, and the Ameri-

can Law Institute says boards review and approvecorporate plans and actions that the board and prin-cipal senior executives consider major.

Prior qualitative research also suggests that non-executive directors influence diversification andacquisitions in at least two ways: direct influ-ence on developing and approving diversifica-tion and acquisition strategy and indirect influ-ence through their function as sounding boards formanagement (Lorsch and MacIver, 1989; Demband Neubauer, 1992; McNulty and Pettigrew,1996). Finally, including non-executive directorsin our theoretical discussion and empirical analysesenables us to present more fine-grained and com-prehensive analyses of the importance of the elitesdemographic characteristics on corporate strategy(e.g., by examining differences across alternativedefinitions of corporate elites).

The prior discussion suggests the following twocore hypotheses:

Hypothesis 1: Firms with more finance corpo-

rate elites are more likely to engage in high

levels of diversification.

Hypothesis 2:Firms with more finance corporateelites are more likely to engage in greater acqui-

sition activity.

Note that the previous arguments focus exclusivelyon the association between corporate elitesfunctional background experiences and corporatestrategy, and implicitly treat corporate elitesas a homogeneous group distinguished onlyby the presence/absence of certain functionalbackground experiences. However, a considerationof agency theory would suggest that thehomogeneity assumption may be problematic

and that corporate elites strategy preferencesand dispositions are determined not only by

their functional background experiences, but alsoby their governance role (Fama and Jensen,1983). Agency theorists argue specifically thatshareholders and executives may have differentinterests and that the function of governancearrangements such as the board of directors isto align their interests (e.g., Fama and Jensen,1983). While all directors are assumed to actin the interest of the shareholders, non-executivedirectors are especially important because oftheir presumed independence vis-a-vis corporateexecutives (Fama, 1980). Indeed, one of the

most commonly prescribed remedies to increasecorporate boards governance effectiveness is toincrease the number of non-executive directorsrelative to executive directors (Walsh and Seward,1990).

Distinguishing between executive and non-ex-ecutive directors is particularly important whenfocusing on decisions where top managersand shareholders interests may differ, such asdiversification and acquisition activities. Fromthis perspective, the most important determinantof corporate elites interests in diversificationand acquisitions is rooted in whether theyhold the position of executive director vs.non-executive director. Agency theorists arguethat the specialization of management andrisk-bearing functions between executives andshareholders creates different interests betweenexecutive vs. non-executive directors (Fama,1980). Shareholders can use the stock marketsto diversify their investment portfolios and hedgeagainst the risk of any particular corporationsfailure. However, executives often have muchof their wealth, particularly their human capital,tied to the corporation for whom they work

(Beatty and Zajac, 1994). Amihud and Lev (1981)and Tosi and Gomez-Mejia (1989) argue thatexecutives therefore prefer a higher level of firmdiversification than would shareholders, since itprovides them with a form of personally valueddiversification. In addition, greater diversification,as a mechanism for growing the size of thecorporation, is often associated with higherexecutive compensation, social prominence, andpublic prestige (Jensen, 1986; Gomez-Mejiaand Wiseman, 1997). Moreover, to the extentthat acquisitions are a standard approach todiversification (Salter and Weinhold, 1978: 166),



5/18


it also follows that executive directors may havea stronger preference for acquisitions than non-

executive directors. In fact, Jensen (1986) notesthat acquisitions often are the result of executivesusing firms free cash flows on wasteful attemptsto grow the firm beyond a level that is valuemaximizing for shareholders.2

Non-executive directors, however, when com-pared with executives, do not face the sameemployment risk nor do they receive the sameperks of public attention and prestige. They aretherefore more likely to act in accordance with thedirector-role expectations and protect the interestsof the shareholders they represent. These argu-

ments suggest that finance executive directors, whoface the problem of under-diversified wealth, arelikely to prefer greater corporate diversificationand more acquisitions than finance non-executivedirectors. Indeed, non-executive directors are oftenseen as helping to counteract executives prefer-ences for (and maintenance of) greater diversi-fication (Goodstein and Boeker, 1991; Westphal,1998). This discussion above suggests the follow-ing hypotheses:

Hypothesis 3: Having more finance executive

directors (as opposed to more finance non-

executive directors) will be associated with high

levels of diversification.

Hypothesis 4: Having more finance executive

directors (as opposed to more finance non-

executive directors) will be associated with

greater acquisition activity.

The agency theory argument regarding the impor-tance of positional differences can also be extendedto focus on the CEO, who occupies a uniqueposition in corporate governance. In large U.S.

corporations, the CEO is typically the most vis-ible individual, and most closely associated withthe overall performance of the firm. This sug-gests that, among the group of executive direc-tors, a CEO may have preferences that differ fromthose of other directors. Indeed, Fama and Jensen

2 While Jensen (1986) also suggests that free cash flow can besquandered through unnecessary overinvestment in R&D andother capital investments, the opportunities for growing the firm(for the benefit of its management) are greatest when diversifying(Hypothesis 3), which is more often accomplished via acquisition(Hypothesis 4). We thank an anonymous reviewer for raising thisissue.

(1983) explicitly emphasize the importance of dis-tinguishing between CEOs and the other execu-

tive directors. They argue that even though CEOsoften dominate corporate decision processes, theother executive directors serve an important func-tion as the non-executive directors primary sourceof information about both firm and CEO perfor-mance. The non-CEO executive directors are inthe precarious position of being beholden to theCEO because of their lower position in the corpo-rate hierarchy, but also having to act as directorsand monitor and control the CEO on behalf of theshareholders.

These arguments suggest that the non-CEO

executive directors interests cannot a priori beassumed to align only with the CEOs interests.Several researchers have argued that executivesrarely constitute cohesive teams. For example,Ocasio (1994) views CEO succession in the firstdecade of a CEOs tenure as an instance of circula-tion of power emphasizing shifting political coali-tions and incessant political struggles. He arguesthat conflict among corporate elites is an importantforce for change, and finds that having more execu-tive directors increases CEO succession under eco-nomic adversity. Hambrick (1994) notes similarlythat most top management teams would be betterdescribed as top management groups, given theirlow levels of behavioral integration (defined as thedegree to which top managers engage in mutualand collective interaction). The potential lack ofcohesion among executives has led some schol-ars to note that strategic leadership occurs withina complex social system of multiple leaders withmultiple agendasboth private and publicthatreflect multiple realities and the needs of multipleconstituencies (Jackson, 1992: 346).

It may be particularly useful to distinguishbetween CEOs and non-CEO corporate elites when

studying the association between functional back-ground experiences in finance and diversificationand acquisition strategies because the CEO islikely to have a particularly strong interest inboth. Specifically, individuals occupying the CEOrole are likely to gain the most from diversifica-tion, as they benefit most directly from the socialperquisites that accompany growing the scale andscope of their corporations, and they are also likelyto have the lowest level of personal (human andfinancial) wealth diversification. Moreover, sinceCEOs are the most visible top managers, theymay be particularly interested in growing the firm



6/18


through acquisitions: who can doubt the appealof the titles awarded by the financial commu-

nity; being called a gunslinger, white knight,or raider can quicken anyones blood (Hayesand Abernathy, 1980: 76). Hayward and Ham-brick (1997) suggest that the visibility of CEOsand the accompanying public attention influencetheir acquisition behavior, and they show that CEOmedia praise actually increases the premiums theirfirms are willing to pay for acquisitions, whichsuggests that CEOs are influenced by the publicattention they receive. We expect therefore that,ceteris paribus, finance CEOs are likely to favorpersonally prestigious and risk-reducing corporate

strategies such as diversification and acquisitionto a greater extent than would finance non-CEOcorporate elites (i.e., all other executive and non-executive directors):3

Hypothesis 5: Having a finance CEO (as op-

posed to finance non-CEO corporate elites) will

be associated with high levels of diversification.

Hypothesis 6: Having a finance CEO (as op-

posed to finance non-CEO corporate elites) will

be associated with greater levels of acquisition

activity.

METHOD

Sample

To test the hypotheses, we generated a stratified,random sample of 200 industrial corporations fromthe 1985 Fortune 500, i.e., 40 corporations fromthe Fortune 100, 40 from the Fortune 101200,etc. While the benefits of random sampling arewell understood by researchers, we believe that themerits of random sampling have not been fully rec-

ognized and exploited in prior empirical work ondiversification and acquisitions (e.g., Hoskisson,Johnson, and Moesel, 1994; Haunschild et al.,1999). Specifically, the confidence with which onecan draw conclusions from empirical studies ofstrategic phenomena is significantly limited if the

3 While CEOs stronger preferences for diversification may leadfirms to pursue diversification even when it is not in the interestsof their shareholders, we are not suggesting that shareholdervalue destruction always results. Rather, we simply suggestthat the differences in preferences for diversification make itmore likely that CEOs would choose diversification over otherstrategic options than would other non-CEO corporate elites.

sample is constructed of firms that have (as arequirement for inclusion in the sample) experi-

enced the phenomenon under study. Our sample istherefore constructed independently of the depen-dent variables included in the study to avoid prob-lems associated with sample selection bias. Therandom sample also includes firms that were lateracquired by other firms, thereby avoiding survivorbias. The timeframe of the study is 1985 to 1995.

Data on acquisition activity were taken fromthe Securities Data Corporations database (SDCincludes all acquisitions over $5M) and diversifica-tion and financial data came from Compustat. Dataon corporate elite demography and board struc-

ture and composition were obtained from The Dunand Bradstreet Reference Book of Corporate Man-

agement, Standard and Poors Register of Corpo-rations, Directors, and Executives, Whos Who in

America, Whos Who in Finance and Industry, andcorporate proxy statements. Consistent with priorupper-echelons research, the study focuses only oncorporate elites, i.e., individuals who are execu-tives in the focal firm or other industrial firms.Observations (corporationyear) were eliminatedif data were missing for more than a quarter ofexecutives and non-executive directors or if CEO

data were missing (cf. Westphal and Zajac, 1997).The final sample consists of 1329 observations.

Dependent and independent variables

Following previous diversification and corporateelites research (e.g., Wiersema and Bantel, 1992;Hoskisson et al., 1993; Westphal, 1998), we mea-sure diversification using the entropy measure ofdiversification (Palepu, 1985):

n

i=1

Pi ln(1/Pi )

where P is the share of segment is sales ofthe firms total sales (dollar values) and (1/Pi )is used as a weight to account for the impor-tance of the segment for total sales. This measureis useful because it takes into consideration boththe number of segments each firm operates in aswell as the importance of each segment. How-ever, the differences in revealed preference fordiversification may be particularly pronounced inthe case of unrelated diversification as opposed to



7/18


related diversification.4 We therefore test the diver-sification hypotheses on both related and unrelated

diversification, and defined highly diversified firmsas firms whose degree of related and unrelateddiversification were one standard deviation abovethe mean level of diversification. Markides (1995)notes that it is practically impossible to know a pri-ori exactly when firms are over-diversified becauseeach firm has a different optimal limit dependingon their internal resources and competitive envi-ronment. He therefore used a similar approachto define over-diversified firms as firms whosediversification levels were one-half of a standarddeviation above the overall mean (our definition

is thus more conservative).5

Acquisition activity ismeasured in total dollar amounts per year (sum ofall acquisitions per year).

We begin by testing our predictions using themost aggregate corporate elite unit of analysis:the full board of directors. We then disaggregatethis group of corporate elites into two subgroups:executive and non-executive directors. By dis-tinguishing between executive and non-executivedirectors with similar demographic characteristics,we explicitly take into consideration how thesegroups different functions in corporate gover-nance influence their strategy preferences and dis-positions. Finally, given the CEOs unique gov-ernance position, we take the distinction betweenexecutive and non-executive directors one step fur-ther and distinguish between the CEO and the othercorporate (executive and non-executive) directors.

This process of disaggregation leads us to usefour models for each dependent variable, with themain independent variables differing across eachmodel. The first model in each set is a baselinemodel that only contains control variables. Thesecond model focuses on the number of financecorporate elites, defined here by the number of

both executive and non-executive directors withfunctional background experiences in finance (notethat all models control for the total number of cor-porate elites, obviating the need for a proportionalmeasure). Based on individuals entire pre-CEOcareer histories, we define finance background as

4 Related diversification arises from operating in several 4-digit(SIC) segments within a 2-digit segment, whereas unrelateddiversification arises from operations in several 2-digit segments(Hoskisson et al., 1993).5 To test the sensitivity of our results to different definitions ofhighly diversified, we also used Markides (1995) definition aswell as continuous measures and obtained similar results.

dominant (at least 5 years more in finance than inother functions) functional experience in finance

(or accounting and law) (note that we do notinclude executives CEO tenure when determin-ing their dominant functional background experi-ence). The third model splits the number of corpo-rate elites with finance backgrounds into financeexecutive and non-executive subsets. Note thatmost upper-echelons research equates the exec-utive directors with the top management team(e.g., Finkelstein and Hambrick, 1990; Finkel-stein, 1992; Haleblian and Finkelstein, 1993). Thefourth model distinguishes between finance CEOs(dummy variable: 1 if finance, 0 otherwise) and the

other finance non-CEO directors (including exec-utive and non-executive directors).

Models 3 and 4 thus represent different disag-gregations of the corporate elite unit of analysis,which allows distinguishing the effects of demog-raphy across agency positions: the executive vs.non-executive directors and the CEO vs. the otherexecutive and non-executive directors. In sum, thefour models, when taken together, allow compari-son of the effects of corporate elites demographicpreferences on corporate strategy across differentgovernance positions. All demographic variables

are lagged 1 year.

Control variables

Power is an important factor in studies of strate-gic decision making. Finkelstein (1992) showsthat power-weighted top management teams arestronger predictors of diversification and acqui-sitions than teams that are not power-weighted.Zajac and Westphal (1996) show similarly thatCEOs preferences for successors are realizedmore often when accompanied with greater power.

Since the CEO typically is the single most power-ful member of the corporate elite and often consid-ered the most important determinant of other cor-porate elites influence (Lorsch and MacIver, 1989;Hambrick, 1994), we control for CEO power. Bycontrolling for CEO power, we attempt to elimi-nate the alternative explanation that differences ineffects across governance positions are explainablesolely in terms of differences in power; i.e., whenCEOs are very powerful, non-CEO corporate elitesmay have little influence on corporate strategy.

Following previous research, we use two differ-ent indicators of CEO power: CEO duality (where



8/18


the same individual is both CEO and chairper-son of the board) and CEO relative board tenure

(defined as the CEOs board tenure divided by theaverage tenure of the other board members). It iseasier for both executive and non-executive direc-tors to exert influence when different individualsoccupy the CEO and chairperson roles becausesuch a governance structure provides directors withthe opportunity to organize around multiple indi-viduals with formal power. Similarly, to the extentCEOs have relatively long board tenure, they havebeen instrumental in recruiting the other directors,which makes it more difficult for them to disagreewith the CEOs. These two indicators are among

the most used indicators of power in corporate eliteresearch (e.g., Mallette and Fowler, 1992; DAveniand Kesner, 1993; Zajac and Westphal, 1996). Inthe analyses reported here, we entered the powerindicators as main effects; however, in separateanalyses, we also controlled for all the differentpowerdemography interaction effects. Neither ofthe power demography interactions was signif-icant, nor did they change the results reportedhere, which suggests that the differences acrossgovernance positions reported here are not simplyexplainable in terms of power differences.

We also control for firm size, which has beenshown to be associated with diversification lev-els and restructuring activity (Montgomery, 1982;Hoskisson et al., 1994). Firm size is defined andmeasured as the natural logarithm of total firmassets. While firm size is an indirect measure ofthe resources available to the firm to pursue diver-sification and acquisition strategies, Jensen (1986)argues that a more precise measure is the amountoffree cash flow available. Chatterjee and Werner-felt (1991) operationalize this concept, and suggestthat the availability of internal funds or unuseddebt capacity favors higher levels of diversifica-

tion. They define internal funds in terms of thedebt to market value ratio (natural logarithm ofthe ratio of long-term debt to market value) and thedebt capacity in terms of the current ratio (ratio ofcurrent assets to current liabilities). We also con-trol for firm performance, which has been shownto be associated with diversification and acquisi-tion activities (e.g., Markides, 1995; Wiersema andBantel, 1992). It is defined and measured as ROA(return on assets).

Prior diversification level has also been shownto be associated with acquisition activity (Marki-des, 1995; Hoskisson et al., 1994). This variable

is measured at time t using the entropy measuredescribed above, and we control for both lin-

ear and curvilinear (using squared terms) effects(Markides, 1995). Specifically, we control fortotal diversification, which is defined as the sumof related and unrelated diversification (Palepu,1985). Finally, the analyses also control for theyear of observation. Hoskisson and Hitt (1994)and Markides (1995) argue that firms have reducedtheir level of diversification throughout the periodstudied here. This suggests controlling for the yearof observation as firms are expected to be lessdiversified over time. We use year dummies in allthe analyses (separate analyses using a single con-

trol for year produced similar results). The finan-cial control variables are lagged 1 year. Table 1provides summary statistics and bivariate correla-tions for all data pooled.

Analysis

We used random effects logistic regression to ana-lyze whether demography and governance positionaffect the likelihood of being highly diversified.The final data structure is a pooled time series,where firm year represents the observation, andthe dependent variable is a dichotomous variable

coded one if the firm is highly diversified andzero otherwise. Unobserved heterogeneity, whichmay occur because each firm contributes multipleobservations that are not independent from eachother, is always a potential problem in pooledtime series (Petersen and Koput, 1991). A commonapproach to addressing problems of unobservedheterogeneity is to insert additional firm-specificerror terms that are either fixed over time foreach firm (fixed-effects models), or vary randomlyover time for each firm (random-effects models)(Sayrs, 1989). We used random-effects models for

the following reasons. First, fixed-effects modelstypically produce biased estimates of the fixedeffects when the time period is relatively short(Chintagunta, Jain, and Vilcassim, 1991; Heck-man, 1981). While the time frame of this studyis 10 years, some firms contribute fewer than 10observations because they were merged with otherfirms or because of missing data. Second, sincemost of the firms are either highly diversified ornot throughout the whole period, the models can-not be estimated using the fixed-effect approachbecause this approach requires variance in bothdependent and independent variables to assure that



9/18


Table1.

Summarystatisticsandbivariatecorrelations

Variable

Mean

S.D.

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

1.

Relateddiversification

0.17

0.3

8

1.0

0

2.

Unrelateddiversification

0.18

0.3

8

0.2

1

1.0

0

3.

Acquisitions(millionsof$)

104.86

637.8

7

0.0

0

0.0

0

1.0

0

4.

Financecorporateelites(#)

2.69

1.5

4

0.0

3

0.0

8

0.0

2

1.0

0

5.

Financeexecutiveelites(#)

1.23

1.0

6

0.0

1

0.0

0

0.0

1

0.6

0

1.0

0

6.

Financenon-executiveelites(#)

1.46

1.2

3

0.0

5

0.1

0

0.0

2

0.7

4

0.1

0

1.0

0

7.

FinanceCEO(#)

0.26

0.4

3

0.0

3

0.1

1

0.0

1

0.2

6

0.3

5

0.0

2

1.0

0

8.

Financenon-CEOelites(#)

2.43

1.4

9

0.0

2

0.1

1

0.0

2

0.9

6

0.5

2

0.7

5

0.0

3

1.0

0

9.

Totalnumberofcorporateelites

9.01

2.5

4

0.1

1

0.1

4

0.0

5

0.5

3

0.3

9

0.3

3

0.0

1

0.5

5

1.0

0

10.

CEOrelativeboardtenure

1.64

1.3

2

0.0

3

0.0

2

0.0

2

0.0

5

0.0

8

0.0

1

0.0

5

0.0

7

0.1

2

1.00

11.

CEOduality

0.83

0.3

8

0.0

7

0.0

5

0.0

4

0.0

1

0.0

7

0.0

5

0.0

3

0.0

2

0.0

1

0.22

1.0

0

12.

Assets(ln)

7.61

1.1

6

0.0

4

0.0

6

0.2

0

0.1

2

0.1

2

0.0

5

0.0

3

0.1

4

0.2

7

0.02

0.1

3

1.0

0

13.

Debt/marked(ln)

0.96

1.1

7

0.1

0

0.0

6

0.0

3

0.1

2

0.0

5

0.1

1

0.0

8

0.1

0

0.0

4

0.03

0.0

1

0.0

8

1.0

0

14.

Currentratio

1.80

0.6

6

0.0

4

0.0

7

0.0

2

0.1

1

0.0

8

0.0

7

0.0

3

0.1

1

0.0

8

0.04

0.1

5

0.4

1

0.2

3

1.0

0

15.

ROA(%)

4.77

7.0

5

0.0

2

0.0

1

0.0

6

0.0

1

0.0

4

0.0

5

0.0

0

0.0

1

0.0

7

0.05

0.0

4

0.0

0

0.5

7

0.2

5

1.0

0

16.

Diversification(t)

0.84

0.5

6

0.3

1

0.4

1

0.0

5

0.0

6

0.0

2

0.0

9

0.1

4

0.0

2

0.0

0

0.01

0.0

1

0.1

7

0.1

2

0.1

6

0.0

3

1.00

17.

Diversification

2

(t)

1.03

1.0

2

0.2

8

0.4

2

0.0

3

0.0

2

0.0

3

0.0

5

0.1

1

0.0

1

0.0

3

0.00

0.0

1

0.1

5

0.1

0

0.1

4

0.0

3

0.94

1.0

0

Correlationcoefficientsabove0.0

6aresignificantatthep

c and

yit = c if y < c

where c is the threshold for censoring (zero inthis case).6 A DurbinWatson test showed that

autocorrelation is not a problem (DurbinWatsonstatistic = 1.99). To test the stability of the results,we recoded the dependent variable into dichoto-mous variables and used a random effects logisticregression procedure to test the hypotheses. Again,the results are fundamentally the same as thoseobtained using the tobit procedure. We finally used

6 A likelihood-ratio test comparing the random-effects modelwith the pooled tobit model was insignificant, suggesting thatthe panel-level variance component is unimportant and separateanalyses confirmed that the results did not differ across thesemodels.

maximum-likelihood Wald tests to test the signif-icance of differences between regression coeffi-

cients (Greene, 1997).

RESULTS

Diversification

Table 2 reports the results of random-effects logis-tic regression analyses of the predictors of highdiversification. Models 2 and 6 suggest that firmswith more finance corporate elites are more likelyto be highly diversified, as suggested in Hypoth-esis 1. However, comparing the two models alsosuggests that the effect of finance corporate elitesis stronger in the case of unrelated diversifica-tion than related diversification. The disaggrega-tion of the corporate elites into executive and non-executive directors is shown in Models 3 and 7.They show that neither the finance executive direc-tor nor the finance non-executive director unitsof analysis predict the likelihood of high relatedor high unrelated diversification. Moreover, thedifference between the finance executive direc-tor and finance non-executive director effects is

not statistically significant in either of the cases.These results indicate that Hypothesis 3, whichhad argued that finance executive directors areassociated with more diversification than financenon-executive directors, cannot be supported.

This non-finding might appear surprising, giventhe prevalent use of the insider (executive) vs.outsider (non-executive) director variable in topmanagement team research (e.g., Finkelstein andHambrick, 1990; Haleblian and Finkelstein, 1993).However, we investigated this issue further inempirical analyses not reported here, and found

that the main reason for these non-significant find-ings was that the finance non-CEO executiveswere more like the finance non-executives than thefinance CEOs, which interestingly provides furthersupport for the logic of our further disaggregationbetween CEOs and non-CEO elites, the results ofwhich are shown in Models 4 and 8 below.

Models 4 and 8, which distinguish betweenfinance CEOs and finance non-CEO directors,show that firms with finance CEOs are signifi-cantly more likely to be highly diversified than arefirms with non-finance CEOs. Having a financeCEO (as opposed to finance non-CEO directors)



11/18


Table 2. Random effects logistic regression: high diversificationa

Related diversification Unrelated diversification

(1) (2) (3) (4) (5) (6) (7) (8)

Finance corporate 0.23 0.41

elites (#) (0.17) (0.15)Finance executive 0.27 0.01

elites (#) (0.30) (0.22)Finance non- 0.00 0.04

executive elites (#) (0.23) (0.19)Finance CEO (#) 1.84 1.31

(0.50) (0.45)Finance non-CEO 0.18 0.68

elites (#) (0.15) (0.18)

Total number of 0.03 0.05 0.07 0.01 0.15 0.51 0.23 0.00

corporate elites (0.08) (0.13) (0.15) (0.11) (0.08) (0.14) (0.12) (0.11)CEO relative 0.26 0.31 0.20 0.10 0.04 0.16 0.23 0.03

board tenure (0.15) (0.13) (0.13) (0.13) (0.15) (0.16) (0.14) (0.15)CEO duality 1.07 1.16 1.06 1.29 0.27 0.61 0.36 0.50

(0.50) (0.51) (0.49) (0.54) (0.52) (0.57) (0.55) (0.57)Assets (ln) 0.08 0.00 0.21 0.39 0.42 0.88 0.38 0.94

(0.15) (0.18) (0.23) (0.19) (0.24) (0.29) (0.30) (0.29)Debt/market (ln) 0.46 1.14 0.31 1.06 0.09 0.82 1.09 0.52

(0.24) (0.27) (0.28) (0.29) (0.21) (0.22) (0.21) (0.20)Current ratio 0.11 0.83 0.36 0.86 0.01 0.24 0.07 0.98

(0.42) (0.38) (0.44) (0.41) (0.42) (0.40) (0.41) (0.42)Return on assets 0.08 0.05 0.07 0.05 0.05 0.09 0.05 0.06

(ROA) (0.03) (0.03) (0.03) (0.03) (0.04) (0.04) (0.03) (0.04)Year 1986 0.24 0.17 0.25 0.23 1.15 0.93 0.90 1.09

(0.67) (0.67) (0.68) (0.67) (0.69) (0.68) (0.64) (0.75)

Year 1987 0.58 0.63 0.63 0.69 2.15 2.09 1.56 2.28(0.67) (0.68) (0.68) (0.68) (0.72) (0.72) (0.65) (0.78)

Year 1988 0.33 0.43 0.34 0.60 2.34 2.43 2.01 2.83

(0.69) (0.69) (0.69) (0.70) (0.74) (0.73) (0.70) (0.81)Year 1989 0.34 0.16 0.41 0.04 3.28 3.35 2.72 3.48

(0.69) (0.70) (0.70) (0.72) (0.80) (0.83) (0.77) (0.86)Year 1990 0.04 0.12 0.06 0.35 2.77 2.55 2.07 2.76

(0.68) (0.70) (0.70) (0.72) (0.80) (0.77) (0.76) (0.86)Year 1991 0.27 0.44 0.26 0.88 1.83 2.09 1.42 2.03

(0.69) (0.71) (0.71) (0.76) (0.79) (0.77) (0.72) (0.84)Year 1992 0.81 0.99 0.77 1.62 1.75 1.88 1.44 1.87

(0.72) (0.73) (0.75) (0.75) (0.77) (0.77) (0.74) (0.82)Year 1993 1.28 1.48 1.28 1.95 2.60 2.60 2.03 2.86

(0.70) (0.73) (0.72) (0.75) (0.79) (0.81) (0.77) (0.88)Year 1994 1.31 1.57 1.28 2.08 3.46 3.56 2.81 3.72

(0.72) (0.76) (0.76) (0.77) (0.79) (0.85) (0.84) (0.87)Year 1995 0.16 0.52 0.19 1.06 3.93 4.32 3.82 4.38

(0.72) (0.79) (0.73) (0.77) (0.81) (0.89) (0.93) (0.93)Constant 4.71 4.54 3.04 8.06 6.26 5.92 2.82 10.99

(1.96) (21.71) (2.23) (2.14) (2.18) (2.17) (2.49) (2.65)

2 24.37 46.05 25.86 52.07 37.59 55.73 59.74 49.16

d.f. 17 18 19 19 17 18 19 19n 1329 1329 1329 1329 1329 1329 1329 1329

a Unstandardized coefficients. One-way tests for hypothesized effects, two-way tests otherwise.p < 0.10; p < 0.05; p < 0.01; p < 0.001



12/18


is also more strongly associated with both highrelated and high unrelated diversification ( 2(1) =

10.69, p < 0.01 and 2(1) = 14.51, p < 0.001),thus supporting Hypothesis 5. Interestingly, hav-ing finance non-CEO directors actually reducesthe likelihood a firm is highly diversified, whichprovides further support for the agency argumentthat CEOs and other corporate elites may havemarkedly different revealed preferences for diver-sification. The differences across Models 24 and68 suggest the value of distinguishing betweenindividuals occupying the CEO position and allthe other corporate elites to understand more fullythe association between functional background and

diversification.These results illustrate the subtleties in theempirical relationships between corporate elitesfunctional background experiences and diversifi-cation, and also suggest that the use of aggre-gate units of analysis may be problematic. Specifi-cally, the results show that aggregate effects, whichappear to indicate no significant relationships, infact may mask significant disaggregate effects. Thepositive and significant finance CEO effect wouldhave remained undetected had we not parsed outthe differences in governance position. The resultsthus illustrate the importance of considering bothdemography and governance position in analysesof corporate elites effects on strategy. Stated dif-ferently, our study reveals the danger of mistakenlyviewing all corporate elites as part of a like-mindedteam.

Acquisitions

Table 3 reports the results of random-effects tobitregression analyses of acquisition activity. Model 2shows that, contrary to Hypothesis 2, having morefinance corporate elites actually decreases a firms

propensity for acquisition activity. However, Mod-els 3 and 4 show (similar to the diversificationfindings discussed above) that failing to considerdifferences in governance positions masks impor-tant differences in predicting acquisition activity.Specifically, Model 3 indicates that the surpris-ingly strong negative effect of finance corporateelites on acquisitions in Model 2 is based on anegative effect of finance non-executive directors,not finance executive directors. However, sincethe effect of finance executive directors is not sig-nificantly different (F (1, 1308) = 2.90; p > 0.05)from the effect of finance non-executive directors,

Hypothesis 4 cannot be supported: finance execu-tive directors are not associated with more acquisi-

tion activity than finance non-executive directors.Finally, Model 4 shows once again the empir-

ical importance of our theoretical choice to com-pletely disaggregate the units of analysis. It showsthat the insignificant finance executive directoreffect in Model 3 in fact masks a significantfinance CEO effect. Moreover, the finance CEOeffect is significantly larger than the effect offinance non-CEO directors (F (1, 1308) = 6.56;p < 0.05). This supports Hypothesis 6, whichargued that finance CEOs compared to financenon-CEO directors are associated with more acqui-

sition activity. Thus, the overall negative financedirector effect in Model 2 is more appropriatelyexpressed in terms of a positive finance CEO effectand a negative finance non-CEO director effect.This conclusion is again, as noted above, vali-dated by the overall improvements in the signifi-cance of the acquisition models. Separate analysesalso showed that the non-CEO executive direc-tors align more closely with the non-executivedirectors, even when controlling for power interac-tions. These results thus confirm the diversificationresults, suggesting that it is critically important toconsider the importance of governance positionswhen analyzing the effects of demographic char-acteristics on corporate strategy.

To assess the substantive significance of usingdisaggregate units of analysis, we also calculatedthe dollar equivalents of the acquisition modelstobit coefficients and their two constitutive parts.The tobit coefficients indicate the marginal propen-sity or capacity of a firm to engage in acquisitionactivities and can be transformed into dollar valuesto provide more intuitive and substantive interpre-tation of the effects (Breen, 1996). Expressed indollar values, replacing a non-finance CEO with a

finance CEO results in roughly $103 million morespent on acquisitions per year (these values arecalculated at the mean level of the independentvariables (Breen, 1996)). Following McDonald andMoffitt (1980), these effects can be decomposedinto two subeffects which show that most of thefinance CEO effect derives from increasing theprobability of doing acquisitions ($65 million or63 percent), whereas a smaller part derives fromincreasing the size of the acquisitions ($38 millionor 37 percent). The marginal effect of finance non-CEO directors is quite different. It reduces theaverage size of acquisitions by roughly $53 million



13/18


Table 3. Random-effects tobit regression: acquisitionsa

(1) (2) (3) (4)

Finance corporate elites (#) 91.59

(42.22)Finance executive elites (#) 26.76

(56.51)Finance non-executive elites (#) 137.65

(50.57)Finance CEO (#) 199.98

(120.79)Finance non-CEO elites (#) 128.13

(44.77)

Total number of corporate elites 21.00 8.39 6.67 21.00(21.79) (25.54) (25.57) (26.00)

CEO relative board tenure 52.07 50.59 57.87 51.58(44.22) (43.93) (44.40) (44.11)

CEO duality 210.19 217.23 242.42 186.04(156.20) (156.26) (157.30) (156.44)

Assets (ln) 437.26 429.10 424.20 434.16

(53.31) (53.20) (53.26) (53.07)Debt/market (ln) 54.44 39.34 39.59 44.17

(58.78) (59.17) (59.23) (59.14)Current ratio 116.71 103.46 109.35 99.11

(91.99) (92.06) (92.37) (92.16)Return on assets (ROA) 24.45 25.30 24.99 25.16

(10.75) (10.77) (10.83) (10.86)Diversification level (t) 670.78 624.42 600.31 675.03

(282.81) (283.52) (284.44) (284.05)Diversification level2 (t) 368.43 347.75 342.62 363.16

(162.81) (162.98) (163.39) (162.92)Year 1986 375.97 388.87 393.11 375.17

(242.79) (242.40) (243.13) (242.17)Year 1987 121.54 124.94 130.06 116.84

(253.51) (253.36) (254.05) (252.83)Year 1988 210.09 198.79 204.07 193.08

(248.77) (248.62) (249.23) (248.09)Year 1989 503.14 501.12 510.31 498.44

(243.59) (243.26) (244.02) (242.55)Year 1990 30.95 25.27 28.14 16.71

(268.56) (268.62) (269.66) (268.63)Year 1991 130.35 136.56 123.82 150.65

(273.56) (273.46) (274.46) (273.47)Year 1992 88.09 86.81 99.17 53.93

(259.30) (258.82) (259.86) (259.11)

Year 1993 192.34 179.25 196.38 153.80(262.86) (262.75) (263.83) (262.63)Year 1994 415.85 403.59 414.85 381.26

(258.23) (257.92) (258.82) (257.54)Year 1995 425.98 418.05 436.56 395.27

(257.40) (257.05) (258.13) (256.64)Constant 4968.19 4914.82 4922.84 4984.92

(564.84) (564.76) (566.28) (564.68)

2 115.62 120.38 123.30 126.94

d.f. 19 20 21 21n 1329 1329 1329 1329

a Unstandardized coefficients. One-way tests for hypothesized effects, two-way tests otherwise.p < 0.10; p < 0.05; p < 0.01; p < 0.001



14/18


a year split between $13 million less for firmsalready acquiring and $40 million less from lower-

ing the probability of engaging in the activity. Theproblems of aggregate units of analysis are clearlyillustrated by comparing the detailed informationobtained from the disaggregated finance CEO andfinance non-CEO director variables on the onehand and the aggregate finance corporate direc-tor variable on the other hand. Had we stoppedwith the aggregate-level analysis, we would sim-ply have concluded that the marginal effect offinance corporate directors is to reduce acquisitionsby roughly $40 million a year.

Supplemental analyses

Our prediction regarding an expected positive rela-tionship between finance CEOs and highly diver-sified firms was based on our theorizing aboutfinance CEOs having relatively stronger prefer-ences for diversification (relative to non-financeCEOs). To further distinguish between financeCEOs preferences for high levels of diversifica-tion vs. their particular expertise to create share-holder value through high levels of diversification,we collected additional data on firm performanceand estimated a new set of regression models(see Appendix). If finance CEOs were using theirexpertise to create shareholder value through diver-sification, one would expect that highly diversifiedcorporations with finance CEOs performed bet-ter than other highly diversified corporations. Wedefined performance in terms of relative changes intotal shareholder value ([market capitalizationt+dividendst market capitalizationt1]/market capi-talizationt1) and used the multiplicative interac-tion approach (Gupta and Govindarajan, 1993) totest whether there was a positive interaction effecton performance of having a finance CEO and being

highly diversified.7As the Appendix shows, the interaction effects

are insignificant, indicating that highly diversi-fied corporations with finance CEOs performedno better than other highly diversified firms. Thissuggests that the positive relationship that wehypothesized and found between finance CEOs andhigh diversification is not attributable to a financeCEOs unique expertise in creating shareholder

7 We also tested used alternative performance indicators suchas return on assets and price/earnings ratio and found similarresults.

value through high levels of corporate diversifica-tion. In other words, we are now able to show that

finance CEOs engage in significantly higher diver-sification, as hypothesized, and that this diversifi-cation is not creating shareholder value.8

IMPLICATIONS AND CONCLUSIONS

This study has sought to develop and test a theo-retical framework that (1) combines insights fromboth the agency and the upper-echelons perspec-tives, and (2) highlights the importance of consid-ering how both demography and position affect

the relationship between corporate elites and cor-porate strategy. We argued and found empiri-cally that focusing on only one aspect at theexpense of the other may lead to an unfortu-nately narrow choice of units of analysis that canoversimplify or even mask the true relationshipbetween corporate elites and the scope of the firm.

For example, there first appeared to be no sup-port (or contradictory support) for the demograph-ically based notion that corporate elites functionalbackground experiences affect diversification andacquisition activities. However, by taking gover-nance position into account and disaggregating

elites into seemingly more relevant governancepositions, we demonstrated that the insignificantor contradictory effects of corporate elite demogra-phy mask significantly different subgroup effects.In the case of both unrelated diversification andacquisitions, the aggregate-level effects masked apositive finance CEO effect and a negative financenon-CEO effect. In other words, by pushing theagency theory argument that differences in gover-nance positions imply differences in strategy pref-erences to its logical limit, we provided clearerinsights into the role of elites demography on

corporate strategy. Given the increased popular-ity of aggregate units of analysis, such as thetop management team or the board of directors,documented by Finkelstein and Hambrick (1996),the results of the present study suggest that cau-tion should be exercised when interpreting researchusing these aggregate units of analysis.

8 Of course, this does not imply that finance CEOs typicallymake worse decisions than non-finance CEOs when engagingin high levels of diversification, or that they never increaseshareholder value. Rather, the absence of shareholder valuecreation due to high diversification by finance CEOs reflectsthe central tendencies found in our large sample of firms.



15/18


Some scholars argue that aggregate executiveunits of analysis, such as the top management

team, are superior to the CEO unit of analysisbecause they explain more variance than the CEOunit of analysis (Hage and Dewar, 1973; Ban-tel and Jackson, 1989; Finkelstein, 1992). Whileaggregate units of analysis that include the CEOmay explain more variance than the CEO unitalone, our study makes a different point. We do notsuggest excluding the CEO, non-CEO executives,or non-executives from analyses of corporate elitesand corporate strategy. Instead, we suggest thatit is critically important to aggregate or disaggre-gate along the appropriate governance positions.

In fact, our study suggests the value of includingthe CEO, non-CEO executives, and non-executivedirectorsbut distinguishing between them.

We would therefore advocate precisely the oppo-site approach: rather than focusing on the cor-porate elites as an aggregate whole, one shoulddistinguish between all the different subgroups ofcorporate elites that occupy similar governanceposition. Where this is not feasible, an alterna-tive strategy would be to return to the simple CEOunit of analysis. While this strategy has the dis-advantage of not considering the other corporateelites and making it impossible to address group-level phenomena such as group heterogeneity (e.g.,Wiersema and Bantel, 1992), it is at least rela-tively unambiguous (our study showed that thecompletely disaggregated model, which isolatedthe CEO effects, generally had greater predictivesignificance).

While we have focused specifically on theagency context to address the interplay betweengovernance position and demography, future re-search could also examine whether other con-textual factors might have similar influence oncorporate elites demographic preferences. Indeed,

some scholars have begun focusing on relation-ship between different organizational and envi-ronmental contexts and corporate elites structureand composition (e.g., Keck and Tushman, 1993;Keck, 1997). However, our findings still highlightthe need for corporate elite-based research to con-sider how the relationships under study may differaccording to the alternative governance positionscorporate elites occupy.

Another contribution that we hope emerges fromour joint demography/agency perspective is thefocus on when one is likely to observe phenom-ena such as very high levels of diversification.

While agency researchers have alleged such behav-iors in broad terms, they have not isolated where

and why certain corporate elites are more likelyto exhibit these specific behaviors. We believe ourstudy is distinctive in actually assessing the extentto which demographically based preferences (suchas those stemming from a particular functionalbackground) lead to extreme behaviors (rather thansimply above-average levels of behavior) amongindividual corporate elites. We hope that our the-oretical recognition of the aggregation problem inTMT research, along with our attempt to very care-fully analyze this issue empirically, will representan important contribution in redirecting research in

this area.Finally, we would also like to note that, irre-

spective of the disaggregation question, the studysfindings highlight that corporate elites do in factinfluence corporate strategies above and beyondeconomic factors such as prior performance, re-source scarcity, and firm size. In this way, wesee the study as contributing to the larger bodyof corporate strategy research on diversificationand acquisitions. Given that numerous prior studieshave used either the agency or the upper-echelonsperspectives to study these phenomena, this empir-ical context was very well suited for highlightinghow our attempt at a more nuanced, syntheticapproach could add to that line of inquiry. In thisway, we also hope our study is seen as providing abehavioral complement to the more traditional eco-nomic perspectives on diversification and acquisi-tions.

ACKNOWLEDGEMENTS

We would like to thank J. Richard Harrison, Matt

Kraatz, Willie Ocasio, Jim Westphal, and seminarparticipants at Northwestern University for theirhelpful comments. We thank Rick Johnson andBob Hoskisson for allowing us to access someof their data, and Noriko Shiomi for providingvaluable assistance in data collection. Both authorscontributed equally.

REFERENCES

American Law Institute. 1984. Principles of CorporateGovernance. Draft 2.



16/18


Amihud Y, Lev B. 1981. Risk reduction as a managerialmotive for conglomerate mergers. Bell Journal of

Economics 12: 605616.Bantel KA, Jackson SE. 1989. Top management and

innovations in banking: does the composition of thetop team make a difference? Strategic Management

Journal , Summer Special Issue 10: 107124.Beatty RP, Zajac EJ. 1994. Managerial incentives,

monitoring, and risk bearing: a study of executivecompensation, ownership, and board structure in initialpublic offerings. Administrative Science Quarterly 39:313335.

Beyer JM, Chattopadhyay P, George E, Glick WH, ogil-vie dt, Pugliese D. 1997. The selective perception ofmanagers revisited. Academy of Management Journal40: 716737.

Breen R. 1996. Regression Models: Censored, Sample

Selected, or Truncated Data . Sage: Thousand Oaks,CA.

Business Roundtable. 1990. Corporate Governance andAmerican Competitiveness.

Cannella AA Jr, Monroe MJ. 1997. Contrasting perspec-tives on strategic leaders: toward a more realisticview of top managers. Journal of Management 23:213237.

Chatterjee S, Wernerfelt B. 1991. The link betweenresources and type of diversification: theory andevidence. Strategic Management Journal 12(1):3348.

Chattopadhyay P, Glick WH, Miller CC, Huber GP.1999. Determinants of executive beliefs: comparing

functional conditioning and social influence. StrategicManagement Journal 20(8): 763789.Chintagunta PK, Jain DC, Vilcassim NJ. 1991. Investi-

gating heterogeneity in brand preferences in logit mod-els for panel data. Journal of Marketing Research 28:417428.

DAveni RA, Kesner IF. 1993. Top managerial prestige,power and tender offer response: a study of elite socialnetworks and target firm cooperation during takeovers.Organization Science 4: 123151.

Dalton DR, Daily CM, Ellstrand AE, Johnson JL. 1998.Meta-analytic reviews of board composition, leader-ship structure, and financial performance. Strategic

Management Journal 19(3): 269290.Dearborn DC, Simon HA. 1958. Selective perception: a

note on the departmental identifications of executives.Sociometry 21: 140144.

Demb A, Neubauer FF. 1992. The Corporate Board.Oxford University Press: Oxford.

Fama EF. 1980. Agency problems and the theory of thefirm. Journal of Political Economy 88: 288307.

Fama EF, Jensen MC. 1983. Separation of ownership andcontrol. Journal of Law and Economics 26: 301325.

Finkelstein S. 1992. Power in top management teams:dimensions, measurement, and validation. Academy of

Management Journal 35: 505538.Finkelstein S, Hambrick DC. 1990. Top-management-

team tenure and organizational outcomes: the mod-erating roles of managerial discretion. AdministrativeScience Quarterly 35: 484503.

Finkelstein S, Hambrick DC. 1996. Strategic Leadership:Top Executives and Their Effects on Organizations.

West: St Paul, MN.Fligstein N. 1987. The intraorganizational power strug-gle: rise of finance personnel to top leadership inlarge corporations, 1919 1979. American Sociologi-cal Review 52: 4458.

Fligstein N. 1990. The Transformation of CorporateControl. Harvard University Press: Cambridge, MA.

Gomez-Mejia LR, Wiseman RM. 1997. Reframing exec-utive compensation: an assessment and outlook. Jour-nal of Management 23: 291374.

Goodstein J, Boeker W. 1991. Turbulence at the top: anew perspective on governance structure changes andstrategic change. Academy of Management Journal 34:306330.

Greene WH. 1997. Econometric Analysis . Prentice-Hall:

Englewood Cliffs, NJ.Gupta AK, Govindarajan V. 1993. Methodological issues

in testing contingency theories: an assessment ofalternative approaches. In Creative and Innovative

Approaches to the Science of Management, Ijiri Y(ed). Quorum: Westport, CT; 453 471.

Hage J, Dewar R. 1973. Elite values versus organiza-tional structure in predicting innovation. Administra-tive Science Quarterly 18: 279290.

Haleblian J, Finkelstein S. 1993. Top management teamsize, CEO dominance, and firm performance: themoderating roles of environmental turbulence anddiscretion. Academy of Management Journal 36:844863.

Hambrick DC. 1994. Top management group: a concep-

tual integration and reconsideration of the team label.In Research in Organizational Behavior, Staw BM,Cummings LL (eds). JAI Press: Greenwich, CT;171213.

Hambrick DC, Mason PA. 1984. Upper echelons: theorganization as a reflection of its top managers.

Academy of Management Review 9: 193206.Haunschild PR, Henderson AD, Davis-Blake A. 1999.

CEO demographics and acquisitions: network effectsof educational and functional background. In Cor-

porate Social Capital and Liability, Leenders RTAJ,Gabbay SM (eds). Kluwer: Boston, MA; 266283.

Hayes RH, Abernathy WJ. 1980. Managing our way toeconomic decline. Harvard Business Review 58(4):6777.

Hayward MLA, Hambrick DC. 1997. Explaining thepremiums paid for large acquisitions: evidence ofCEO hubris. Administrative Science Quarterly 42:103127.

Heckman JJ. 1981. The incidental parameters problemand the problem of initial conditions in estimatinga discrete time-discrete data stochastic process. InStructural Analysis of Discrete Data with Econometric

Applications, Manski C, McFadden D (eds). MITPress: Cambridge, MA; 179195.

Hoskisson RE, Hitt MA. 1994. Downscoping: How toTame the Diversified Firm. Oxford University Press:New York.

Hoskisson RE, Hitt MA, Johnson RA, Moesel DD. 1993.Construct validity of an objective (entropy) categorical



17/18


measure of diversification strategy. Strategic Manage-ment Journal 14(3): 215235.

Hoskisson RE, Johnson RA, Moesel DD. 1994. Corpo-rate divestiture intensity in restructuring firms: effectsof governance, strategy, and performance. Academy of

Management Journal 37: 1207 1251.Jackson SE. 1992. Consequences of group composition

for the interpersonal dynamics of strategic issueprocessing. In Advances in Strategic Management,Vol. 8, Shrivastava P, Huff AS, Dutton JE (eds). JAIPress: Greenwich, CT; 345382.

Jensen MC. 1986. Agency costs of free cash flow,corporate finance, and takeovers. American Economic

Review 76: 323329.Jensen MC, Meckling WH. 1976. Theory of the

firm: managerial behavior, agency costs, andownership structure. Journal of Financial Economics

3: 305360.Judge GG, Griffiths WE, Hill RC, Lee T-C. 1985. The

Theory and Practice of Econometrics. Wiley: NewYork.

Keck SL. 1997. Top management team structure: differ-ential effects by environmental context. OrganizationScience 8: 143156.

Keck SL, Tushman ML. 1993. Environmental andorganizational context and executive team structure.

Academy of Management Journal 36: 1314 1344.Lane PJ, Cannella AA Jr, Lubatkin MH. 1998. Agency

problems as antecedents to unrelated diversification:Amihud and Lev reconsidered. Strategic Management

Journal 19(6): 555578.Lorsch JW, MacIver E. 1989. Pawns or Potentates:

The Reality of Americas Corporate Boards . HarvardBusiness School Press: Boston, MA.

Mallette P, Fowler KL. 1992. Effects of board com-position and stock ownership on the adoption ofpoison pills. Academy of Management Journal 35:10101035.

Markides CC. 1995. Diversification, Refocusing, andEconomic Performance. MIT Press: Cambridge, MA.

McDonald JF, Moffitt RA. 1980. The uses of tobitanalysis. Review of Economics and Statistics 62:318321.

McNulty T, Pettigrew AM. 1996. The contribution,power and influence of part-time board members.Corporate Governance 4: 160179.

Michel JG, Hambrick DC. 1992. Diversification posture

and top management team characteristics. Academy ofManagement Journal 35: 935.

Montgomery CA. 1982. The measurement of firm diver-sification: some new empirical evidence. Academy of

Management Journal 28: 299307.Ocasio W. 1994. Political dynamics and the circulation

of power: CEO succession in U.S. industrial

corporations, 1960 1990. Administrative ScienceQuarterly 39: 285312.

Palepu K. 1985. Diversification strategy, profit perfor-mance and the entropy measure. Strategic Manage-ment Journal 6(3): 239255.

Palmer D, Barber BM. 2001. Challengers, elites, andowning families: a social class theory of corporateacquisitions in the 1960s. Administrative ScienceQuarterly 46: 87120.

Petersen T, Koput KW. 1991. Density dependence inorganizational mortality: legitimacy or unobservedheterogeneity? American Sociological Review 56:399409.

Salter MS, Weinhold WA. 1978. Diversification viaacquisition: creating value. Harvard Business Review56(4): 166176.

Sayrs LW. 1989. Pooled Time Series Analysis. Sage:

Newbury Park, CA.Song JH. 1982. Diversification strategies and the

experience of top executives of large firms. StrategicManagement Journal 3(4): 377380.

Tobin J. 1958. Estimation of relationships for limiteddependent variables. Econometrica 26: 2436.

Tosi HL, Gomez-Mejia LR. 1989. The decoupling ofCEO pay and performance: an agency theoryperspective. Administrative Science Quarterly 34:169189.

Waller MJ, Huber GP, Glick WH. 1995. Functionalbackground as a determinant of executive perception.

Academy of Management Journal 38: 943974.Walsh JP. 1988. Selectivity and selective perception:

an investigation of managers belief structures and

information processing. Academy of ManagementJournal 31: 873896.

Walsh JP, Seward JK. 1990. On the efficiency of internaland external corporate control mechanisms. Academyof Management Review 15: 421458.

Westphal JD. 1998. Board games: how CEOs adaptto increases in structural board independence frommanagement. Administrative Science Quarterly 43:511537.

Westphal JD, Zajac EJ. 1997. Defections from theinner circle: social exchange, reciprocity, and thediffusion of board independence in U.S. corporations.

Administrative Science Quarterly 42: 161183.Wiersema MF, Bantel KA. 1992. Top management team

demography and corporate strategic change. Academy

of Management Journal 35: 91121.Zajac EJ, Westphal JD. 1996. Director reputation,

CEO/board power, and the dynamics of boardinterlocks. Administrative Science Quarterly 41:507529.



18/18


APPENDIX: FIXED-EFFECTS OLS REGRESSION: TOTAL SHAREHOLDERRETURNa

(1) (2)

High related diversification (0 1) 0.04(0.06)

High unrelated diversification (01) 0.02(0.06)

Finance CEO (#) 0.04 0.05(0.04) (0.07)

Finance CEO * High related diversification 0.01(0.02)

Finance CEO * High unrelated diversification 0.06(0.07)

Acquisition (01) 0.01 0.01(0.02) (0.02)

Divestiture (01) 0.05 0.05

(0.02) (0.02)Finance non-CEO elites (#) 0.03 0.03

(0.01) (0.01)Total number of corporate elites 0.02 0.02

(0.01) (0.01)CEO relative board tenure 0.02 0.01

(0.01) (0.03)CEO duality 0.01 0.01

(0.03) (0.03)Assets (ln) 0.36 0.36

(0.05) (0.05)

Debt/market (ln) 0.16

0.16

(0.03) (0.03)Current ratio 0.06 0.06

(0.03) (0.03)Return on assets (ROA) 0.00 0.00

(0.00) (0.00)Constant 3.06 3.07

(0.38) (0.38)

R2 0.23 0.23F(22, 1003) 13.42 13.44

n 1179 1179

a Unstandardized coefficients.p < 0.10; p < 0.05; p < 0.01; p < 0.001. All models include year dummies.


Documents

Corporate Elites and Corporate Strategy