Defections from theInner Circle: SocialExchange, Reciprocity,and the Diffusion ofBoard Independence inU.S. Corporations

James D. WestphalUniversity of Texas at AustinEdward J. ZajacNorthwestern University

© 1997 by Cornell University.0001-8392/97/4201-0161/$1.00.

Both authors contributed equally to thepaper. The helpful comments of RonBurt, Jerry Davis, Ranjay Gulati, PaulHirsch, Mark Shanley, and Brian Uzzi areappreciated.

This study seeks to reconcile traditional sociologicalviews of the corporate board as an instrument of elitecohesion with recent evidence of greater board activismand control over top management. We propose thatCEO-directors may typically support fellow CEOs byimpeding increased board control over management butthat CEO-directors may also foster this change if theyhave experienced it in their own corporation. Drawing onsocial exchange theory, we develop and test the argu-ment that these CEO-directors may experience a reversalin the basis for generalized social exchange with othertop managers from one of deference and support to oneof independence and control. Using data from a largesample of major U.S. corporations over a recent ten-yearperiod, we show (1) how CEO-directors "defect" from thenetwork of mutually supportive corporate leaders, (2)how defections have diffused across organizations andover time, and (3) how this has contributed to increasedboard control, as measured by changes in board struc-ture, diversification strategy, and contingent compensa-tion. We also provide evidence that a social exchangeperspective can explain the diffusion of these changesbetter than more conventional perspectives on networkdiffusion that emphasize imitation or learning.*

With few exceptions, organizational researchers and corpo-rate governance experts have historically viewed corporateboards of directors as "rubber stamps" for managennentinitiatives (Herman, 1981) or as "tools" of top management(Pfeffer, 1972: 219). From this viewpoint, boards are popu-lated either by (1) inside directors who can ill afford tocriticize their superiors or themselves, (2) uninformedoutsiders who are unable to evaluate top management, or(3) more knowledgeable outside directors who are CEOsthemselves and whose empathy for their fellow chiefexecutive officers may diminish their willingness to monitorthem actively. Interlocking directorships among top manag-ers have also traditionally been viewed as contributing to acohesive "inner circle" of organizational elites accountableonly to themselves (Useem, 1984). From this traditionalperspective, overlapping board memberships provide acommunication network for managerial elites that helps topreserve their corporate power (Useem, 1984; Davis, 1991;Mizruchi, 1996).

More recently, however, the business press has usednumerous vivid cases to illustrate a dramatic shift towardincreased board activism and control over top management(e.g.. Fortune, 1993; Business Week, 1994). This apparentchange among large corporations is not easily reconcilablewith the depiction of corporate boards as a passive grouprepresenting a cohesive inner circle of corporate leaders,since it suggests that boards are increasingly likely to actindependently of top management's preferences. Recentattempts to explain the causes of this change towardincreased board monitoring, or internal corporate control(Walsh and Seward, 1990), have usually referred to externalforces, notably increased activism by large, institutionalinvestors {Fortune, 1993; Davis and Thompson, 1994).Several recent empirical studies suggest, however, that the

161/Administrative Science Quarterly, 42 (1997): 161-183

magnitude of institutional investor ownership does notnecessarily increase the willingness of boards to challengemanagement (e.g.. Daily, 1996; Kim and Ocasio, 1995;Sundaramurthy, 1996; Westphal and Zajac, 1997). Whileinstitutional investors may be able to exert their will onspecific issues where they have voting power, it is not clearwhether and how they can overcome longstanding socialrelationships and norms of mutual support that are assumedto exist between managers and directors.

The present study proposes a more comprehensive explana-tion of the rise in board independence that takes intoaccount the social exchange relationships among corporateleaders, bringing into question the cohesiveness of the innercircle. While we do not deny that macro-social forces suchas institutional shareholder activism may have played a rolein sparking greater board independence from management inlarge U.S. corporations, we suggest that the diffusion ofboard independence may have resulted from micro-mecha-nisms relating to social and psychological dynamics withinthe inner circle of corporate leaders.

Outside directors who are themselves CEOs of other largecorporations may play a pivotal role in determining whether aboard has a passive or an active orientation (Lorsch andMaclver, 1989: 18), and they may thus have a more complexrole in corporate governance than previously assumed in theliterature. Specifically, we propose that although CEO-directors may typically support fellow CEOs by impedingincreased board control over management, CEO-directorsexperiencing increased board control in their own corpora-tions may also experience a reversal in their perceived basisfor social exchange with other top managers. This change inthe norms of reciprocity underlying generalized exchangecould then lead CEO-directors to "defect" from the networkof corporate leaders. This study shows how these specificinstances of "defection" from a previously mutually support-ive network have diffused across organizations and overtime, and how this has created the current situation ofincreased board independence in large U.S. corporations.

When prior governance research has examined diffusionprocesses, it has been to establish the spread of mecha-nisms for increased managerial entrenchment such aspoison pills or golden parachutes (e.g., Davis, 1991, 1992;Westphal and Zajac, 1994). Governance researchers havenot considered how mechanisms for increased boardindependence or control nnay diffuse through an interorgani-zational network. This may be attributable to the fact thatmost researchers typically view board interlocks in terms ofa social cohesion model or "a general comnnunicationsystem" that "allows firms to learn about their businessenvironment" (Davis, 1992: 8). With respect to corporategovernance, network ties are thought to spread informationabout mechanisms that protect the inner circle of businesselites from market discipline (Useem, 1984). Increased boardindependence, however, can weaken rather than protect theinner circle, and thus its diffusion may not be easily ex-plained by traditional network perspectives.

In this study we develop a new perspective on network tiesamong corporate elites that emphasizes the generalized

162/ASQ, March 1997

Defections

social exchange relationships that exist between CEO-directors and top managers and how social and psychologi-cal factors may affect the dynamics of that relationship. Thisperspective suggests that generalized norms of reciprocityamong corporate leaders and changes in those norms canimpede or foster the diffusion of board independence. Wealso control for the conventional communication perspec-tives on diffusion described above, thus enabling us todistinguish support for the social exchange perspective ondiffusion developed in this study from alternative explana-tions. We then examine the generalizability of our frameworkto explain the diffusion of two other changes reflectingincreased board independence that have generally beenconsidered inimical to managerial preferences and thought tohave been driven by external pressures: decreased corporatediversification and increased compensation contingency. Wetest hypotheses using event history analyses of changes inboard structure, corporate strategy, and CEO compensationamong the largest U.S. corporations over a ten-year period.

There are several alternative theoreticalexplanations that are sometimes invokedin prior governance research to explainwhy boards would generally defer toCEOs. One is that outside directors feelindebted to a CEO who has offered thema directorship position and are thereforelikely to support the CEO (Main, O'Reilly,and Wade, 1995). Such a cooptation viewdiffers from ours in that it takes a strictlydyadic and intraorganizational focus (i.e.,a specific director owes a CEO for theboard appointment) and thus does notaddress the spread of greater or lesserboard control across firms.

THEORY

Reciprocity and Board Independence

Preventing increased board independence. One reasonwhy boards have traditionally been characterized as rubberstamps or tools of top management is that outside directorswho are also CEOs of their own firms may be reluctant tochallenge or criticize CEOs on whose boards they sit (Mace,1971; Lorsch and Maclver, 1989; Lawler, 1990). We proposethat generalized norms of reciprocity among CEOs who alsoserve as outside board members may represent a primary,social psychological mechanism hindering increased boardindependence. The principle of reciprocity refers to a rule ofbehavior in social exchange situations, and the more com-monly used phrase "norm of reciprocity" highlights thesocial obligation underlying the principle. Gouldner (1960)used the term to refer to the mutual reinforcement by twoparties of each other's actions, but as Ekeh (1974: 48) noted,generalized norms of reciprocity refer to the situation inwhich "an individual feels obligated to reciprocate another'saction, not by directly rewarding his benefactor, but bybenefiting another actor implicated in a social exchangesituation with his benefactor and himself." In such general-ized social exchange relationships, reciprocations are alsogeneralized, involving multiple actors (rather than just two)and indirect (rather than direct) benefits.^

In such a system of social interaction, A may help B,expecting reciprocation not from B but, rather, from anotheractor, such as C or D, sometime in the future. This impliesthat such a system relies on the existence of sufficient trustthat the giver believes he or she will be reciprocated bysomeone else, somewhere else in the future. L§vi-Strauss's(1969: 266) research on social exchange and reciprocitysuggests that "for a system [of generalized exchange] tofunction harmoniously," exchange partners need to be "ofequal status and prestige." In addition, social psychologicalresearch has shown that high levels of group solidarity orcohesion facilitate cooperative behavior among group

163/ASQ, March 1997

members (of,, Dawes, 1992) and that individuals benefittingfrom an altruistic or prosocial act are more likely to engagein similar behavior toward others, so that helping behaviorspreads through a social structure (Krebs and Miller, 1985;Komorita, Hilty, and Parks, 1991),

The CEO-director network is an arena likely to be character-ized by a generalized reciprocity among top managers andCEO-directors, First, given that corporate CEOs are arelatively homogenous and cohesive group (Useem, 1984),they may feel a generalized obligation to support fellowCEOs in board meetings and during periods of poor firmperformance (Mace, 1971), Thus, CEOs enjoying minimalboard independence at their home companies may, asCEO-directors, hinder any attempts to increase the indepen-dence of other boards on which they sit. In the context ofthe present study, the preceding discussion suggests thefollowing proposition;

Proposition 1: The greater the proportion of CEO-directors on aboard, the lower the likelihood of an increase in board indepen-dence.

The diffusion of increased board independence. While thediscussion thus far has emphasized how generalizedreciprocity can result in cooperative behavior among individu-als, other possible outcomes, such as mutual noncooperativebehavior or defection, are also possible (Axelrod, 1984),Similarly, while social solidarity may promote generalizedreciprocal cooperation among individuals, it can likewiseprovide a context for generalized reciprocal "retaliation,"defined broadly as the repayment of injurious or otherwiseundesired acts, regardless of emotional content (i,e,,retaliation simply represents one category of exchangereciprocity) (Gouldner, 1960; 172; Walster, Berscheid, andWalster, 1973), Where CEO-directors have experienced anincrease in board independence at their home companies,their felt obligation to support fellow CEOs (i,e,, by resistingincreased independence) may diminish or even reverseitself. When generalized exchange partners are no longer ofequal status and prestige, such inequality may "force arupture" in the system of exchange (Levi-Strauss, 1969;266), As Ekeh (1974; 48) noted; "the principle of reciprocityhas a built-in change-generating factor," since equality insocial exchange is needed for continuity of social interaction;when this expectation is frustrated, the social exchangesituation is threatened. Given that a CEO's power andinfluence is closely tied to his or her status among CEOs, aCEO-director who has lost power to the board in his or herhome company may perceive himself or herself as unequalin status to those CEOs who enjoy greater control over theirboards and thus greater influence over strategy, compensa-tion, or perquisites. While disempowered CEOs mayalleviate the resultant inequality in several ways, oneresponse is to diminish their support for CEOs in otherorganizations where they serve as outside directors, thusreducing the power and status of fellow corporate leadersand returning balance to the social exchange relationship.

The preceding discussion suggests that CEO-directorssubjected to increased board independence at their homecompanies may not only decrease their resistance to greater

164/ASQ, March 1997

Defections

board control but may actually induce such changes in otherboards on which they sit. Thus, while CEO-directors maytypically help sustain CEO dominance over the board,effectively preventing greater board independence, they mayspur increased independence when they have experienced itthemselves, resulting in "chains of negative reciprocity" orretaliation (Collins, 1990: 419). In this way, increased boardindependence can spread through the director network ofcorporate leaders. As the prevalence of director experiencewith greater board independence increases, the board'sorientation should tend toward greater independence andactivism rather than passivity. This suggests the followingproposition:

Proposition 2: The greater the proportion of CEO-directors on aboard who have experienced an increase in board independence attheir home connpanies, the greater the likelihood of such change inthe focal company.

Hypotheses

Reciprocity and change in board structure. While it isdifficult to observe and measure directly the level of boardindependence in large corporations, a number of structuralvariables have been used in the governance literature toassess differences in levels of board independence acrossorganizations. We consider three such variables. Managerialhegemony theorists, agency theorists, and the businesspress have generally all argued that separation of the CEOand board chair positions (i.e., allocation of each position toseparate individuals) enhances the board's independentmonitoring capacity (Crystal, 1991; Beatty and Zajac, 1994;Finkelstein and D'Aveni, 1994). In contrast, CEOs holdingboth positions are thought to possess greater formalauthority and heightened informal power over board mem-bers (Harrison, Torres, and Kukalis, 1988). More generally,given that the board chairman is responsible for monitoringand evaluating CEO decision making, uniting both roles inone person represents a formalized conflict of interest.Similarly, separating these roles indicates an increase inboard independence. This logic suggests the followinghypotheses, consistent with propositions 1 and 2:

Hypothesis 1 (HI): The greater the proportion of CEO-directors ona board, the lower the likelihood of a separation of the CEO andboard chair positions in the firm.

Hypothesis la (Hia): The greater the proportion of CEO-directorson a board who have experienced a separation of the CEO andboard chair positions at their home companies, the greater thelikelihood of a separation of the CEO and board chair positions inthe firm.

Agency theorists and advocates of board reform have alsolong maintained that outside directors are better able tocontrol management decision making (Zahra and Pearce,1989). Since insiders are "beholden to CEOs for their jobs"(Fredrickson, Hambrick, and Baumrin, 1988: 262), they maybe less willing to challenge or override CEO initiatives whenshareholder interests are threatened (Fama and Jensen,1983; Kosnik, 1987). Outsiders may also be better positionedto evaluate managerial performance impartially (Beatty andZajac, 1994). Thus, increases in the ratio of outside to insidedirectors can be viewed as reflecting increased board

165/ASQ, March 1997

independence from management. This suggests the follow-ing hypotheses;

Hypothesis 2 (H2): The greater the proportion of CEO-directors ona board, the lower the likelihood of increases in the ratio of outsideto inside directors.

Hypothesis 2a (H2a): The greater the proportion of CEO-directorson a board who have experienced increases in the ratio of outsideto inside directors at their home companies, the greater thelikelihood of such an increase in the firm's board.

While the two manifestations of board independencediscussed above are based on the prevalence and allocationof formal board roles, independence may also be enhancedor reduced through more subtle, demographic mechanisms.Abundant evidence suggests that demographic similarlyproduces bias in evaluation decisions. For instance, Wayneand Liden (1995) showed how demographic similarityenhanced mutual affect in superior-subordinate dyads,resulting in more positive performance appraisals (see alsoTurban and Jones, 1988), By contrast, demographic dissimi-larity between board members and top managers shouldminimize bias and enhance independence and objectivity indecision control activities. Recent empirical evidenceindicates that greater demographic distance between theCEO and the board can increase the tendency for directorsto challenge managerial preferences on behalf of shareholderinterests (Westphal and Zajac, 1995), Accordingly, theappointment of demographically dissimilar new directorsshould enhance the board's attitudinal and social indepen-dence from management. In general, therefore, CEO-directors who have experienced a reduction in demographicsimilarity with the board at their home companies shouldundergo changes in their perceived social exchange relation-ships with CEOs of other companies on whose boards theysit who have not experienced this reduction. This suggeststhe following hypotheses:

Hypothesis 3 (H3): The greater the proportion of CEO-directors ona board, the lower the likelihood of an increase in demographicdistance between the CEO and the board.

Hypothesis 3a (H3a): The greater the proportion of CEO-directorson a board who have experienced an increase in demographicdistance from the board at their home companies, the greater thelikelihood of such an increase in the firm's board.

Reciprocity and unrelated diversification. Social exchangeprocesses may also influence the diffusion of other changesthat are thought to reflect greater board independence andthat have generally been considered aversive to top manage-ment, including reduction in unrelated diversification andgreater compensation contingency. The pursuit of diversifica-tion as a corporate strategy has been the subject of increas-ing debate. According to both managerialist and agencyresearchers, top managers have personal incentives topursue diversification beyond the level at which shareholderwealth is maximized (Hill and Snell, 1988; Jensen, 1989;Baysinger and Hoskisson, 1990), From these perspectives,top managers may diversify into largely unrelated businessesto enhance their personal power, compensation, and status(Marris, 1964), while stabilizing income streams and minimiz-ing employment risk (Amihud and Lev, 1981), Since stock-

166/ASQ, March 1997

Defections

holders can diversify their investment portfolios more easilythan CEOs can diversify their employment, shareholdersshould favor lower levels of unrelated diversification. Onewould expect that boards of directors, as representatives ofshareholder interests, would resist management's attemptsto diversify into largely unrelated businesses and wouldencourage the reduction of existing levels of diversificationwhere possible.

Jensen (1989) has suggested that passive boards have beenunwilling to question or resist top management's prefer-ences for unrelated diversification. From our perspective,such passivity may be understood in terms of the existenceof generalized norms of reciprocity among corporate leaders.To the degree that CEO-directors frequently empathize withthe focal CEO's desire to minimize employment risk andstabilize income, while also perceiving a general obligation tosupport the preferences of fellow CEOs, the presence ofCEO-directors on the board may inhibit board-initiated effortsto minimize or reduce unrelated diversification. At the sametime, however, CEOs who have been forced into reducingunrelated diversification, thus sacrificing their personalpreferences for risk reduction, power, and status, may seekto rectify perceived inequities by reducing their support forunrelated diversification or even actively advocating divesti-ture in those firms on whose boards they sit as directors. Asa result, while CEO-directors may, under normal circum-stances, typically help sustain strategies of unrelateddiversification, they may foster reductions in it when theyhave experienced it themselves. Taken together, andparalleling the discussion of board independence, wepropose the following hypotheses:

Hypothesis 4 (H4): The greater the proportion of CEO-directors on aboard, the lower the likelihood of a decrease in unrelated diversifi-cation.

Hypothesis 4a (H4a): The greater the proportion of CEO-directorson a board who have experienced a reduction in unrelated diversifi-cation at their home companies, the greater the likelihood of suchchange in the focal company.

Reciprocity and compensation contingency. Contingentcompensation contracts that link pay to firm performanceserve to align the interests of CEOs as agents with thepreferences of shareholders as principals, thus promotingshareholder wealth (Jensen and Murphy, 1990). From anagency theory perspective, the provision of long-termincentives such as stock options, performance shares, orrestricted stock is a primary mechanism by which corporateboards effect incentive alignment (Kerr and Kren, 1992;Gibbs, 1993; Beatty and Zajac, 1994). But a significantnumber of large U.S. corporations make only limited use oflong-term incentive compensation in designing CEO compen-sation contracts (Jensen and Murphy, 1990), which West-phal and Zajac (1994) have attributed to managerial entrench-ment. From a normative agency theory perspective, CEOs asrisk-averse agents prefer less risk in their compensationcontracts (Harris and Raviv, 1979). By making compensationcontingent on future firm performance, long-term incentivesadd uncertainty to a CEO's compensation. Moreover,stock-based compensation effectively increases the CEO's

167/ASQ, March 1997

non tradeable investment in the firm and reduces thediversification of his or her investment portfolio (Beatty andZajac, 1994),

To the extent that generalized norms of reciprocity amongcorporate leaders constitute a relatively fundamental sourceof CEO power over the board (as suggested above), havingsympathetic CEO-directors on the board may be a primarymechanism by which CEOs avoid contingent compensation,thus inhibiting increased incentive alignment. By contrast,just as CEOs experiencing a decrease in unrelated diversifi-cation at their home companies may reduce their oppositionto those changes elsewhere, CEO-directors subjected togreater compensation contingency may likewise reduce theirresistance to incentive compensation at other companies.Thus, norms of retaliation among CEO-directors mayfacilitate the spread of incentive alignment through thedirector network of corporate leaders. This suggests thefollowing hypotheses;

Hypothesis 5 (H5): The greater the proportion of CEO-directors ona board, the lower the likelihood of an increase in compensationcontingency.

Hypothesis 5a (H5a): The greater the proportion of CEO-directorson a board who have experienced an increase in compensationcontingency at their home companies, the greater the likelihood ofsuch a change in the focal firm,

METHOD

Sample and Data Collection

The population for this study included the largest U,S,industrial and service firms, as listed in the 1982 Forbes andFortune 500 indexes The Forbes 500 uses multiple listswhose overlap depends on the specific size measure used;we included those firms that qualified according to two ormore size measures. Firms were excluded from the finalsample if complete demographic data were unavailable formore than one-quarter of the outside directors in each yearor if complete diversification or compensation data wereunavailable. This procedure yielded 422 companies, T-testsrevealed no significant differences in size, measured as logof sales, or profitability between the initial and final samples.

Data were collected for the years 1982 to 1992, inclusive.Academic research and anecdotal evidence suggest thatincreases in board independence, as manifested by control-enhancing changes in board structure, greater compensationcontingency, and reductions in corporate diversification, mayhave accelerated during this period (Davis and Thompson,1994; Westphal and Zajac, 1994), While the spread of suchchanges has been attributed to increased activism byinstitutional investors during the 1980s, we analyze theeffect of social exchange factors internal to the network ofcorporate leaders during this period.

Information on CEO and board member characteristics,including the affiliations of CEO-directors, was obtained fromthe Dun and Bradstreet Reference Book of CorporateManagement, Standard and Poor's Register of Corporations,Directors, and Executives, and Who's Who in Finance andIndustry; data on board structure and compensation were

168/ASQ, March 1997

Defections

collected from proxy statements. Finally, we obtaineddiversification data from Standard & Poor's COMPUSTATBusiness Segment Tapes; size and performance data wereprovided by COMPUSTAT and the Center for Research inSecurity Prices (CRSP).

Dependent Variables

Board structure. To analyze the likelihood of a separation ofthe CEO and board chair positions [separation), we created adichotomous variable, coded as 1 in a given year if thepositions were separate in that year but not in the prior year,and 0 otherwise. Separation of the CEO and board chairpositions often occurs when the CEO is replaced (Vancil,1987; Harrison, Torres, and Kukalis, 1988). As a result, aCEO-director's experience with separation may involve (1)having lost the CEO position as well as the board chairposition, (2) having lost only the CEO position and remainingas board chair, or (3) having lost only the board chair positionand remaining as CEO. In all three scenarios, however, theCEO witnesses separation of the two positions, experiencesthe associated increase in board power, and loses the statusassociated with holding both positions. Thus, we combinedall three varieties of separation experience. A separateanalysis indicated that when cases of separation accompany-ing CEO succession are removed (scenarios 1 and 2 above),the results reported below remain unchanged.

Increases in the ratio of outside to inside directors [increasedoutsider ratio) were also measured dichotomously, coded 1if the ratio of outside to inside directors was greater in thecurrent year than in the prior year, and 0 otherwise.

To analyze the likelihood of increased demographic distancebetween the CEO and the board [increased dissimilarity), wefirst measured CEO-board similarity across several dimen-sions for each year. Similarity was assessed in terms ofthree characteristics commonly employed in the top man-agement team literature (Hambrick and Mason, 1984).Functional background similarity and education similaritywere measured with a variant of Blau's (1977) index ofheterogeneity, defined as [P)^, where P,- is the proportion ofCEO-board member dyads sharing the rth category (Murray,1989). For functional background, this measure indicates thesquared proportion of CEO-board member dyads in whichboth individuals have primary experience in the same core,functional area (Hambrick and Mason, 1984); for educationalbackground, it represents the proportion of such dyads inwhich both individuals possess the same type of degree.Following prior research (Wiersema and Bantel, 1992), wemeasured degree type as the highest university degreeobtained in five categories: arts, sciences, engineering,management or economics, and law. When specialties werenot listed, B.S. and M.S. degrees were classified as sciencedegrees. Age similarity was measured with an analog of thecoefficient of variation (Tsui and O'Reilly, 1989):

where S, is the CEO's age, S,- indicates the age of director /

169/ASQ, March 1997

(excluding the CEO), and n represents the number of boardmembers. This measure was converted to an indicator ofsimilarity by subtracting each firm's coefficient from thehighest value in the sample. We then created a dichotomousmeasure of increased demographic distance, coded 1 ifCEO-board similarity decreased across all three dimensions.The results were also unchanged using increased similarityalong one or more dimensions. To the extent that each ofthe three demographic characteristics indicates differentattitudes and behavioral tendencies (Hambrick and Mason,1984), this composite measure indicates relatively broadlybased increases in demographic distance between the CEOand the board.

While it is possible to use continuous rather than dichoto-mous measures of some of the dependent variables, wechose to use dichotomous measures for two reasons; (1)our theoretical argument implies that the increase in boardindependence itself (i,e,, even a small one) is a meaningfulevent, since it requires a fundamental change in howCEO-directors perceive the CEO-board relationship; and (2)the range of size in such increases is very restricted formost of our dependent variables and highly correlated withthe dichotomous measures. For instance, boards rarelyincrease the outsider ratio by adding more than two individu-als to the board, so that the magnitude of such increases isvery similar across cases.

Compensation contingency. Compensation contingencywas calculated for each year as the total value of long-termincentive grants divided by total direct compensation. Stockoptions were valued using the Black-Scholes (1973) method,which estimates option value based on the historical pricevolatility of the underlying security. Other grants (e,g,,restricted stock, performance shares) were valued accordingto the market price at date of grant (Crystal, 1984), Allcompensation values were adjusted for inflation to represent1990 constant dollars using the Consumer Price Index, Toanalyze the likelihood of increased compensation contin-gency (increased contingency), we created a dichotomousmeasure, coded 1 in a particular year if compensationcontingency increased from the prior year, and 0 otherwise.

Diversification, To test hypotheses on change in unrelateddiversification, we used Palepu's (1985) entropy measure,which takes into account the number of segments in whicha firm operates and weights each segment according to itscontribution to total sales. It is defined as follows;

where P is the dollar value of sales attributed to segment /and In(1/P,) is the weight for each segment i, or the loga-rithm of the inverse of its sales. Reduced diversification wasoperationalized as an absolute decrease in the entropymeasure of ,10 or more, to capture relatively significantchange and to exclude change that reflects random alter-ations in segment sales levels. The results were similarwhen we operationalized reduced diversification as anabsolute decrease in the entropy measure of ,05 or more,

170/ASQ, March 1997

Defections

Independent Variables

The proportion of outside directors who were CEO-directorswas assessed simply as the number of outside directorscurrently then also serving as chief executive at anotherFortune or Forbes 500 company, divided by the total numberof outsiders on the board. To measure the proportion ofoutside CEO-directors who had experienced increased boardindependence in their home companies, we created fiveindependent variables, three for board structure and oneeach for diversification and compensation contingency. Eachindicates the number of outside CEO-directors serving asCEO at another Fortune or Forbes 500 company who havealso experienced change along the relevant dimension withinthe past two years, divided by the total number of outsiderson the board. The results were generally robust to differentlags, such as one year or three years.

In separate analyses we operationalized the presence ofCEO-directors as the absolute number of such directors onthe board, rather than as the proportion of outsiders. Theresults were substantively unchanged. Because our argu-ment implicity compares the social obligations of CEO-directors to those of other outsiders, we used the propor-tional measure in the final models presented below.

Controls

We controlled for the possibility that more traditional networkdiffusion processes involving imitation or social learningcould influence the diffusion of increased board indepen-dence (e.g., Galaskiewicz and Wasserman, 1989; Davis, 1991;Haunschild, 1993; Palmer, Jennings, and Zhou, 1993).According to this perspective, organizations are more likelyto adopt an innovation when they have director ties to prioradopters, because such contacts help resolve uncertaintyabout the consec|uences of adoption (Davis, 1991). As notedby Palmer, Jennings, and Zhou (1993), directors who haveparticipated in a particular structural, strategic, or policychange at other firms may become more forceful advocatesof such change because they have behaviorally committedthemselves to it. These directors may also be more persua-sive in advocating change because they can provide vivid,firsthand accounts of how the innovation affected organiza-tional processes and outcomes (Davis, 1991). Overall,according to this perspective, director ties serve as ''informa-tion conduits" that facilitate imitation and learning, largelybecause information provided by fellow corporate leadersmay be trusted more than information derived from othersources (Davis, 1991: 594; Haunschild, 1993; Mizruchi,1996). This would suggest that CEO-directors' experiencewith increased board independence at other firms mayincrease the likelihood of such change at the focal firmsimply by providing information that encourages imitation.

Thus, to distinguish the social exchange explanation fordiffusion from this more traditional explanation, we controlledfor the prior experience CEO-directors had with increases inboard independence while serving as outside directors onother boards. Specifically, we developed a set of variables toindicate the number of outside CEO-directors who experi-enced change along the relevant dimension as an outside

171/ASQ, March 1997

director at a Fortune or Forbes 500 company other than theirhome company within the previous two years, divided bythe total number of outsiders on the board. Significanteffects for these variables would provide evidence consis-tent with the typical imitation or learning perspective ondiffusion through interlocks. The absence of significanteffects for these variables, together with support for thehypotheses, would provide evidence consistent with thesocial exchange perspective on diffusion developed in thisstudy.

Because pressure from institutional investors is commonlyconsidered a primary external determinant of increasedboard control and reduced diversification, as discussedabove, we included institutional investor ownership as acontrol variable in all models (Bethel and Liebeskind, 1993).Institutional ownership was defined as equity held bypension funds, banks and trust companies, savings andloans, mutual fund managers, and labor union funds dividedby total common stock (Hansen and Hill, 1991).

There is some research suggesting that poorly performingorganizations are more likely to separate the CEO and boardchair positions (Harrison, Torres, and Kukalis, 1988) and toappoint outsiders to the board (Hermalin and Weisbach,1988). Prior firm performance may also be related tosubsequent changes in compensation contingency (Westphaland Zajac, 1994) and diversification (Wiersema and Bantel,1992). Thus we included two measures of firm performance(return on assets and excess stock returns) as controlvariables in all analyses.

In addition, because there is some evidence for relationshipsbetween firm size and various dimensions of board structure(Beatty and Zajac, 1994; Finkelstein and D'Aveni, 1994),diversification (Bergh, 1995), and compensation contingency(Beatty and Zajac, 1994), we included log of sales in allanalyses. Also, since changes in board structure may affectchanges in compensation contingency and diversification, wecontrolled for changes in board structure in these specificanalyses. Finally, we also controlled for industry and timeeffects by including dummy variables indicating primary,two-digit Standard Industrial Classification codes, as well asdummies for each year in all analyses. To conserve space,coefficients for these variables are not reported in the tabies.

Analysis

Hypotheses were tested using discrete-time event historyanalysis (Allison, 1982, 1984; Yamaguchi, 1991). Eventhistory models are especially appropriate for analyzinglongitudinal data with time-varying covariates when thedependent variable is a discrete event. In this case, theevents of interest are (1) control-enhancing changes in boardstructure (measured by separation of the CEO and boardchair positions, increases in the ratio of outside to insidedirectors, and increased demographic distance between theCEO and the board); (2) increases in compensation contin-gency; and (3) decreases in corporate diversification. Thuswe developed five models. Changes in board structure,compensation contingency, and diversification were ob-served from 1983 to 1992, yielding 4,220 firm-years of data.

172/ASQ, March 1997

Defections

The discrete-time model can be expressed in the following,logistic regression form (Allison, 1984):

= awhere P,U) is the probability of change in CEO characteristicsor increased CEO-board similarity in year M, Xn^ aretime-varying independent variables hypothesized to influencethe risk or likelihood of change, and b^ are the estimatedcoefficients. P,- is defined as:

such that R,{t) increases monotonically with b^,(,(M) and canassume any value between zero and one. Given time-varyingcovariates, a firm's likelihood of change is updated over timeas the values of independent and control variables change.As the model indicates, all independent and control variableswere lagged by one year.

Since the dependent variables of interest represent relativelyfundamental and long-lasting alterations in board orientationtoward the CEO, we model only the likelihood of the firstevent during the time period (i.e., the firm is removed fromthe risk set following adoption of the changes). This isconsistent with our interest in explaining when and whyboards began to assert their independence from top man-agement; subsequent continued increases in board indepen-dence are less central to the study, and we assume theyrepresent continuance of the board's new orientation. Also,significant reversals across the five dependent variables inthis study were quite rare. Moreover, for the model ofCEO-board chair separation, the initial risk set excludescases for which the two positions are already separate.Finally, to ensure that the results were not dependent onunspecified, time-specific factors, we included year dummyvariables in all models (Allison, 1982, 1984), as noted above.

In general, discrete-time models provide an adequateapproximation of continuous-time models, which estimateinstantaneous rates of change, when the conditional prob-ability of event occurrence is small, e.g., 0.1 or smaller(Clogg and Eliason, 1987). In this study, the likelihood ofchange in any given year is less than .1 for all dependentmeasures.

RESULTS

Table 1 provides the means, standard deviations, andbivariate correlations for all data pooled. The results of eventhistory analyses of changes in board structure are shown inTable 2. HI , H2, and H3 predicted a negative associationbetween the proportion of the board composed of CEO-directors and the likelihood of control-enhancing changes inboard structure. In general, the results support thesehypotheses. Consistent with HI , the proportion of the boardcomposed of CEO-directors is significantly and negativelyrelated to the likelihood of separation of the CEO and boardchair positions. H3 is also supported: A greater presence ofCEO-directors on the board significantly diminishes thelikelihood of a decrease in demographic similarity betweenthe CEO and the board. Finally, the results provide somesupport for H2, in that the proportion of the board composed

173/ASQ, March 1997

of CEO-directors is negatively related to increases in theratio of outside to inside directors.

Table 1

Descriptive Statistics and Pearson Correiation Coefficients {N = 4220)

Variables

1. CEO-directors on board2. CEO-directors experiencing

change at home company:(a) Separation(b) Increased dissimilarity(c) Increased outsider ratio(d) Increased contingency(e) Reduced diversification

3. CEO-directors experiencingchange as outside director:(a) Separation(b) Increased dissimilarity(c) Increased outsider ratio(d) Increased contingency(e) Reduced diversification

4. Separation5. Increased dissimilarity6. Increased outsider ratio7. Increased contingency8. Reduced diversification9. Return on assets

10. Excess stock returns11. Log of sales12. Institutional ownership

Variables

3. CEO-directors experiencingchange as outside director:(d) Increased contingency(e) Reduced diversification

4. Separation5. Increased dissimilarity6. Increased outsider ratio7. Increased contingency8. Reduced diversification9. Return on assets

10. Excess stock returns11. Log of sales12. Institutional ownership

Mean

.376

.068

.071

.081

.090

.082

.147

.167

.166

.219

.182

.041

.044

.064

.089

.060

.068

.0137.8141

.356

3c

.07

.11

.03

.04

.09-.01

.07-.03-.06-.02

.01

S.D.

.245

.166

.161

.157

.194

.162

.184

.172

.170

.206

.175

.198

.205

.245

.286

.238

.050

.135

.224

.208

3d

.13

.05

.03

.06

.06

.01

.01-.02

.03-.01

1

.27

.23

.34

.22

.28

.22

.29

.28

.26

.33-.22-.20-.08-.11-.26-.14-.15

.07-.08

3e

.06

.05

.08

.07

.12

.00

.04

.08-.04

2a

.21

.12

.19

.33

.16

.13

.10

.13

.05

.25

.14

.05

.12

.15

.07

.03-.02

.23

4

.19- . 1 4 •

.22

.24

.09

.13

. 0 2 •

.09

2b

.13

.20

.15

.08

.22

.11

.12

.09

.10

.34-.13

.12

.07

.10

.08

.03

.12

5

-.19.15.18.04.12

-.03.13

2c

-.29-.25

.07

.13

.19

.11

.11

.02-.01

.18-.05-.08-.04-.09

.03-.02

6

-.25-.20-.06-.08

.09

.16

7

.11

.07

.17

.00

.08

2d

.12

.12

.10

.08

.16

.12

.11

.06-.09

.24

.13

.10

.07-.05

.19

8

.05

.08

.18

.20

2e

.10

.13

.12

.10

.18

.09

.07-.13

.14

.32-.03-.11

.05

.15

9

.32-.11

.07

3a

.09

.13

.14

.17

.08

.07-.02

.03

.02

.01

.02-.01

.00

10

-.14-.03

3b

.10

.16

.14

.06

.05

.04

.02

.03

.04

.02-.01-.03

11

.15

HI a, H2a, and H3a predicted a positive relationship betweenthe proportion of the board composed of CEO-directors whoexperienced control-enhancing changes in board structure attheir home companies and the likelihood of such changes atthe focal company. The results shown in Table 2 providestrong and consistent evidence for these hypotheses.Consistent with HI a, separation of the CEO and board chairpositions is more likely in the presence of CEO-directorswho have experienced separation themselves. The findingsalso afford strong evidence for H3a: The larger the propor-tion of the board composed of CEO-directors who experi-enced an increase in demographic distance from the CEO athome, the greater the likelihood of such change at the focalcompany. Finally, the results also support H2a: Increases inthe ratio of outside to inside directors are more likely whenthe board includes CEO-directors who have experienced

174/ASQ, March 1997

Defections

Table 2

Event History Analyses of Changes in Board Structure*

Unstandardized Coefficients

independent variabies

1. CEO-directors on board

2. CEO-directors experiencingchange at honne company

3. CEO-directors experiencingchange as outside director

4. Return on assets

5. Excess stock returns

6. Log of sales

7. Institutional ownership

Constant

Chi-squareNumber of firm-years

Separation

-4 .100-(1.938)1.918—(.355)

-.319(.570)

- 4 . 9 3 4 —(1.236)- .089—(.034).064

(.124)1.230*(.723)

2.548-*(1.013)

109.12**—3170

Dissimilarity

- 2 . 2 5 1 -(1.062)1.274*-*(.239).189

(.403)-4.523*-*(1.025)-.023(.028).057

(.115).543

(.411)4.036*—*(.955)

97.44*-*3801

Outsider ratio

-1.196*(.878).593-

(.346)-.231(.354)

-2.333*(1.391)- . 0 4 9 -(.022).187*

(.094).446

(.355)4 .532—(.767)

38.34**-3555

*p< .10;**p< .05;-*p< .01;**-p< .001.• Coefficients for industry and year dummy variables are not reported. Standard errors are in parentheses. T-tests areone-tailed for hypothesized effects, two-tailed for control variables.

such a change at their home companies. In general, there-fore, while the presence of CEO-directors typically preventsan increase in board independence, the presence of CEO-directors who experienced an increase in independence at

Table 3

Event History Analysis of Reduced Diversification*

independent variabies Unstandardized coefficients

1. CEO-directors on board

2. CEO-directorsexperiencing change

3. CEO-directors experiencingchange as outside director

4. Return on assets

5. Excess stock returns

6. Separation

7. Dissimilarity

8. Outsider ratio

9. Log of sales

10. Institutional ownership

Constant

Chi-squareNumber of firm-years

-2.499(.730)*-1.871(.233)*--.312(.280)

-1.529(1.119)-.032(.023).797(.241)*-.514(.293)*1.019(.526)*.018(.097).758(.377)-3.754(.767)*-

108.87-**3639

*p< .10;**p< .05;-*p< .01;*-*p< .001.* Coefficients for industry and year dummy variables are not reported. Stan-dard errors are in parentheses, f-tests are one-tailed for hypothesized effects,two-tailed for control variables.

175/ASQ, iViarch 1997

their home companies actually stimulates such a change atthe focal company.

The next set of analyses show whether the presence ofCEO-directors has a comparable effect on two other ex-amples of changes in large U.S. corporations: a decrease inunrelated diversification and an increase in compensationcontingency. Table 3 provides the results of event historyanalysis of a decrease in diversification. In support of H4, thepresence of CEO-directors on the board lowers the likelihoodof a reduction in diversification. The findings also supportH4a: The presence of CEO-directors enduring reduceddiversification at home enhances the likelihood of suchchange at the focal company. The results of event historyanalysis of an increase in compensation contingency areprovided in Table 4. Consistent with H5, the proportion ofthe board composed of CEO-directors is negatively related tothe likelihood of an increase in compensation contingency.Moreover, the findings afford strong support for H5a: Thegreater the proportion of the board composed of CEO-directors who experienced an increase in compensationcontingency at their home companies, the greater thelikelihood of such a change in the focal company.

Finally, the results also indicate that variables included tocontrol for more traditional perspectives on interlock diffu-sion are consistently nonsignificant. The results in Table 2show that a CEO-director's experience with control-enhanc-ing changes in board structure as an outside director on

Table 4

Event History Analysis of Increased Compensation Contingency*

Independent variables Unstandardized coefficients

1. CEO-directors on board -.906(.603)*

2. CEO-directors 1.114experiencing change (.203)****

3. CEO-directors experiencing -.120change as outside director (.246)

4. Return on assets -.829(1.029)

5. Excess stock returns -.045(.022)~

6. Separation .461(.227)-

7. Dissimilarity .659(.274)-

8. Outsider ratio .417(.371)

9. Log of sales -.087(.076)

10. Institutional ownership .520(.306)*

Constant 3.500(.638)—

Chi-square 92.21 —Number of firm-years 3353

•p< .10;-p< .05;—p< .01;—p< .001.* Coefficients for industry and year dummy variables are not reported. Stan-dard errors are in parentheses. T-tests are one-tailed for hypothesized effects,two-tailed for control variables.

176/ASQ, March 1997

Defections

Other boards is not significantly related to such a change atthe focal firm. Similarly, as shown in Tables 3 and 4, thelikelihood of a reduction in diversification or an increase incompensation contingency is not independently affected bythe presence of CEO-directors who have experienced such achange as outside directors at other firms. Overall, then, theresults provide strongly consistent evidence that the diffu-sion of increased board independence and control, asmanifested by specific changes in board structure, greatercompensation contingency, and reduced diversification, isaffected by CEO-directors' experience at the CEOs' homecompanies, consistent with the social exchange perspective,but not by CEO-directors' experience as outside directors onother boards, contrary to a traditional network diffusionperspective,

DISCUSSION

The findings provide strong evidence that control-enhancingchanges in board structure, as well as related changes incorporate strategy and executive incentive compensation,are influenced by social and psychological dynamics operat-ing within the inner circle of corporate leaders. Empiricalanalyses yielded a highly consistent pattern of resultssupporting the theoretical proposition that, although general-ized norms of reciprocity among corporate elites maytypically lead CEO-directors to inhibit organizational changesthat are contrary to the preferences of fellow top managers,CEO-directors actually induce such change when they haveexperienced it themselves. Thus, it appears that the socialexchange framework can explain the specific conditionsunder which a variety of phenomena that are consideredaversive to CEOs can spread through the network ofCEO-directors,

The first set of findings revealed a negative relationshipbetween the proportion of the board composed of CEO-directors and the likelihood of changes in board structure.This suggests that, consistent with a social exchange modelemphasizing generalized norms of reciprocity, CEO-directorsperceive a generalized obligation to support other CEOs, Inthis system of social interaction among status equals,sufficient trust exists that CEO-directors believe theirsupport for CEOs will be reciprocated indirectly by someoneelse, sometime in the future. This finding is consistent withsocial psychological research showing that group solidarity ormutuai identification can generate cooperative behavior(Dawes, 1992) and that individuals benefiting from prosocialacts frequently engage in similar behavior toward generalizedothers, causing helping behavior to spread through a socialstructure (e,g,, Krebs and Miller, 1985; Komorita, Hilty, andParks, 1991), Thus the present study provides a theoreticalexplanation and large-sample empirical evidence for howboard norms emphasizing deference to the CEO that arecommon across firms (Whisler, 1984) can be maintainedover time.

At the same time, however, our additional analyses indicatedthat the relative prevalence of directors who experienced an

177/ASQ, March 1997

increase in board independence actually increased thelikelihood of such a change at the focal firm, where theyappeared to spur, rather than resist, such change. This resultis consistent with the notion that when generalized ex-change partners are no longer of equal status and prestige,such inequality may "force a rupture" in the system ofexchange (Levi-Strauss, 1969: 266), so that social solidarityamong corporate leaders can provide a context for general-ized retaliation or defection, rather than cooperation (cf.Gouldner, 1960; Axelrod, 1984). The underlying socialpsychological mechanisms for such defection can also beunderstood from an equity or social comparison perspective(Festinger, 1954; Walster, Berscheid, and Walster, 1973:152), wherein recently disempowered CEOs perceive theirstatus as having worsened relative to comparison-otherCEOs and thus act to reduce their support of other CEOs onwhose boards they sit. Overall, the results suggest that thesocial and psychological factors relating to the CEO-directors'perceived exchange relationship with top managers canpredict and explain the diffusion of board independence iniarge U.S. corporations.

Further analyses demonstrate the generalizability of ourproposed framework for understanding the diffusion ofdefection among a connected set of previously cooperativeactors. The findings indicate that while the presence ofCEO-directors typically reduces the likelihood of a decreasein unrelated diversification, the presence of CEO-directorswho have experienced a decrease in diversification at theirown firms increases the likelihood of such a change at thefocal company. Furthermore, a very similar pattern of resultsemerges in analyses of increases in compensation contin-gency. Thus defection or retaliation among corporate elitesappears to facilitate the spread of reduced diversification andincentive alignment (both of which are manifestations ofgreater board independence), as CEO-directors subjected tosuch pressures in their home companies diminish theirresistance or even actively support such change elsewhereso as to restore balance to their social exchange relation-ships.

Additional results provided further evidence for this interpre-tation by allowing us to rule out alternative explanations.While experience with increased board independence at theCEO-directors' home companies consistently predicted thelikelihood of change at the focal firm, we found that CEO-directors' experience as outside directors on other boardswas consistently unrelated to such change (across fivedifferent indicators). This finding is inconsistent with alterna-tive, more conventional perspectives on network diffusionemphasizing the role of imitation and social learning (e.g.,DiMaggio and Powell, 1983; Galaskiewicz and Wasserman,1989; Davis, 1991; Haunschild, 1993; Palmer, Jennings, andZhou, 1993). The diffusion of control-enhancing change inboard structure does not appear to result simply from thespread of information and awareness about such changethrough director ties to other boards.

The results also suggest that diffusion cannot be describedhere as a process in which directors who participated inincreasing board control on other boards became socialized

178/ASQ, March 1997

Defections

into believing that such change is effective, leading them topush for it at other companies (cf, Burt, 1987), Instead,these additonal findings are consistent with our generalizedsocial exchange perspective, according to which CEO-directors would not generally seek to spread changes thatare aversive to CEOs' interests, even if their prior exposureto such changes on other boards led them to believe thatsuch changes might be desirable for other constituents.Rather, CEO-directors will spread such changes only as aresult of their own experiences as CEOs, which lead themto restore balance to social exchange relationships withfellow corporate leaders that have been disturbed by theCEOs' loss of control over their own boards. Moreover, ourfocus on CEO-directors enables a precise comparison ofhome company vs, outside director experience, since anyobserved differences cannot be attributed to differences inthe kind of director studied.

We also conducted a separate test of the learning-basedalternative explanation for our results. Conventional perspec-tives on diffusion might suggest that directors experiencedwith increased board independence might spread suchpractices to improve firm performance where it is poor. Totest this notion, we estimated supplementary models thatincluded interactions between CEO-director experience andprior performance of the focal company. The results of theseanalyses were consistently insignificant and thus did notsupport the alternative explanation. These results providestronger evidence that the observed effect of CEO-directorexperience reflects the proposed social exchange factors,rather than economic considerations or learning.

Our findings stress the importance of social and psychologi-cal forces within the inner circle, rather than emphasizinghow external forces influence a cohesive class of managerialelites. The findings not only support the social exchangemechanism, which operates within corporate elites, but theyalso show that ownership by institutional investors is onlyweakly related to the govemance and strategic changesanalyzed in this study. It may be that external forces play amore significant role in precipitating diffusion (cf, Fligstein,1991) by prompting initial defections. Further, the relativelyactive market for corporate takeovers of the early to mid-1980s may have created a political atmosphere conduciveto diminished elite solidarity (Davis and Thompson, 1994),thus precipitating initial defections. It is important to note,however, that we observe the spread of defection throughthe network of heretofore mutually supportive corporateleaders continuing well beyond the period of widespread andsignificant takeover threat to large corporations.

Our perspective suggests an alternative behavioral mecha-nism by which changes in governance arrangements,corporate strategy, and possibly other organizational phe-nomena spread across organizations. While existing theoreti-cal perspectives on network diffusion highlight the role ofcommunication in facilitating the spread of organizationalphenomena (Rogers, 1983; Davis, 1991), our perspectiveemphasizes the role of changing social obligations in drivingdiffusion: CEO-directors do not support greater board controlbecause they know it to be a legitimate or effective change

179/ASQ, March 1997

in governance arrangements from sitting on boards that havemade similar changes but because their experience withincreased board control in their home corporations haschanged their generalized social exchange relationship withfellow CEOs. Defection restores balance to the socialexchange relationship, while supporting other CEOs at thesame level would only maintain the new and unequal statusquo.

Our study also sheds light on the specific form of defection.Gouldner (1960: 172) suggested that reciprocity is often"homeomorphic" or "identical in form . . . with respect tothe things exchanged [and] the circumstances under whichthey are exchanged." Additional analyses suggested that thepresence of CEO-directors who experienced one kind ofincrease in board independence at their home companiesgenerally did not increase the likelihood of different kinds ofincreases elsewhere. CEOs experiencing the separation ofCEO and board chair positions, for example, were likely toinduce that specific change but not the other changesexamined in this study. Thus it appears that CEO-directorstend toward homeomorphic retaliation as a response toincreased board independence at their home companies.

The network of CEO-directors examined in this study sharescertain characteristics with a positive social exhange net-work, as conceived by Cook and Emerson (1978: 725), inthat exchange, or the use of power in one relation, iscontingent upon exchange in the other relation. Whereasequity concerns are static and restrain the use of power inCook and Emerson's model, however, we view equity as adynamic, social psychological mechanism that can facilitateas well as inhibit the exercise of power.

While the theory developed in this study is not concerneddirectly with the notion of trust, some might interpret ourfindings as counter to the notion that trust governs mostinterpersonal and interorganizational relationships. Relation-ships with trust, however, are neither extremeiy fragile norimpervious to change, and we propose, consistent withAxelrod (1984), that trust among exchange partners is mostrobust under stable environmental conditions and leastrobust under circumstances of substantial contextualdisruptions. The latter condition more accurately describesthe period under study. To the extent that there are fewerexternal disruptions in the corporate governance arena overthe next ten years, there may be fewer defections.

This study does not suggest, moreover, that board indepen-dence has become universal among large firms. Zajac andWestphal (1996) provided evidence suggesting that, inrecent years, managers may have used their influence overthe director selection process to avoid the potential forincreases in board independence. They found that powerfulCEOs favor the selection of directors who have not experi-enced increases in board independence, thus avoidingdirectors who are at risk of defecting. Conversely, boardsthat have already asserted their independence are morelikely to appoint new directors who have experiencedincreases in independence. Overall, then, it is not obviousthat board norms favoring independence and control have

180/ASQ, March 1997

Defections

universally replaced norms favoring passivity and support.Both sets of norms can be found across boards of largefirms, and mechanisms such as director selection are usedto reinforce them. Ironically, the emergence of board controlas a viable, competing norm may have raised solidarity orgroup awareness among those boards adhering to the oldnorms of support, even as it raises the prospect of defec-tion. By recognizing the existence of social forces forstability and for change in corporate governance practices, aswell as the role of the intercorporate network in impedingand impelling such changes, researchers can develop agreater understanding of seemingly inconsistent argumentsand results regarding the control of U,S, corporations.

REFERENCES

Allison, Paul D.1982 "Discrete-time methods for

the analysis of event histo-ries," In S, Leinhardt (ed,).Sociological Methodology:61-98, San Francisco:Jossey-Bass,

1984 Event History Analysis,Newbury Park, CA: Sage,

Amihud, Yakov, and Baruch Lev1981 "Risk reduction as a manage-

rial motive for conglomeratemergers," Bell Journal ofEconomics, 12: 605-617,

Axelrod, Robert1984 The Evolution of Cooperation,

New York: Basic Books,

Baysinger, Barry, and Robert E,Hoskisson1990 "The composition of boards of

directors and strategic control:Effects on corporate strat-egy," Academy of Manage-ment Review, 15: 72-87,

Beatty, Randolph P,, and EdwardJ, Zajac1994 "Top management incentives,

monitoring, and risk sharing: Astudy of executive compensa-tion, ownership and boardstructure in initial publicofferings," AdministrativeScience Quarterly, 39:313-336,

Bergh, Donald P,1995 "Size and relatedness of units

sold: An agency theory andresource-based perspective,"Strategic ManagementJournal, 16: 221-239,

Bethel, Jennifer E., and JuliaLiebeskind1993 "The effects of ownership

structure on corporaterestructuring," StrategicManagement Journal, 14:15-32,

Black, Fischer, and Myron Scholes1973 "The pricing of options and

corporate liabilities," Journalof Political Economy, 81:637-659,

Blau, Peter M.1977 Inequality and Heterogeneity,

Glencoe, IL: Free Press,

Burt, Ronald S,1987 "Social contagion and innova-

tion: Cohesion versus struc-tural equivalence," AmericanJournal of Sociology, 92:1287-1335,

Business Week1994 "How to handle a CEO,"

February 21: 56-61,

Clogg, Clifford C, and Scott R,Eliason1987 "Some common problems in

log-linear analysis," Sociologi-cal Methods and Research,15: 8^4 ,

Collins, Randall1990 Theoretical Sociology, New

York: Harcourt BraceJovanovich,

Cook, Karen S., and Richard M,Emerson1978 "Power, equity and commit-

ment in exchange networks,"American Sociological Review,43: 721-739,

Crystal, Graef S.1984 Questions and Answers on

Executive Compensation,Englewood Cliffs, NJ: Pren-tice-Hall,

1991 In Search of Excess: TheQvercompensation of Ameri-can Executives, New York:W,W, Norton,

Daily, Catherine M,1996 "Governance patterns in

bankruptcy reorganizations,"Strategic ManagementJournal, 17: 355-375,

Davis, Gerald F,1991 "Agents without principles?

The spread of the poison pillthrough the intercorporatenetwork," AdministrativeScience Quarterly, 36:583-613,

1992 "The interlock network as aself-reproducing socialstructure," Paper presented atthe Advancement of Socio-Economics Annual Confer-ence, Irvine,

Davis, Gerald F,, and Tracy A,Thompson1994 "A social movement perspec-

tive on corporate control,"Administrative ScienceQuarterly, 39: 141-173,

Dawes, Robyn M.1992 "Social dilemmas, economic

self-interest, and evolutionarytheory," In Donald R, Brownand J,E, Keith Smith (eds,).Recent Research in Psychol-ogy: Frontiers of MathematicalPsychology: 1-27, New York:Springer-Verlag,

DiMaggio, Paul J,, and Walter W,Powell1983 "The iron cage revisited:

Institutional isomorphism andcollective rationality inorganizational fields," Ameri-can Sociological Review, 48:147-160,

Ekeh, Peter P,1974 Social Exchange Theory: The

Two Traditions, Cambridge,MA: Harvard University Press,

Festinger, Leon1954 "A theory of social compari-

son processes," HumanRelations, 7: 117-140,

Fama, Eugene F., and Michael C,Jensen1983 "The separation of ownership

and control," Journal of Lawand Economics, 26: 301-325,

181/ASa, March 1997

Finkelstein, Sydney, and RichardA. D'Aveni1994 "CEO duality as a double-

edged sword: How boards ofdirectors balance entrench-ment avoidance and unity ofcommand." Academy ofManagement Journal, 37:1079-1108.

Fiigstein, Neil1991 "The structural transformation

of American industry: Aninstitutional account of thecauses of diversification." InWalter W. Powell and Paul J.DiMaggio (eds.), The NewInstitutionalism in Organiza-tional Analysis: 311-337.Chicago: University of ChicagoPress.

Fortune1993 "The king is dead: Why more

boards are waking up andpushing out the CEO."January 11: 34-50.

Fredrici<son, James W., Donaid C.iHambrick, and S. Baumrin1988 "A model of CEO dismissal."

Academy of ManagementReview, 13: 255-270.

Galaskiewicz, Joseph, and StanieyWasserman1989 "Mimetic processes within an

interorganizational field: Anempirical test." AdministrativeScience Ouaterly, 34: 454-479.

Gibbs, Phiiip A.1993 "Determinants of corporate

restructuring: The relativeimportance of corporategovernance, takeover threat,and free cash flow." StrategicManagement Journal, 14:51-68.

Gouidner, Aivin W.1960 "The norm of reciprocity: A

preliminary statement."American Sociological Review,25: 161-178.

Hambrick, Donald C, and PhyliisP. iVIason1984 "Upper echelons: The

organization as a reflection ofits top managers." Academyof Management Review, 2:193-206.

Hansen, Gary S., and CharlesW. L Hiil1991 "Are institutional investors

myopic? A time series studyof four technology-drivenindustries." Strategic Manage-ment Journal, 12: 1-16.

Harris, Milton, and Artur Raviv1979 "Some results on incentive

contracts with imperfectinformation." Journal ofEconomic Theory, 20: 231-259.

Harrison, J. Richard, David L.Torres, and Sai Kukaiis1988 "The changing of the guard:

Turnover and structuralchange in the top-manage-ment positions." Administra-tive Science Ouarterly, 33:211-232.

Haunschiid, Pamela R.1993 "Interorganizational imitation:

The impact of interlocks oncorporate acquisition activity."Administrative ScienceQuarterly, 38: 564-592.

Hermaiin, B. and iVI. Weisbach1988 "The determinants of board

composition." Rand Journal ofEconomics, 19: 589-605.

Herman, Edward S.1981 Corporate Control, Corporate

Power. New York: CambridgeUniversity Press.

Hiii, Charies W. L, and Scott A.Sneli1988 "External control, corporate

strategy, and firm perfor-mance in research-intensiveindustries." Strategic Manage-ment Journal, 9: 577-590.

Jensen, iVIichaei C.1989 "The eclipse of the public

corporation." Harvard Busi-ness Review, 67: 61-74.

Jensen, Michaei C, and Kevin J.iVIurphy1990 "Performance pay and

top-management incentives."Journal of Political Economy,98: 226-263.

Kerr, Jeffrey L., and Lesiie Kren1992 "Effect of relative decision

monitoring on chief executiveofficer compensation."Academy of ManagementJournal, 35: 370-397.

Kim, Hyosun, and Wiliiam Ocasio1995 "investor exit, investor voice:

Institutional investor, perfor-mance monitoring, and CEOsuccession." Working Paper,MIT.

Komorita, S. S., J. A. Hiity, andC. D. Parks1991 "Reciprocity and cooperation

in social dilemmas." Journalof Conflict Resolution, 35:494-518.

Kosnik, Rita D.1987 "Greenmail: A study of board

performance in corporategovernance." AdministrativeScience Ouarterly, 32:163-185.

Krebs, Dennis L., and D. T. Miiier1985 "Altruism and aggression." In

Gardner Lindzey and ElliotAronson (eds.). The Handbookof Social Psychology, 1: 1-26.Hillsdale, NJ: Eribaum.

Lawier, Edward E.1990 Strategic Pay: Aligning

Organizational Strategies andPay Systems. San Francisco:Jossey-Bass.

Levi-Strauss, Ciaude1969 The Elementary Structures of

Kinship. Boston: BeaconPress.

Lorsch, Jay W., and EiaineiViaciver1989 Pawns or Potentates: The

Reality of America's CorporateBoards. Boston: HarvardBusiness School Press.

iVIace, iVIyles L.1971 Directors: Myth and Reality.

Boston: Harvard BusinessSchool Press.

iVlain, Brian G., Charies A. O'Reiiiyili, and James Wade1995 "The CEO, the board of

directors and executivecompensation: Economic andpsychological perspectives."Industrial and CorporateChange, 4: 293-332.

iVIarris, Robert L.1964 The Economic Theory of

Managerial Capitalism.London: MacMillan.

iVIizruchi, Mark S.1996 "What do interlocks do? An

analysis, critique, and assess-ment of research on interlock-ing directorates." AnnualReview of Sociology, 22:271-298.

Murray, Aian i.1989 "Top management heteroge-

neity and firm performance."Strategic ManagementJournal, 10: 125-141.

Paiepu, Krishna G.1985 "Diversification strategy, profit

performance, and the entropymeasure." Strategic Manage-ment Journal, 6: 239-255.

182/ASQ, March 1997

Defections

Palmer, Donald A., P. DevereauxJennings, and Xueguang Zhou1993 "Late adoption of the multidi-

visional form by large U.S.corporations: Institutional,political, and economicaccounts." AdministrativeScience Quarterly, 38:100-131.

Pfeffer, Jeffrey1972 "Size and composition of

corporate boards of direc-tors." Administrative ScienceQuaterly, 17: 218-228.

Rogers, Everett M.1983 Diffusion of Innovations. New

York: Free Press.

Sundaramurthy, Chamu1996 "Corporate governance within

the context of antitakeoverprovisions." Strategic Man-agement Journal, 17: 377-394.

Tsui, Anne S., and Charles A.O'Reilly1989 "Beyond simple demographic

effects: The importance ofrelational demography insuperior-subordinate dyads."Academy of ManagementJournal, 32: 402^23.

Turban, Daniel B., and Allan P.Jones1988 "Supervisor-subordinate

similarity: Types, effects, andmechanisms." Journal ofApplied Psychology, 73:228-234.

Useem, Michael1984 The Inner Circle. Oxford:

Oxford University Press.

Vancil, Richard F.1987 Passing the Baton: Managing

the Process of CEO Succes-sion. Boston: Harvard Busi-ness School Press.

Walsh, James P., and James K.Seward1990 "On the efficiency of internal

and external corporate controlmechanisms." Academy ofManagement Review, 15:421^58.

Walster, Elaine, Ellen Berscheid,and G. William Walster1973 "New directions in equity

research." Journal of Person-ality and Social Psychology,25: 151-176.

Wayne, Sandy J., and Robert C.Liden1995 "Effects of impression

management on performanceratings: A longitudinal study."Academy of ManagementJournal, 38: 232-260.

Westphal, James D., and EdwardJ. Zajac1994 "Substance and symbolism in

CEOs' long-term incentiveplans." Administrative ScienceQuaterly, 39: 367-390.

1995 "Who shall govern? CEQ/board power, demographicsimilarity, and new directorselection." AdministrativeScience Quarterly, 40: 60-83.

1997 "The symbolic managementof stockholders: Corporategovernance reforms andshareholder reactions."Working Paper, University ofTexas at Austin.

Wiersema, Margarethe F., andKaren A. Bantel1992 "Top management team

demography and corporatestrategic change." Academyof Management Journal, 35:91-121.

Whisler, Thomas J.1984 Rules of the Game: Inside the

Board Room. Homewood, IL:Dow-Jones Irwin.

Yamaguchi, Kazuo1991 Event History Analysis.

Newbury Park, CA: Sage.

Zahra, Shaker A., and John A.Pearce1989 "Boards of directors and

corporate financial perfor-mance: A review and integra-tive model." Journal ofManagement, 15: 291-334.

Zajac, Edward J., and James D.Westphal1996 "Director reputation, CEQ-

board power, and the dynam-ics of board interlocks."Administrative ScienceQuarterly, 41: 507-529.

183/ASa, March 1997