Pay and Equity

Strategic Management JournalStrat. Mgmt. J., 31: 629–651 (2010)

Published online EarlyView in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/smj.827

Received 5 December 2005; Final revision received 30 October 2009

THE EFFECT OF CEO PAY DEVIATIONS ON CEOWITHDRAWAL, FIRM SIZE, AND FIRM PROFITS

ERIC A. FONG,1* VILMOS F. MISANGYI,2 and HENRY L. TOSI, JR.3,4

1 University of Alabama in Huntsville, College of Business Administration, Departmentof Management and Marketing, Huntsville, Alabama, U.S.A.2 The Pennsylvania State University, Smeal College of Business, Department ofManagement and Organization, University Park, Pennsylvania, U.S.A.3 University of Florida, Warrington College of Business Administration, Departmentof Management, Gainesville, Florida, U.S.A.4 Bocconi University, School of Management, Strategy and EntrepreneurshipManagement Department, Milan, Italy

We build upon previous work on the effects of deviations in CEO pay from labor markets to assesshow overcompensation or undercompensation affects subsequent voluntary CEO withdrawal,firm size, and firm profitability, taking into account the moderating effect of firm ownershipstructure. We find that CEO underpayment is related to changes in firm size and CEO withdrawal,and that the relationship between CEO underpayment and CEO withdrawal is stronger in owner-controlled firms. We also show that when CEOs are overpaid, there is higher firm profitability;a relationship that is weaker among manager-controlled firms. We then discuss the implicationsthat these findings have for future research. Copyright 2010 John Wiley & Sons, Ltd.

INTRODUCTION

CEO compensation has been a topic of great inter-est for several decades in the business press (i.e.,articles in Fortune magazine and surveys in Busi-ness Week ) as well as the academic literature (see,for example, Bebchuk and Fried [2004] for one ofthe more recent inquiries). Typically this researchaddresses the ‘hot’ question of whether CEOs areoverpaid, and is based upon the convention thatthe financial performance of the firm is the basic

Keywords: CEO compensation; CEO turnover; CEOpower; equity; labor markets; overpayment*Correspondence to: Eric A. Fong, University of Alabama inHuntsville, College of Business Administration, Departmentof Management and Marketing, 202 Business AdministrationBuilding, Huntsville, AL 35 899, U.S.A. E-mail: [email protected]

determinant of appropriate levels of CEO com-pensation. Recently, however, some scholars haveargued that inquiries regarding the proper levelof CEO compensation should give more consid-eration to ‘relative evaluation within an industry’(Miller, 1995: 1381) and to the role that the ‘exec-utive labor market’ (Ezzamel and Watson, 1998:221) plays in CEO compensation.

Though few in number, studies that have utilizedthis relative evaluation, or labor market, approachhave produced some interesting findings; for exam-ple, it has been shown how the compensation ofboard of director members is related to the payof CEOs in the focal firm (e.g., O’Reilly, Main,and Crystal, 1988; Porac, Wade, and Pollock,1999). Also, studies have shown that CEO under-and/or overpayment (i.e., relative to the goingrate of the executive labor market) has effects on

Copyright 2010 John Wiley & Sons, Ltd.

630 E. A. Fong, V. F. Misangyi, and H. L. Tosi, Jr.

future levels of CEO compensation (Ezzamel andWatson, 1998) as well as on the compensationand turnover of lower-level managerial employ-ees (Wade, O’Reilly, and Pollock, 2006). In otherwords, these studies show that considerations offairness, or equity, play a role in the determina-tion of CEO pay. At the same time, this researchhas highlighted the importance of looking at thepotential consequences of CEO under- and/or over-payment.

The current study builds upon this body ofresearch to investigate how the under- and over-payment of CEOs relative to the CEO labor mar-ket rate may lead to actions that could theoret-ically help CEOs to resolve their own sense ofequity. Specifically, both increasing the size ofthe firm as well as voluntarily withdrawing fromthe firm present underpaid CEOs with viable andfair alternatives for redressing their underpaymentsituation. An overpaid CEO, on the other hand,would seek to improve firm profitability as themost desirable and fair alternative for redress-ing the situation. Furthermore, extant theory andresearch regarding the managerial discretion con-ferred by different types of ownership structurewould suggest that the firm’s ownership structuremay moderate each of these relationships.

We investigate these questions using a sampleof approximately 3,000 observations with respectto 900 CEOs from large U.S. publicly held firmscovering 30 industries over a 10-year time period.We follow previous research that has investigatedthe effects of CEO pay differentials, and thus startwith the notion that the CEO labor market is animportant referent that affects CEOs’ perceptionsof fairness and how they react to it (Wade et al.,2006; Watson et al., 1996). Our findings supportthe premise that CEOs react to fairness considera-tions in ways that have major implications for thefirm and its objectives.

THEORETICAL BACKGROUND

Previous research suggests that significant devia-tions in CEO pay from the ‘going market rate’ (i.e.,in the executive labor market) will have equity-oriented effects. Theoretically, to the extent thatan informationally efficient executive labor mar-ket exists, then the substantial underpayment ofan executive relative to the market rate wouldresult in that executive being bid away by other

firms prepared to pay the market rate (Ezzameland Watson, 1998; Fama, 1980). In other words,boards of directors must consider market pay whendetermining the CEO’s compensation because ‘formotivational, recruitment, and retention reasons, afirm’s compensation committee has to ensure thatits senior executives are paid. . .the compensationlevel typically paid by similar firms to compara-ble individuals occupying similar posts’ (Ezzameland Watson, 1998: 221). Miller (1995) found thatchanges in CEO compensation were more directlytied to industry referents (i.e., industry-relativeevaluations of performance) than to firm referents(i.e., prior firm performance), leading him to con-clude that equity theoretic concerns may be thenext logical step in developing our understandingof CEO compensation.

These studies, and others, suggest that the levelof CEO pay, among other effects, may be explainedby social comparison processes (e.g., Festinger,1957), which underlie equity theory oriented con-cerns (Adams, 1965). O’Reilly et al. (1988) founda positive association between CEO compensation(in a sample of large U.S. firms) and the compen-sation of compensation committee directors, whichled them to conclude that social comparisons occurin the CEO compensation setting process. Poracet al. (1999) found that boards of directors (of asample of S&P 500 firms) use social comparisonprocesses in a somewhat political manner, and thussuggest that such processes operate in the setting ofCEO compensation. Though boards generally usewithin-industry comparisons in setting CEO pay,it was also found that when those comparisonsplaced focal firms in an unfavorable light (e.g.,poor performance relative to comparison organi-zations), then their boards expanded their com-parisons beyond industry boundaries seemingly toprotect both the CEO and the board from criticism.

Social comparison processes have also been theunderpinning of studies that examine the relation-ship that under- and overpayment deviations fromthe going executive labor market rate have withsuch outcomes as managerial job satisfaction (Wat-son et al., 1996), as well as subordinates’ pay andturnover (Wade et al., 2006). Watson et al. (1996)investigated the relationship between the compen-sation and job satisfaction of nonowner managers(of small and medium-sized U.K. firms), examin-ing whether under- and overpayment (as calculatedby taking the difference between the managers’actual pay and their predicted pay based upon

Copyright 2010 John Wiley & Sons, Ltd. Strat. Mgmt. J., 31: 629–651 (2010)DOI: 10.1002/smj

Effect of CEO Pay Deviations 631

their human capital, firm, and industry characteris-tics) was related to job satisfaction. They foundthat ‘though the absolute level of pay was notrelated to job satisfaction. . .overpayment (under-payment) relative to their estimated comparisonlevel is reflected by significantly higher (lower)levels of job satisfaction’ (Watson et al., 1996:576).

The study by Wade et al. (2006) examinedwhether the under- and overpayment of CEOs rel-ative to the going executive labor rate (as mea-sured by the residuals of a ‘CEO wage equation’)affected the compensation of lower-level managerswithin the CEOs’ firms (in a sample of large,U.S. publicly held firms). In essence, their resultssuggest that CEOs recognize that they themselvesare under- or overpaid (relative to other CEOs)and that this influences the wage-setting processof their subordinates such that the latter are alsopaid above or below their respective labor mar-ket rates (i.e., that equity prevails within the firm)and this affects firm effectiveness (via reducedemployee turnover, etc.). Indeed, their findingsalso show that ‘inequity, expressed as compara-tive underpayment with the CEO (internal under-payment inequity) and with the average wage forthe job in other organizations (external underpay-ment inequity), was associated with higher lev-els of [subordinate] turnover’ (Wade et al., 2006:539). Thus, their study highlights the inherent rolethat social comparison processes play in the work-ings of any labor market, including that of CEOs.Furthermore, it suggests that in addition to theassumption of the economic matching of wagesto ability, the workings of the labor market alsoimplicitly requires ‘that there should not be strongfeelings of injustice’ (Hicks, 1963: 317; cf. Wadeet al., 2006: 528).

In short, these studies show that social compar-ison processes—and thus norms of fairness andconcerns with equity—are integral to the work-ings of the executive labor market and that theyoperate among those who are setting wages (i.e.,boards of directors, compensation committees, andCEOs) as well as those who receive them (i.e.,CEOs and employees). This assumes, then, that‘in order to determine how a person evaluates areward, he must first compare his own inputs andoutcomes to others’ and that social comparison‘theories assume that judgments of fairness mat-ter a great deal in determining peoples’ responsesto the comparative evaluations’ (Wade et al., 2006:

529). We build upon the findings of this researchand its premises, to develop several hypothesespertaining to the actions that CEOs may take toresolve dissonance created when their pay deviatesfrom the going executive labor market rate.

CEO underpayment, firm size, and CEOwithdrawal

Organizational approaches to fairness (see Wadeet al., 2006) suggest that CEOs whose pay deviatessubstantially from the labor market should be moti-vated to reduce the dissonance caused by this situ-ation. Underpaid CEOs can reduce dissonance by(1) increasing their outcomes (i.e., rewards; Green-berg, 1990), (2) altering their situation either mate-rially (i.e., withdrawal; Greenberg, 1990) or cog-nitively (i.e., rationalization; Greenberg, 1989), or(3) reducing their effort (Adams, 1965; Cowherdand Levine, 1992). We believe that the last alter-native, despite the suppositions of ‘effort mini-mization’ or ‘shirking’ posited in agency theory, isgenerally at odds with theoretical accounts of CEOdispositions and their motivations, which suggestthat CEOs tend to have high power motivationas well as reasonably high levels of achievementmotivation (Simon, 1947; Marris, 1964; McClel-land and Boyatzis, 1982; Davis, Schoorman, andDonaldson, 1997). Thus, CEOs will seek to rem-edy the dissonance created in underpayment situ-ations by attempting to affect their rewards, to theextent that any increase in rewards is greater thanthe increases in effort necessary to obtain them(Adams, 1965), and through seeking alternativesituations rather than through simply ‘shirking’ intheir present situation.

As to dissonance reduction via affecting out-comes, both theory and evidence suggest thatincreasing firm size provides a vehicle throughwhich CEOs can increase their rewards monetar-ily (i.e., salary; Gomez-Mejia, Tosi, and Hinkin,1987; Hambrick and Finkelstein, 1995) as well asnonmonetarily (i.e., power, prestige, status, and jobsecurity; Marris, 1964; Simon, 1947; Williamson,1964). Increasing the size of the firm most likelyrequires more effort by the CEO and thus thismarginal product should also result in correspond-ing increases in financial rewards (Roberts, 1959;Henderson and Fredrickson, 1996; Finkelstein andBoyd, 1998). Thus, from a purely extrinsic stand-point, firm growth may not redress the inequity



situation—as both CEO effort and financial out-comes will increase, possibly in a manner thatmay maintain the current inequity. Firm growthstands to reduce dissonance, nonetheless, becauseit also leads to increases in highly valued rewardsthat go well beyond financial gains. Indeed, inaddition to fulfilling CEOs’ power and securityneeds (Barnard, 1938; Simon, 1947; Marris, 1964),growing the firm also provides an opportunityfor CEOs to fulfill their higher order needs (i.e.,achievement; self-actualization) given that they areat the apex of the organization and have lim-ited opportunities for such self-fulfillment (Marris,1964; Williamson, 1964; Davis et al., 1997). At thevery least, this increasing of nonmonetary rewardsshould alter the situation in a manner that greatlyenhances the possibility for CEOs to change theircognition regarding its fairness (Greenberg, 1989).Furthermore, in terms of fairness, because thegrowth of large publicly held firms serves the inter-ests of the CEOs and their subordinates and maycome at the expense of shareholders (Marris, 1964;Mueller, 1972), such an action is completely con-sistent with considerations of norms of fairness.While increasing the outcomes for the underpaidCEO, firm growth doesn’t help those ‘responsible’for the underpayment condition (i.e., shareholdersas represented by boards of directors). In short,when CEOs are underpaid, increasing the size ofthe firm presents a means by which CEO outcomescan be increased in a manner that addresses theirfairness concerns. Thus, we hypothesize that:

Hypothesis 1: CEO underpayment is associatedwith subsequent increases in firm size.

Firm growth, however, is not the only route opento underpaid CEOs for dissonance reduction, as the‘decision to participate in the organization—or toleave the organization’ (March and Simon, 1958:48) is, in part, a function of the existing alternativesavailable to the CEO. Therefore, to the extentthat there are desirable external alternatives, someCEOs may reduce their dissonance by voluntarilywithdrawing from the firm. Although this issuehas not been examined previously with regard toCEOs, there is evidence to suggest that inequity inpay does lead to voluntary turnover among lower-level employees. For example, Wade et al. (2006)show that as subordinate underpayment inequityincreased relative to a CEO’s pay, subordinateturnover also increased. Also, Zenger (1992) found

that in organizations where the emphasis was onrewarding only the best performers, moderatelyhigh performers were more likely to leave giventheir relative underpayment to the best performers.Therefore, we hypothesize that:

Hypothesis 2: CEO underpayment is associatedwith voluntary CEO withdrawals.

CEO overpayment and firm profitability

Being overpaid relative to the going labor mar-ket rate should result in dissonance, and althoughtheoretically this may lead to actions that seek toincrease inputs and/or decrease rewards (Adams,1965), existing evidence tends to support the for-mer (Brockner et al., 1986; Greenberg, 1988). Forexample, Greenberg (1988) found that when man-agers were assigned to offices of higher status thantheir position warranted, they increased their per-formance rather than give up their offices. This ten-dency to increase inputs or effort and not to reduceoutcomes would also be expected among overpaidCEOs, given the CEO’s motivational structure asdiscussed above. High power motivation individ-uals have the need to establish and maintain pres-tige (McClelland, 1975), a need clearly not con-ducive to the reduction of financial rewards (i.e.,outcomes). Furthermore, individuals with achieve-ment needs take personal responsibility for suc-cess; they must feel that success comes from theirown efforts. Thus, they tend to seek feedbackabout success and are concerned with how theirefforts lead to it (McClelland, 1953; 1961; 1965).In an overpayment situation, wherein the feedbackis that a CEO’s financial rewards are more thanfor comparable CEOs with similar firm profitabil-ity, CEOs will subsequently seek to increase theirefforts toward profitability as opposed to attempt-ing to reduce their outcomes because this shouldwork toward reducing overpayment in a man-ner consistent with their power and achievementneeds.

Norms of fairness would also suggest that CEOsincrease their efforts toward the interests of theprincipals (i.e., shareholders as represented by theboard of directors) for whom they are agents. Inshort, when overpaid, CEOs should work towardincreasing firm profitability. Here too, equity the-ory points to how CEO compensation involvesmore than simply extrinsic motivations and



rewards. Given that virtually all CEO compen-sation packages incorporate some type of mech-anisms that link pay to performance (Bebchukand Fried, 2004; Jensen and Murphy, 1990), andthus the overpayment (as does the underpayment)condition occurs in the presence of this ‘incen-tive alignment,’then the overpaid CEO’s increasedeffort toward firm profitability may also lead toincrease the CEO’s financial rewards, therebypotentially perpetuating the overpayment situation.Nevertheless, the increase of effort toward theexpectations of those rewarding the CEO shouldwork toward appeasing the CEO’s felt dissonancein the overpaid situation (i.e., the judgment thathis or her financial rewards are more than war-ranted for the level of his or her inputs) becausesuch dissonance is necessarily based upon a beliefthat he or she is not inputting enough effort (i.e.,otherwise the CEO would have no fairness-basedconcerns). Moreover, to the extent that overpaidindividuals are only moderately dissatisfied withtheir overpayment condition (Adams, 1965; Wal-ster, Walster, and Berscheid, 1978)—as they areconfronted with the conflict between self-interestedand fairness-based concerns when forming theirjudgments about being overpaid (Messick andSentis, 1983; Peters, van den Bos, and Bobocel,2004)—then the potential that the overpaymentcondition may be perpetuated by an increase ineffort toward profitability would not seem to detersuch a CEO action. That is, by increasing his orher efforts toward profitability, the overpaid CEOis doing all that he or she can be expected todo (by themselves or shareholders) toward makingthe payment inequity more just. In short, increas-ing effort toward profitability should mollify theoverpaid CEO, as taking such action addresses theCEO’s fairness-based concerns in a manner con-sistent with the CEO’s motivational structure (i.e.,taking responsibility for his or her success), evenif doing so may potentially produce a subsequentoverpayment condition (which, incidentally, wouldrest easily with his or her own self-interest). Thus,we predict that:

Hypothesis 3: CEO overpayment is associatedwith subsequent increases in firm profits.

The moderating effect of ownership structure

The foregoing arguments presume that CEOs havethe latitude to alter conditions in a manner that

reduces dissonance. In particular, the assumption isthat CEOs have the discretion to pursue managerialobjectives even if they are at the expense of equityholders’ objectives. Scholarship grounded in man-agerial capitalist theory suggests that the separationof firm ownership from firm control is the essen-tial factor that affords CEOs such discretion (Berleand Means, 1932; Marris, 1964; Williamson, 1964;McEachern, 1975). Evidence suggests that CEOsin firms lacking a dominant ownership position(i.e., ‘manager-controlled’ or ‘MC’ firms) havemore leeway to pursue managerial objectives thando CEOs of firms in which there is a domi-nant ownership position (i.e., ‘owner-controlled’ or‘OC’ firms). For instance, studies have found dif-ferences across MC and OC firms in accountingpractices (see Tosi et al., 1999), firm diversifica-tion (Amihud and Lev, 1981; Hill and Snell, 1988,1989), firm performance (see Hunt, 1986), boardof director monitoring (Tosi and Gomez-Mejia,1989, 1994), and the source of annual CEO payraises (Gomez-Mejia et al., 1987; Hambrick andFinkelstein, 1995). Furthermore, several studieshave established the importance of a third categoryof ownership control structure: that of ‘owner-managed’ (OM) firms in which the CEO is thedominant stockholder (McEachern, 1975, 1978;Salancik and Pfeffer, 1980). Therefore, manage-rial capitalist theory suggests that these differencesin firm ownership structure moderate the abovehypothesized relationships.

First, such differences moderate the relation-ship between CEO underpayment and subsequentincreases of firm size. Although all CEOs mayvalue growth for both extrinsic and intrinsic rea-sons as described above, CEOs in MC firms willhave more discretion to pursue this course of actionthan will their counterparts in OC firms (Marris,1964; Williamson, 1964). As an example, stud-ies by Gomez-Mejia et al. (1987) and Hambrickand Finkelstein (1995) have found that annual payraises for CEOs in MC firms were based moreupon changes in firm size than were pay raisesof CEOs in OC firms, as the raises for the lat-ter were more dependent upon changes in per-formance. Several theorists have also argued thatowner-managers may have a tendency to ‘buildempires’ (Knight, 1921; Schumpeter, 1942; Simon,1947) and the findings of McEachern (1978) aresupportive of this contention. Based on these argu-ments, we hypothesize that CEOs of OC firms areless likely to pursue growth than are CEOs of MC



or OM firms when faced with an underpaymentsituation:

Hypothesis 4: Ownership structure moderatesthe relationship between CEO underpaymentand subsequent increases in firm size, such thatthe positive relationship between CEO under-payment and increases in firm size is weakeramong OC as compared to MC or OM firms.

Second, ownership structure moderates the rela-tionship between CEO underpayment and vol-untary CEO withdrawal. Tosi and Gomez-Mejia(1994) show that CEOs in OC firms face moremonitoring than CEOs in MC firms. Additionally,studies by McEachern (1975) and Salancik andPfeffer (1980) show that CEO tenure is more con-tingent on the firm’s profit performance in OCfirms than in MC firms, while CEO tenure is‘virtually buffered from performance concerns’ inOM firms (Salancik and Pfeffer, 1980: 662). Thus,because underpaid CEOs in OC firms face thegreatest board vigilance, it would appear that theywould be more likely to voluntarily withdraw thanwould underpaid CEOs of MC or OM firms:

Hypothesis 5: Ownership structure moderatesthe relationship between CEO underpaymentand CEO voluntary withdrawal, such that thepositive relationship between CEO underpay-ment and CEO voluntary withdrawal is strongeramong OC firms as compared to MC or OMfirms.

Finally, firm ownership structure should alsomoderate the relationship between CEO overpay-ment and subsequent increases in firm profits. Asargued above, existing theory and evidence sug-gests that CEOs in an overpayment condition willtend to address norms of fairness by increasingtheir efforts toward profitability. Theory also sug-gests, nevertheless, that individuals will redressinequities in the least costly manner, and thus maycognitively redefine their situation as equitable(Adams, 1965; Greenberg, 1989; Walster et al.,1978). Although there is no evidence to suggestthat overpaid individuals cognitively redefine theirsituations, CEOs of MC firms would seem to havemore leeway to reduce dissonance ‘cognitively’when overpaid than will their counterparts. Thatis, because profitability concerns are very salient

to CEOs of OC firms (via being highly moni-tored; Tosi and Gomez-Mejia, 1989, 1994) and toowner-manager CEOs (via their significant owner-ship stakes; Demsetz, 1983), these CEOs are muchmore likely to redress their overpayment inequitythrough behavioral responses than are CEOs ofMC firms. In short, to the extent that overpaidCEOs are torn between self-interested and fairness-based concerns (e.g., Messick and Sentis, 1983;Peters et al., 2004), CEOs of MC firms wouldseem to be more likely to be able to cognitivelyredefine their overpayment condition in a man-ner congruent with self-interested motivations, andthus less likely to increase firm profits than CEOsin OC firms or owner-manager CEOs, as the lat-ter will be more likely to follow their concernsfor fairness and thus increase their efforts towardprofitability:

Hypothesis 6: Ownership structure moderatesthe relationship between CEO overpayment andsubsequent increases in firm profitability, suchthat the positive relationship between CEO over-payment and increases in firm profitability isweaker among MC firms as compared to OC orOM firms.

METHODS

Data and sample

The issues under study are most pertinent amonglarge established companies and require observa-tions over time, thus our initial sample was drawnrandomly from the largest (total assets greaterthan $10 million) U.S. publicly traded corpora-tions from the Compustat database in 1995, forwhich we then collected data for the time periodof 1990—1999. The final sample for each anal-ysis varied slightly due to data availability forthe particular variables included in each modelspecification: the withdrawal analysis was basedupon a final sample of 2,955 observations overtime nested within 932 CEOs across 30 industries,after accounting for missing variables and remov-ing observations of CEOs who were fired or diedon the job. With respect to the analyses of changesin firm profitability and firm size, the final sam-ples consisted of 2,690 observations within 912CEOs and 2,666 observations within 908 CEOs,respectively, after accounting for any missing data.



All data used in this study were obtained fromthe Compustat, Execucomp, and Compact Disclo-sure databases as well as proxy statements, 10-Ks,annual reports, and Dun and Bradstreet ReferenceBook of Corporate Management (Dun and Brad-street, 1999). All firm financial and total CEO paydata is adjusted to 1990 dollars using the consumerprice index.

Dependent variables

Change in firm size

We calculated change in firm size as (sizet−sizet−1)/sizet−1, where sizet represents firm size inany given year t and sizet−1 represents firm size inthe previous year. This relative measure of changein firm size helps to control for the fact that largerorganizations have greater resources to put forthto growth (Weinzimmer, Nystrom, and Freeman,1998). Following previous research (e.g., Tosi andGomez-Mejia, 1989; Tosi et al., 2004), firm size ismeasured as a composite index represented by thesingle factor resulting from a principal componentsanalysis (PCA) of two commonly used indicatorsof firm size: the natural log of firm sales (e.g.,Boeker, 1997) and the natural log of the numberof employees (e.g., Miller and Chen, 1994).

Voluntary CEO withdrawal

Following previous research (Parrino, 1997; Shenand Cannella, 2002), we operationalized voluntaryCEO withdrawal as a categorical variable such thatsituations in which the CEO left the position andeither became a director or retained a previouslyheld directorship were coded as equal to 1, and0 otherwise. Shen and Cannella (2002) suggestthat retaining a director position does not representdismissal; a CEO who resigns the position butretains a director position has likely voluntarilywithdrawn from being the CEO. To determinewhether a CEO became a director or retained thedirectorship after a turnover event, we examinefirm proxy statements around a CEO’s departureto determine the cause for departure. The initialnumber of observations of CEO turnover was 259,from which six observations were removed whereturnover was caused by death or where the causeof turnover was clearly labeled as a dismissal.Of the 253 remaining turnover observations, 155involved the CEO becoming a director, or retaininga directorship, after the turnover event.

Change in firm profitability

Firm profitability was measured using return onassets (ROA), calculated as the firm’s net incomedivided by total assets for each year. Similar tochange in firm size, change in firm profitability wascalculated as (ROAt− ROAt−1)/ROAt−1, whereROAt represents firm profitability in any givenyear t and ROAt−1 represents firm profitability inthe previous year.

Independent variables

CEO under- and overpayment

Consistent with previous studies, CEO under- andoverpayment was measured by using the residualsresulting from the regression of the natural log oftotal CEO pay on a set of theoretically relevanthuman capital, organizational, and industry fac-tors shown in previous research to be determinantsof CEO pay (i.e., a ‘CEO wage equation’; Wadeet al., 2006; Watson et al., 1996). The appendixcontains a full description of the measures andmethodology used in this estimation procedure. Inshort, a positive residual represents a condition inwhich the CEO was overpaid because the CEO’sactual total pay was greater than his or her pre-dicted total pay, whereas a negative residual sug-gests underpayment because the CEO’s actual totalpay was less than his or her predicted total pay.

Because the hypothesized effects of under- andoverpayment are not symmetrical, and because it islikely that individuals make comparisons with oth-ers who are better and more expert than themselves(O’Reilly et al., 1988), we used these residuals toconstruct two categorical variables to capture thoseCEOs who were clearly under- or overpaid withrespect to the going labor market rate. CEO under-payment was coded as equal to 1 when the CEO’sresidual wage score was in the lowest quartile ofall CEO residual scores in the sample, and 0 oth-erwise. CEO overpayment was coded as equal to1 when the CEO’s residual wage score was in thehighest quartile in the sample, and 0 otherwise.Thus, the omitted reference group comprises thoseCEOs who were paid more closely to the execu-tive labor market rate (i.e., ‘fairly paid’; CEOs inthe sample whose residuals were in the second andthird quartiles).



Ownership structure

Consistent with previous research, ownershipstructure of the firm was measured using a fivepercent equity-holding threshold (Tosi and Gomez-Mejia, 1989; Hambrick and Finkelstein, 1995) tosort organizations into our previously mentionedthree categories (McEachern, 1975): (1) MC ismeasured as a dummy variable equal to 1 in whichno single equity holder owns five percent or moreof the organization’s common stock, 0 otherwise;(2) OM is measured as a dummy variable equalto 1 in which the CEO owns five percent or moreof the common stock, 0 otherwise; and, (3) OC ismeasured as a dummy variable equal to 1 in whichat least one equity holder, who is not the CEO,owns five percent or more of the common stock,and 0 otherwise. Given the hypotheses, OC firmsare the omitted reference category in the tests ofthe hypotheses for voluntary CEO withdrawal andfirm size, and MC firms are the omitted referencecategory in the tests of hypotheses pertaining tofirm profitability.

Control variables

We controlled for several CEO and organizationalcharacteristics as well as corporate governancevariables shown by previous research to be per-tinent to the current analyses. First, with regard toCEO characteristics, we controlled for CEO ageand near retirement in the analysis of CEO with-drawal to further distinguish voluntary turnover(withdrawal) from nonvoluntary turnover. Nearretirement was measured as a dummy variableequal to 1 if the CEO was age 63 or older, and0 otherwise. Controlling for CEO age and nearretirement should help to isolate voluntary with-drawal because they account for turnover due toage (DeFond and Park, 1999) and forced retire-ment (Shen and Cannella, 2002). CEO tenure wasalso incorporated into all of the analyses, measuredas the number of years the CEO has held his orher current position, as this has been shown to be aproxy for the political processes (i.e., CEO power)that may affect CEO compensation (e.g., Finkel-stein and Hambrick, 1989; Hill and Phan, 1991)and withdrawal (Salancik and Pfeffer, 1980).

In terms of organizational characteristics,resource availability was incorporated as a controlvariable in all of the analyses under the assumptionthat greater organizational slack (Cyert and March,

1992) offers more opportunities for CEOs to takeactions that affect organizational outcomes (Ham-brick and Finkelstein, 1987). It was measured asa composite index of the firm’s average retainedearnings (net of depreciation) and dividend payoutratio (dividends per share divided by earning pershare) (Bourgeois, 1981). Firm profitability, mea-sured as ROA, was included as a control in theanalyses of voluntary CEO withdrawal and changein firm size. Firm size, measured as described ear-lier (i.e., PCA composite index), was entered as acontrol in the analyses of withdrawal and changein firm profitability. We also controlled for priorchange in firm size and prior change in firm prof-itability, which capture these respective changesprior to the initial year we observe under- or over-payment for a given CEO, in the change in firmsize and the change in firm profitability analyses,respectively. Thus, our analyses account for thesetypes of changes prior to the initial observation ofunder- or overpayment. The degree of total diver-sification was also included in the analysis of thechange in firm size, as this may be related to firmgrowth. It was measured using the entropy measureof diversification (Jacquemin and Berry, 1979),

T D =N∑

i=1

Pi ln(1/Pi),

where N is the number of industry segments inwhich the firm does business (defined by the four-digit standard industrial classification [SIC] code),and Pi is the percentage of firm sales in the ithindustry segment.

Finally, we controlled for several corporate gov-ernance variables, as the discretion to pursue man-agerial objectives is purported to be constrainedby a ‘bundle of governance mechanisms’ (Redikerand Seth, 1995: 87; Jensen, 1993). Director owner-ship was included because shareholdings by boardmembers other than the CEO may also represent apowerful check on the CEO (Fama, 1980). Direc-tor ownership is measured as the total percentageof shares held by the directors (less the percent-age of shares held by the CEO, since the CEO isincluded in this total). Institutional ownership mayalso constitute a potential check on managerial dis-cretion (Davis and Thompson, 1994). To eliminateany overlap between the Security Exchange Com-mission’s 13(f) reporting of the percentage of afirm’s common voting shares held by institutions



and institutional holdings captured in the owner-controlled ownership category (i.e., institutionalholdings of 5% or more), we used the residualsof a regression of the average percentage of afirm’s common voting shares held by institutionson the average percentage of five percent block-holding to measure institutional ownership. Theoutsider ratio, the most commonly used indicatorof board power in corporate governance research(Dalton et al., 1998), was calculated as the numberof outside directors divided by the total numberof directors. CEO duality was measured with adummy variable equal to 1 if the CEO holds boththe chief executive officer and chairman of theboard positions, and 0 otherwise, because such,duality should afford CEOs more power (Finkel-stein, 1992). Incentive pay, measured as the pro-portion of CEO total pay that comprises long-termincentive pay (i.e., long-term incentive pay, stockoptions, restricted and unrestricted stock grants,and deferred compensation), was also included asit is a key governance mechanism used by boardsto help align managerial interests with sharehold-ers’ interests.

Analytical method

The data of our study consists of a panel designin the form of repeated observations over timenested within firms, which are in turn nested withinindustries. Therefore, we used hierarchical linearmodeling (HLM) (Raudenbush and Bryk, 2002)as our statistical analytic technique because itaffords a modeling of the relationship over timethat CEO under- and overpayment has with theoutcome in question while simultaneously account-ing for the nesting of these relationships withinfirms and industries, thereby providing unbiasedand efficient estimates of the regression coeffi-cients and their standard errors despite the depen-dence among observations (Bryk and Raudenbush,1989). It does so by simultaneously modeling threelevels of analysis (for illustration purposes, wepresent the model specification for the change infirm size analysis as all other analyses are of asimilar nature):

Ytij = π0ij + π1ij (ROA)t−1,ij + π2ij

(CEO underpayment)t−1,ij + π3ij

(CEO overpayment)t−1,ij

+ π4ij (CEO tenure)t−1,ij

+ π5ij (Incentive pay)t−1,ij + εtij (1)

π0ij = β00j + β01j (MC) + β02j (OM)

+ β0Qj(Control V ariables) + r0ij (2)

π1ij = β10j (2a)

π2ij = β20j + r2ij (2b)

π3ij = β30j + r3ij (2c)

π4ij = β40j (2d)

π5ij = β50j (2e)

β00j = γ000 + µ00j (3)

β01j = γ010 (3a)

β02j = γ020 (3b)

β0Qj = γ0Q0 + µ0Qj (3c)

β10j = γ100 (3d)

β20j = γ200 + µ20j (3e)

β30j = γ300 + µ30j (3f)

β40j = γ400 + µ40j (3g)

β50j = γ500, (3h)

where the indices t , i, and j denote time, firms,and industries with t = 1, 2, . . ., nij time periodswithin firm i in industry j ; i = 1, 2, . . ., Ij firmswithin industry j ; and j = 1, 2, . . ., J industries.

In this modeling, Equation 1 models the timelevel in which the change in firm size in the focalyear (i.e., year t) of firm i in industry j (Ytij ) isregressed upon the previous year’s CEO under-payment and CEO overpayment (i.e., both in yeart − 1) of firm i in industry j as well as the pre-vious year’s firm ROA, CEO tenure, and incentivepay (i.e., year t − 1) of firm i in industry j . Allof these time-level relationships were grand meancentered (see Hofmann and Gavin, 1998), and thusthe intercept of Equation 1, π0ij , represents themean change in firm size across time for firm i

in industry j , adjusted for the effect of the time-varying variables.

The intercept of Equation 1, π0ij , is modeledsimultaneously as the outcome in Equation 2, andis regressed on the ownership structure variables(MC and OM ) and control variables (directorownership, institutional ownership, outsider ratio,duality, resource availability, prior change in firmsize, and total diversification); thus β0Qj represents



the vector of coefficients β03j through β09j ) vari-ables expected to explain between-firm variance.The intercept of Equation 2, β00j , thus representsthe mean change in firm size of all firms in indus-try j adjusted for these firm-level predictors (again,given grand mean centering).

Equation 3 simultaneously models β00j as adependent variable in a cross-sectional industrymodel, where γ000 is the grand mean of change infirm size. Furthermore, as shown in Equations 1,2, and 3, each level of analysis has its own uniquerandom error term: etij represents the across-timeresidual; rij the between-firm residual; and, µj

the between-industry residual. Finally, as theseequations also show, HLM models the slopes ofthe relationships at both the time and the firm lev-els as outcome variables at the higher levels ofanalysis (Equations 2a–2e and 3a–3h), and thesewere modeled at the subsequent levels as fixed orrandom based upon what best fit the data (Rauden-bush and Bryk, 2002).

The above modeling demonstrates severaladvantages of using HLM for this analysis. First,the partitioning out of the variance (i.e., into acrosstime, between-firm, and between-industry) effec-tively controls for industry effects and thus allevi-ates the need to adjust variables (i.e., subtractingout industry means, industry dummy variables, etc)in an attempt to control for such effects (Bloomand Milkovich, 1998). Second, the modeling ofthe slopes as outcomes allows for the testing of themoderating hypotheses regarding ownership struc-ture by regressing the slope of the relationship pro-posed to be moderated (in this case, π2ij , the rela-tionship between CEO underpaymentt−1,ij andchange in firm size (Ytij )) on the ownership struc-ture variables. This is accomplished by modelingthe ownership structure variables in Equation 2babove:

π2ij = β20j + β21j (MC) + β22j (OM) + r2ij

(2b)

Finally, because the analysis of CEO withdrawalinvolves a binary outcome, we used a hierarchi-cal generalized linear model (HGLM); this proce-dure has the positive attributes of HLM as wellas appropriately tests binary outcomes using aBernoulli sampling model and logit link (Rauden-bush and Bryk, 2002).

RESULTS

The simple correlation matrix of the variablesincluded in the tests of the hypotheses appears inTable 1.

Tables 2–4 report the results of the hypothesistesting. All of the tables have the same format:Model 1 reports the models containing just thecontrol variables, Model 2 reports the main effectsof CEO under- and overpayment on the dependentvariables, and Model 3 reports the results pertain-ing to the moderating effect of ownership structureon these relationships. The measure of model fitis reported in the form of a likelihood ratio (LR)test comparing the more restrictive model to thenull model (model with no predictors) (Rauden-bush and Bryk, 2002).

Table 2 shows the results with regard to the rela-tionship between CEO underpayment and changein firm size. The results shown in Model 2 sug-gest that there is a significant positive relation-ship between CEO underpayment and change infirm size (p < 0.05), thus supporting Hypothesis1. Compared to CEOs paid more closely to thelabor market rate, significantly greater increasesin firm size are observed from underpaid CEOs.As Model 3 in Table 2 shows, however, own-ership structure does not moderate the relation-ship between CEO underpayment and change infirm size. Contrary to Hypothesis 4, the rela-tionship between CEO underpayment and changein firm size does not differ across ownershipstructures.

Hypothesis 2 predicted a positive relationshipbetween CEO underpayment and voluntary CEOwithdrawal. As Model 2 in Table 3 shows, thishypothesis is supported: there is a significant rela-tionship between CEO underpayment and volun-tary CEO withdrawal (p < 0.05), which suggeststhat underpaid CEOs are more likely to voluntar-ily withdraw from the CEO position than are CEOswho are paid more closely to the labor market rate.Model 3 in Table 3 shows that Hypothesis 5 isalso supported as ownership structure moderatesthis relationship as predicted: underpaid CEOs inOC firms are more likely to voluntarily withdrawthan are underpaid CEOs in MC (p < 0.001) andOM (p < 0.001) firms.

Finally, Table 4 reports the results of the testsof the relationship between CEO overpaymentand change in firm profitability. As Model 2shows, the results are consistent with Hypothesis 3:



Tabl

e1.

Mea

ns,

stan

dard

devi

atio

ns,

and

corr

elat

ions

for

vari

able

s

Var

iabl

esM

ean

s.d.

12

34

56

78

910

1112

1314

1516

1718

1920

1.C

EO

over

paym

ent

0.25

0.43

1.00

2.C

EO

unde

rpay

men

t0.

260.

43−0

.34∗

1.00

3.C

hang

ein

firm

size

−0.0

82.

680.

01−0

.01

1.00

4.C

hang

ein

firm

profi

tabi

lity

−0.1

819

.37

0.01

−0.0

30.

011.

00

5.C

EO

with

draw

al0.

050.

23−0

.02

0.06

∗0.

00−0

.04

1.00

6.Pr

ior

chan

gein

firm

profi

tabi

lity

1.33

38.7

60.

00−0

.02

0.03

0.00

0.00

1.00

7.Pr

ior

chan

gein

firm

size

0.00

1.82

0.01

0.02

−0.0

10.

000.

000.

001.

00

8.M

anag

er-c

ontr

olle

d0.

050.

220.

010.

000.

010.

000.

01−0

.01

0.00

1.00

9.O

wne

r-m

anag

ed0.

230.

420.

02−0

.02

−0.0

1−0

.03

−0.0

4∗−0

.02

0.03

−0.0

8∗1.

0010

.In

stitu

tiona

low

ners

hip

0.01

0.23

0.00

−0.0

6∗0.

05∗

−0.0

20.

000.

000.

00−0

.11∗

−0.2

0∗1.

0011

.D

irec

tor

owne

rshi

p0.

120.

16−0

.01

0.09

∗0.

010.

000.

030.

00−0

.04∗

−0.1

2∗0.

08∗

−0.2

7∗1.

0012

.D

ualit

y0.

690.

46−0

.02

−0.0

6∗0.

010.

00−0

.11∗

0.05

∗0.

030.

10∗

0.18

∗0.

06∗

−0.1

1∗1.

0013

.O

utsi

der

ratio

0.73

0.15

−0.0

8∗0.

01−0

.01

−0.0

10.

05∗

0.02

0.02

0.04

∗−0

.28∗

0.21

∗−0

.07∗

−0.0

6∗1.

0014

.R

esou

rce

avai

labi

lity

0.01

0.99

0.08

∗−0

.10∗

0.00

0.00

−0.0

20.

010.

00−0

.27∗

0.08

∗0.

04∗

0.08

∗−0

.08∗

−0.0

7∗1.

0015

.Fi

rmpr

ofita

bilit

y0.

010.

20−0

.06∗

0.04

∗−0

.01

−0.3

1∗−0

.04∗

0.02

0.01

0.04

∗0.

09∗

0.26

∗0.

030.

06∗

0.00

0.04

∗1.

0016

.Fi

rmsi

ze0.

101.

73−0

.09∗

0.10

∗0.

02−0

.01

0.02

0.01

0.05

∗0.

33∗

−0.1

7∗0.

27∗

−0.1

5∗0.

13∗

0.17

∗−0

.33∗

0.26

∗1.

0017

.To

tal

dive

rsifi

catio

n0.

170.

30−0

.03

−0.0

10.

01−0

.01

−0.0

1−0

.03

0.01

0.05

∗−0

.07∗

0.09

∗−0

.15∗

0.04

∗0.

07∗

−0.0

7∗0.

07∗

0.26

∗1.

0018

.In

cent

ive

pay

0.36

0.32

0.09

∗−0

.09∗

0.00

0.00

−0.0

2−0

.02

0.02

0.07

∗−0

.25∗

0.24

∗−0

.14∗

−0.0

4∗0.

16∗

−0.0

4∗0.

010.

11∗

0.03

1.00

19.

Nea

ring

retir

emen

t0.

160.

37−0

.02

−0.0

20.

02−0

.01

−0.0

9∗0.

06∗

−0.0

5∗−0

.05∗

0.15

∗−0

.10∗

0.00

0.14

∗−0

.15∗

0.01

0.03

−0.0

4∗0.

05∗

−0.1

7∗1.

0020

.C

EO

age

54.4

58.

47−0

.02

−0.0

5∗0.

030.

01−0

.10∗

0.05

∗−0

.12∗

0.03

0.09

∗−0

.07∗

−0.0

7∗0.

19∗

−0.0

8∗−0

.04

0.03

0.05

∗0.

12∗

−0.1

8∗0.

68∗

1.00

21.

CE

Ote

nure

8.27

8.17

−0.0

2−0

.01

0.01

−0.0

2−0

.23∗

0.02

−0.0

7∗0.

010.

34∗

−0.0

8∗−0

.10∗

0.28

∗−0

.21∗

0.07

∗0.

09∗

−0.0

6∗0.

01−0

.19∗

0.42

∗0.

44∗

∗p

<0.

05,

two-

taile

d.



Table 2. Results of the tests of the change in firm size hypothesesa

Variables Change in firm size

Model 1 Model 2 Model 3

Intercept −0.09∗ −0.11∗ −0.11∗

(0.05) (0.05) (0.05)

Resource availability 0.00 0.00 0.00(0.01) (0.01) (0.01)

Manager-controlled 0.04 0.03 0.11†(0.04) (0.05) (0.07)

Owner-managed −0.04 −0.05 −0.02(0.03) (0.03) (0.04)

Director ownership −0.13 −0.18 −0.03(0.19) (0.18) (0.17)

Institutional ownership 0.16 0.17 0.23(0.13) (0.13) (0.14)

Duality 0.08† 0.10∗ 0.08†(0.05) (0.05) (0.05)

Outsider ratio −0.13 −0.15 −0.12(0.14) (0.16) (0.14)

Total diversification −0.03 −0.01 −0.03(0.06) (0.06) (0.06)

Firm profitability 0.05 0.04 0.03(0.07) (0.07) (0.07)

CEO tenure 0.00 0.01 0.00(0.01) (0.01) (0.01)

Incentive pay 0.22 0.19 0.20(0.18) (0.16) (0.16)

Prior change in firm size −0.11 −0.05 −0.09(0.08) (0.03) (0.08)

CEO overpayment −0.08 −0.09(0.05) (0.06)

CEO underpayment 0.14∗ 0.15†(0.06) (0.08)

Slope of the relationship between CEO underpayment and change in firm size

Manager-controlled −0.16(0.10)

Owner-managed −0.11(0.12)

χ 2 30.53 96.04∗∗∗ 91.93∗∗∗

a Robust standard errors in parentheses.† p < 0.10; ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001, two-tailed.

there is a significant positive relationship betweenCEO overpayment and change in firm ROA (p <

0.05), suggesting that greater increases in ROAare observed from overpaid CEOs as compared toCEOs paid more closely to the labor market rate.Model 3 in Table 4 shows the results with regard toHypothesis 6, which predicted a moderating effectof ownership structure on the relationship betweenCEO overpayment and change in firm ROA. The

results show that the relationship between CEOoverpayment and change in firm ROA is weaker inMC firms compared to OC firms (p < 0.05), thussupporting Hypothesis 6. The results comparingMC firms to OM firms are only weakly supported(p < 0.10; two-tailed test); because the results arein the predicted direction, however, they suggestsupport for Hypothesis 6 based upon a one-tailedtest (p < 0.05).



Table 3. Results of the tests of the CEO voluntary withdrawal hypothesesa

Variables CEO withdrawal


Intercept −2.87∗∗∗ −2.82∗∗∗ −2.64∗∗∗

(0.17) (0.14) (0.11)

Resource availability −0.14∗∗∗ −0.13∗∗∗ −0.11∗∗∗

(0.03) (0.02) (0.02)

Manager-controlled −0.07 −0.05 0.39†(0.34) (0.21) (0.15)

Owner-managed −0.48∗ −0.39∗ −0.06(0.24) (0.16) (0.13)

Director ownership 0.11 0.04 −0.03(0.60) (0.43) (0.32)

Institutional ownership −0.72 −0.66∗ −0.53∗

(0.45) (0.32) (0.23)

Duality −0.66∗∗∗ −0.52∗∗∗ −0.45∗∗∗

(0.17) (0.13) (0.11)

Outsider ratio 0.45 0.34 0.18(0.61) (0.45) (0.33)

Firm profitability −0.44 −0.42∗ −0.38∗

(0.28) (0.21) (0.16)

Incentive pay 0.89∗ 0.84∗∗ 0.80∗∗∗

(0.35) (0.28) (0.21)

Firm size 0.13∗ 0.14∗∗∗ 0.12∗∗∗

(0.06) (0.04) (0.03)

CEO age −0.03∗ −0.03∗∗ −0.03∗∗

(0.01) (0.01) (0.01)

CEO tenure −0.37∗∗∗ −0.31∗∗∗ −0.22∗∗∗

(0.04) (0.03) (0.02)

Nearing retirement 0.15 0.37 0.51∗

(0.38) (0.28) (0.21)

CEO overpayment 0.15 0.12(0.14) (0.10)

CEO underpayment 0.27∗ 0.39∗∗

(0.13) (0.12)

Slope of the relationship between CEO underpayment and withdrawal

Manager-controlled −0.73∗∗∗

(0.21)

Owner-managed −1.01∗∗∗

(0.27)

χ 2 371.33∗∗∗ 388.93∗∗∗ 394.45∗∗∗

a Robust standard errors in parentheses.† p < 0.10; ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001, two-tailed.

DISCUSSION

The findings of the current study demonstrate animportant aspect of how the CEO labor mar-ket works—CEO pay that deviates too far aboveor below labor market rates has significant con-

sequences with regard to firm outcomes. As such,our findings contribute to the growing literaturethat has shown that CEO pay inequity not onlyimpacts the CEO compensation setting process, butalso that there are subsequent outcomes that areinfluenced by such deviations. Although previous



Table 4. Results of the tests of the change in firm profitability hypothesesa

Variables Change in firm performance


Intercept −0.28 −0.04 0.13(0.53) (0.43) (0.39)

Resource availability −0.15 −0.09 −0.10(0.11) (0.07) (0.07)

Owner-controlled 0.76 0.13 −0.04(0.50) (0.38) (0.34)

Owner-managed 1.25∗ 0.47 0.28(0.56) (0.46) (0.44)

Director ownership −0.71 −0.55 −0.55(0.65) (0.51) (0.52)

Institutional ownership −1.34† −0.80 −0.80(0.79) (0.66) (0.65)

Duality −0.50∗ −0.36 −0.36(0.25) (0.23) (0.23)

Outsider ratio 0.69 0.14 0.14(0.70) (0.67) (0.67)

Incentive pay 0.27 −0.01 −0.01(0.46) (0.35) (0.36)

Firm size 0.09 0.10 0.09(0.06) (0.06) (0.06)

CEO tenure 0.02 0.01 0.01(0.02) (0.02) (0.02)

Prior change in firm profitability 0.00 0.00 0.00(0.00) (0.00) (0.00)

CEO overpayment 0.67∗ −0.95(0.29) (0.83)

CEO underpayment −0.01 0.36(0.35) (0.40)

Slope of the relationship between CEO overpayment and change in firm profitability

Owner-controlled 1.67∗

(0.86)

Owner-managed 1.78†(1.08)

χ 2 15.15 303.34∗∗∗ 304.49∗∗∗

a Robust standard errors in parentheses.† p < 0.10; ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001, two tailed.

studies have examined how boards of directors(i.e., CEO compensation committee members)react to CEO pay deviations (Ezzamel and Wat-son, 1998; O’Reilly et al., 1988; Porac et al., 1999)and how the overpayment of CEOs affects lower-level managers’ pay and turnover (Wade et al.,2006), our study is among the first to investigatewhether CEO under- and overpayment has effectswith regard to CEOs themselves (see also Wat-son et al., 1996). We did so by examining whether

CEO pay deviations are related to firm outcomesthat would theoretically help CEOs to resolve theirown sense of equity.

Our results suggest that when there is CEO payinequity, as assessed by comparing CEO compen-sation to a compensation equation that representsthe CEO labor market rate for the CEO, CEOs maytake actions to resolve their own equity consider-ations. Organizational theories of equity and fair-ness suggest that people are motivated to maintain



fair relationships with others and to rectify unfairrelationships by making them fair. In the currentcase, this means that CEOs compare themselvesto other CEOs in similar conditions, and find-ing themselves overpaid or underpaid, take actionsto rectify this condition. The explanation is simi-lar to that put forth by Wade et al. (2006), whosuggest CEOs may be interested in resolving fair-ness issues concerning their own pay and otherpeople’s outcomes. By increasing firm profits orfirm size to a greater extent than do more fairlypaid CEOs, over- and underpaid CEOs, respec-tively, are following norms of fairness by rec-tifying their unfair relationship with those whoare ultimately responsible for the CEO’s compen-sation—shareholders. Because the growth of thefirm provides the CEO with intrinsic rewards, inaddition to extrinsic rewards and quite possibly atthe expense of shareholders, our findings suggestthat the situation may be made more equitable foran underpaid CEO through increasing firm size.Another way it appears that underpaid CEOs canresolve their dissonance is through the voluntarywithdrawal from their firms, presumably to pursuewhat they perceive to be more equitable alterna-tives. On the other hand, our results clearly suggestthat increasing firm profitability may make the sit-uation fairer for the overpaid CEO. Given thatthere is a clear expectation that CEOs, be theyowner-managers or hired professional managers,exert their efforts toward the interests of sharehold-ers, increasing their efforts toward firm profitabilityresolves dissonance in a manner congruent withCEO motivational needs.

Our findings with regard to the moderating effectof ownership structure further suggest that the dis-cretion to pursue managerial objectives plays a rolein these relationships. With regard to CEO volun-tary withdrawal, we found that underpaid CEOsin OC firms were more likely to withdraw thanwere CEOs in either MC firms, as the latter arenot constrained by a dominant owner, or CEOsof OM firms, who are the dominant owners them-selves. We found no differences across ownershipstructures, however, with regard to increases infirm size among underpaid CEOs. The main effectin this regard, nevertheless, still tends to confirmmanagerialist psychological contentions: it appearsthat increases in firm size, regardless of the degreeof discretion afforded CEOs, is a viable means forresolving underpayment inequity. Finally, in the

case of firm profitability, our findings are consis-tent with the notion that CEOs in MC firms mayreduce their overpayment dissonance through cog-nitive means more so than do CEOs in OC and OMfirms. To the extent that overpaid CEOs are con-flicted between self-interested and fairness-basedconcerns (Messick and Sentis, 1983; Peters et al.,2004), and because the latter concerns are likelyto weigh more heavily in the judgments of CEOsof OC or OM firms thereby resulting in behav-ioral responses (i.e., increased efforts), overpaidCEOs of MC firms may simply cognitively rede-fine their overpayment in a manner congruent withself-interested motivations. Our results were sup-portive: we found that MC firms with overpaidCEOs were less likely to increase firm profits thanOC or OM firms with overpaid CEOs.

The asymmetry in results between the under-and overpaid conditions, as well as these mod-erating effects of ownership structure, point tosome of the important insights that the equity lensbrings to the study of CEO compensation. Thatis, although all CEOs of publicly traded corpo-rations have compensation packages that tie theirfinancial rewards to firm profits, from a norms offairness standpoint the increasing of effort towardfirm profitability doesn’t resolve dissonance for anunderpaid CEO (i.e., where rewards are not req-uisite to inputs). Because the underpayment pay-ment condition already incorporates the presenceof pay-for-profits mechanisms, any increases ineffort toward such firm performance by the CEO inthis situation would not tend to increase outcomesin a manner that would resolve the CEO’s under-payment inequity. Thus, CEOs who are under-paid, instead, appear to focus their efforts towardincreasing firm size in an attempt to increase theirown outcomes (Marris, 1964; 1998; Tosi et al.,1999; Williamson, 1964) or toward pursuing otheremployment. Overpaid CEOs (i.e., where rewardsare more than requisite to inputs), in contrast, willexert more effort toward firm profitability becausesuch an action is congruent not only with normsof fairness, but also with the power and achieve-ment motivational needs likely to characterizeCEOs (Simon, 1947; Marris, 1964; Williamson,1964; McClelland and Boyatzis, 1982; Davis et al.,1997). In short, once equity considerations aretaken into account, the workings of CEO compen-sation mechanisms, such as incentive alignment,appear to be much more complex than portrayed



by economic accounts that rely solely upon extrin-sic motivations and rewards. Rather than simplybeing shirkers, our study suggests that CEOs fac-ing payment inequity will tend to increase theirefforts, but to what ends these efforts are directedseems to be a function of the inequity condition:when overpaid, CEOs direct their efforts towardshareholders’ interests, whereas when underpaid,CEOs don’t adhere to this conventional wisdom asthey direct their efforts toward their own interests.

Given our findings, it may be tempting to sim-ply conclude that much of the worry concern-ing excess CEO compensation is overblown. Afterall, isn’t the labor market (in tandem with fair-ness norms) working if overpaid CEOs are puttingforth effort toward increasing firm profits (and,furthermore, that underpaid CEOs are the onesmore likely to pursue managerial interests)? Weresist such an interpretation of our results. If noth-ing else, the social comparison processes appar-ently at work in the executive labor market maytend to have a ratchet effect on CEO compensa-tion—extant evidence suggests that CEO pay isalready being ‘ratcheted up’ through a continuouscycle of compensation committees increasing CEOpay in their effort to keep up with the CEO labormarket rate (Bebchuk and Fried, 2004; Ezzameland Watson, 1998). Thus, to interpret our find-ings here as a reason to overpay CEOs wouldonly further contribute to such a ratcheting effect,and such ratcheting can only lead to excessiveCEO pay from a societal standpoint (see Walsh,2008). Our findings with respect to the moder-ating effect of ownership structure only serve tofurther highlight the complexity involved in CEOcompensation in the face of equity concerns. Ifoverpaid CEOs with unconstrained discretion (i.e.,MC firms) redress their overpayment through cog-nitive means, then not all overpayment situationsare alike. It appears that the presence of finan-cial incentives is not the essential ingredient forincreased CEO effort—such rewards appeal to theself-interested side of the conflicted feelings thatan overpaid CEO may experience. Instead, what isrequired is a mechanism that triggers the fairness-based concerns of overpaid CEOs (i.e., the pres-ence of a dominant owner). In the end, our studyand its findings, along with previous research onCEO payment inequity (Wade et al., 2006), clearlypoint to the need for an increased understanding ofhow CEOs’ fairness concerns affect both the CEO

compensation setting process as well as other firmoutcomes.

Limitations and future research implications

When evaluating these results, nevertheless, thereare a few limitations to be kept in mind. Forthe most part, these pertain to the difficulties sur-rounding the measurement of the phenomena understudy. We do not measure dissonance directly.Instead we infer that CEOs recognize when theirpay deviates from the going labor market rate.Also, it is not such a leap to say that CEOscare about their compensation given the researchshowing they can, and may, influence consul-tants (Tosi and Gomez-Mejia, 1989), that orga-nizations should be concerned over the biddingaway of CEOs (Ezzamel and Watson, 1998; Fama,1980), and that CEOs are concerned with fair-ness issues surrounding pay (Wade et al., 2006).Future research that examines CEO perceptionsof the managerial labor market, if possible, andin particular that which investigates the processesunderlying the link between social comparisons,dissonance, and organizational outcomes wouldtherefore stand to greatly enhance our understand-ings of the workings of this labor market. We havealso embraced the assumptions, based upon previ-ous theory and research, that CEOs characteristi-cally possess relatively high power and achieve-ment needs (e.g. Simon, 1947; Marris, 1964;Williamson, 1964; McClelland, 1975; McClellandand Boyatzis, 1982; Davis et al., 1997), and more-over, that the dissonance felt by overpaid CEOsis a function of both self-interested and fairness-based concerns (Walster et al., 1978; Messick andSentis, 1983; Peters et al., 2004). Future researchthat could tap into the degree to which thesemotivations characterize CEOs of publicly tradedcompanies are thus clearly required to further ourunderstanding of CEOs’ fairness judgments, par-ticularly when they are overpaid.

Also noteworthy is that although our modelspecifications afforded a time lag in the tests ofthe hypotheses, as with any study utilizing archivaldata we can only infer causality from the currentfindings. Furthermore, our study’s design inves-tigated whether there was a relationship betweenCEO under- and overpayment and a selection ofsubsequent firm outcomes—we did not exam-ine the specific initiatives taken by these CEOsto address their dissonance. For instance, future



research in this regard could examine whetherunderpaid CEOs tend to employ particular strategicactions (i.e., diversification) to effect firm growth.Moreover, our choice to examine firm profitabil-ity, specifically ROA, was guided by previousresearch as well as because accounting-based indi-cators are more directly attributable to CEO actionsthan are other more market-based proxies (i.e.,shareholder returns, earnings per share; Fryxell andBarton, 1990). Thus, future research could investi-gate whether other measures of firm performanceplay a role in CEOs’ attempts to resolve pay-ment inequity. For example, MC firms have beenfound to be associated with income smoothing, andthe manipulation of earnings and accounting meth-ods (i.e., versus OC firms; for reviews see Hunt,1986; Tosi et al., 1999). Thus, there may be analternative means to that studied here by whichunderpaid CEOs of MC firms may resolve theirunderpayment condition—by increasing profitabil-ity via accounting manipulation, thereby increas-ing their outcomes without an increase in effort.For that matter, since CEOs of OC firms arealso hired managers, such manipulation of perfor-mance indicators also plausibly stands as a meansby which, relative to owner-manager CEOs, theytoo could affect their payment inequity. Futureresearch could therefore investigate whether or notthe manipulation of performance indicators servesas a mechanism by which CEOs faced with pay-ment inequity resolve their dissonance.

With respect to the measurement of CEO pay-ment inequity, while it is unknown as to whatconstitutes an under- or overpayment condition toCEOs themselves, it would appear that employ-ing some type of threshold when studying theeffects of CEO pay inequity is warranted giventhat perceptions of equity in this realm may involveidiosyncratically defined comparison groups (Poracet al., 1999) and that there may be a tendency tomake social comparisons to ‘others who are seenas slightly better or more expert’ (O’Reilly et al.,1988: 262). We followed previous research to mea-sure pay inequity via a CEO wage equation (Ezza-mel and Watson, 1998; Wade et al., 2006; Watsonet al., 1996)—which uses the ‘going market rate’as the referent for inequity. We captured under-and overpayment conditions as being those CEOspaid in the lower and upper quartiles of the sam-ple firms, respectively, thereby accounting for anypotential idiosyncratic, or upward, comparisons. To

the extent that CEOs personalize their social com-parison groups or choose to compare themselvesto those slightly more expert than themselves, afocus upon extremely under- or overpaid CEOs fortheir requisite abilities should compensate for suchoccurrences. For instance, the dissonance of CEOspaid in the upper quartile should not be alleviatedsimply by comparing themselves to slightly moreexpert peers. Given our findings, focusing uponthese extremes appears to serve as a good proxy,at least when it comes to the study of how CEOsattempt to resolve their own pay inequity. Indeed,we performed a post hoc analysis in which wereestimated the models using continuous measuresof under- and overpayment and the results werenot robust to this alternative specification. Thus,although it appears that some type of threshold iswarranted, future research is needed to develop abetter understanding of whether there is any poten-tial asymmetry in the limits to the threshold acrossthe two different inequity conditions (i.e., under-versus overpayment) as well as to understand thedegree to which idiosyncrasy plays a role in CEOsequity judgments.

Whether or not CEOs who are underpaid resolvetheir dissonance through changing the outcomes oftheir present situation via firm growth or throughvoluntarily withdrawing from their firms is, in part,a function of the desirability of the alternativesthat are available to each CEO. Thus, one of themajor limitations of the current study is that wewere not able to measure or evaluate these alterna-tives. Furthermore, although we have hypothesizedthat CEOs resolve their dissonance either throughincreasing firm size or voluntarily withdrawing, itis plausible that a third alternative exists that com-bines these two options: CEOs may initially optto grow their firms and subsequently decide towithdraw if this fails to reduce their dissonance.To investigate this possibility, we performed apost hoc analysis in which we examined whetheror not the relationship between increases in firmsize and CEO underpayment is stronger amongthose CEOs in the sample who stayed with theirfirms throughout the study period as compared tothose CEOs who voluntarily withdrew. The resultsof this post hoc analysis tend to support such athird scenario—while this relationship was pos-itive (p < 0.05) among those firms wherein theCEO stayed, the relationship was not significantamong those firms in which the CEO voluntarilywithdrew. Thus, given the longitudinal nature of



our data and analyses, these findings along withthose in support of Hypotheses 1 and 2 would sug-gest that the alternatives of staying and affectingoutcomes versus pursuing alternative situations arenot mutually exclusive over time and that both areviable options. In the end, although tapping intothe set of alternatives available to CEOs may bean intractable issue for any research effort, as thetrue nature of such an alternative set would seemunknowable to anyone other than to the CEOsthemselves, future research that could investigatethis is surely warranted.

Finally, related to this last issue is our oper-ationalization of CEO withdrawal—because ourhypotheses clearly pertain to voluntary withdrawal,we chose a measure that only considered thoseCEOs who left their position and also became adirector or retained a previously held directorshipon the firm’s board of directors (e.g., Shen andCanella, 2002). Thus, our study represents a verystrict test of voluntary withdrawal, and althoughthis only serves to strengthen our analysis, it doesnot examine those cases where CEOs voluntarilyleft the firm to take other opportunities outsideof the firm and did not also retain a directorship.Given our conservative approach, we performedseveral post hoc analyses to explore the poten-tial boundaries of our findings. The results withregard to the relationship between CEO underpay-ment and CEO voluntary withdrawal (Hypothesis2) and the moderating relationship of ownershipstructure to this relationship (Hypothesis 5) wererobust across all of the following post hoc tests:the reestimation of the models wherein CEO with-drawal excluded all CEOs who are near retire-ment (i.e., 63 years old or older); the reestimationof the models wherein CEO withdrawal excludedretirements identified through proxy statements;the reestimation of the models wherein CEO with-drawal included the CEOs who simply left the firmbut did not stay on the board; and the reestimationof the models wherein CEO withdrawal excludedthose withdrawals that occurred in firms perform-ing below the industry average. In total then, thefindings of this study support the notion that CEOsmay use withdrawal to resolve the dissonance cre-ated by an underpayment situation. Again, giventhat we did not capture the alternatives facedby exiting CEOs, future research directed towardinvestigating this issue are greatly needed to fur-ther develop an understanding of CEO voluntarywithdrawals.

CONCLUSION

The work of social comparison processes in theexecutive labor market has received increasingattention from researchers interested in understand-ing CEO compensation. The findings of our studycontribute to this understanding by suggesting thatCEOs may take actions to resolve their own equityconsiderations, and because the observed outcomesare in a manner consistent with managerial capital-ism and organizational theories of fairness, theseactions have significant consequences for CEOsand shareholders alike. Our findings clearly sug-gest that both of the outcomes associated with theunderpayment of CEOs tend to help CEOs resolvethis inequity, resolutions that come at the expenseof shareholders. Furthermore, this study suggeststhat a focus upon relative labor market pay in thestudy of CEO compensation, rather than the con-ventional approach of using firm performance asthe referent, provides additional insight into theexcess of CEO compensation. Indeed, it suggeststhat overpaying CEOs may have some beneficialeffect in that this condition results in increasedprofitability relative to that gained by CEOs whoare not overpaid. But, such a conclusion must alsoconsider the ratcheting effects in CEO compensa-tion that are very possible once we recognize theworkings of social comparisons and discretionaryprocesses within the labor market context. In theend, it appears that simply thinking about the mag-nitude of CEO pay or the alignment of CEO paywhen studying CEO compensation misses a crucialaspect of CEO self-interest: CEOs may also careabout their pay relative to the going rate in thelabor market. Furthermore, it seems that not allpayment inequity conditions are created equallyin this regard: while self-interested and fairness-based concerns are consistent in the underpaymentcondition, they present a conflict in the overpay-ment condition. When overpaid, CEOs appear tofollow their fairness-based concerns unless theyhave the discretion to indulge their self-interestedconcerns. Therefore, our study and its findings,which builds upon the growing amount of researchthat examines social comparison processes in CEOcompensation, suggest that more future researchinto how CEOs address their own pay inequity isnot only warranted, but can potentially shed muchlight upon the alignment of CEO and shareholderinterests.



ACKNOWLEDGEMENTS

The authors thank Editor Ed Zajac and two anony-mous referees for their guidance, comments, andsuggestions. We also thank James Algina, JasonColquitt, Heather Elms, Rodney Lacey, and WeiShen for their comments on earlier drafts.

REFERENCES

Adams JS. 1965. Inequity in social exchange. InAdvances in Experimental Social Psychology , Berck-owitz L (ed). Academic Press: New York; 267–299.

Amihud Y, Lev B. 1981. Risk reduction as a managerialmotive for conglomerate mergers. Bell Journal ofEconomics 12(2): 605–617.

Barnard CI. 1938. The Functions of the Executive.Harvard University Press: Cambridge, MA.

Bebchuk L, Fried J. 2004. Pay Without Performance:The Unfulfilled Promise of Executive Compensation .Harvard University Press: Cambridge, MA.

Berle A, Means G. 1932. The Modern Corporation andPrivate Property . Macmillan: New York.

Black F, Scholes M. 1973. Pricing of options andcorporate liabilities. Journal of Political Economy81(3): 637–654.

Bloom M, Milkovich GT. 1998. Relationships amongrisk, incentive pay, and organizational performance.Academy of Management Journal 41(3): 283–297.

Boeker W. 1997. Strategic change: the influence ofmanagerial characteristics and organizational growth.Academy of Management Journal 40(1): 152–170.

Bourgeois LJ. 1981. On the measurement of organiza-tional slack. Academy of Management Review 6(1):29–39.

Brockner J, Greenberg J, Brockner A, Bortz J, Davy J,Carter C. 1986. Layoffs, equity theory, and workperformance: further evidence of the impact ofsurvivor guilt. Academy of Management Journal29(2): 373–384.

Bryk AS, Raudenbush SW. 1989. Methodology forcross-level organizational research. Research in theSociology of Organizations 7: 233–273.

Buchholtz AK, Ribbens BA, Houle IT. 2003. The roleof human capital in postacquisition CEO departure.Academy of Management Journal 46(4): 506–514.

Cowherd DM, Levine DI. 1992. Product quality andpay equity between lower-level employees andtop management: an investigation of distributivejustice theory. Administrative Science Quarterly 37(2):302–320.

Cyert RM, March JG. 1992. A Behavioral Theory of theFirm (2nd edn.). Blackwell: Oxford, UK.

Dalton DR, Daily CM, Ellstrand AE, Johnson JL. 1998.Meta-analytic reviews of board composition, leader-ship structure, and financial performance. StrategicManagement Journal 19(3): 269–290.

Davis GF, Thompson TA. 1994. A social movementperspective on corporate control. AdministrativeScience Quarterly 39(1): 141–173.

Davis JH, Schoorman FD, Donaldson L. 1997. Towarda stewardship theory of management. Academy ofManagement Review 22(1): 20–47.

DeFond ML, Park CW. 1999. The effect of competitionon CEO turnover. Journal of Accounting andEconomics 27(1): 35–56.

Demsetz H. 1983. The structure of ownership and thetheory of the firm. Journal of Law and Economics 26:375–390.

Dun & Bradstreet. 1999. Reference Book of CorporateManagement. Dun and Bradstreet Information Ser-vices: New York.

Ezzamel M, Watson R. 1998. Market comparison earn-ings and the bidding-up of executive cash compensa-tion: evidence from the United Kingdom. Academy ofManagement Journal 41(2): 221–231.

Fama EF. 1980. Agency problems and the theory of thefirm. Journal of Political Economy 88(2): 288–306.

Festinger L. 1957. A Theory of Cognitive Dissonance.Stanford University Press: Stanford, CA.

Finkelstein S. 1992. Power in top management teams:dimensions, measurement, and validation. Academy ofManagement Journal 35(3): 505–538.

Finkelstein S, Boyd BK. 1998. How much does the CEOmatter? The role of managerial discretion in the settingof CEO compensation. Academy of ManagementJournal 41(2): 179–199.

Finkelstein S, Hambrick DC. 1989. Chief executivecompensation: a study of the intersection of marketsand political processes. Strategic Management Journal10(2): 121–134.

Fryxell G, Barton S. 1990. Temporal and contextualchange in the measurement structure of financialperformance: implications for strategy research.Journal of Management 16(3): 553–569.

Galaskiewicz J. 1985. Social Organization of an UrbanGrants Economy: A Study of Business Philan-thropy and Nonprofit Organizations . Academic Press:Orlando, FL.

Galaskiewicz J. 1997. An urban grants economyrevisited: corporate charitable contributions in theTwin Cities, 1979–81, 1987–89. AdministrativeScience Quarterly 42(3): 445–471.

Gomez-Mejia LR, Tosi HL, Hinkin T. 1987. Managerialcontrol, performance, and executive compensation.Academy of Management Journal 30(1): 51–70.

Greenberg J. 1988. Equity and workplace status: a fieldexperiment. Journal of Applied Psychology 73(4):606–613.

Greenberg J. 1989. Cognitive reevaluation of outcomesin response to underpayment inequity. Academy ofManagement Journal 32(1): 174–184.

Greenberg J. 1990. Employee theft as a reaction tounderpayment inequity: the hidden cost of pay cuts.Journal of Applied Psychology 75(5): 561–568.

Hambrick DC, Finkelstein S. 1987. Managerial discre-tion: a bridge between polar views of organizationaloutcomes. In Research in Organizational Behavior(Vol. 9), Staw BM, Cummings LL (eds). JAI Press:Greenwich, CT; 369–406.

Hambrick DC, Finkelstein S. 1995. The effects ofownership structure on conditions at the top: the case



of CEO pay raises. Strategic Management Journal16(3): 175–193.

Henderson AD, Fredrickson JW. 1996. Information-processing demands as a determinant of CEOcompensation. Academy of Management Journal39(3): 575–606.

Hicks JR. 1963. Theory of Wages . Macmillan: London,UK.

Hill CWL, Phan P. 1991. CEO tenure as a determinantof CEO pay. Academy of Management Journal 34(3):707–717.

Hill CWL, Snell SA. 1988. External control, corporatestrategy, and firm performance in research intensiveindustries. Strategic Management Journal 9(6):577–590.

Hill CWL, Snell SA. 1989. Effects of ownership structureand control on corporate productivity. Academy ofManagement Journal 32(1): 25–46.

Hofmann DA, Gavin MB. 1998. Centering decisions inhierarchical linear models: implications for researchin organizations. Journal of Management 24(5):623–641.

Hunt HGI. 1986. The separation of corporate ownershipand control: theory, evidence, and implications.Journal of Accounting Literature 5: 85–124.

Jacquemin AP, Berry CH. 1979. Entropy measure ofdiversification and corporate growth. Journal ofIndustrial Economics 27(4): 359–369.

Jensen MC. 1993. The modern industrial revolution, exit,and the failure of internal control systems. Journal ofFinance 48(3): 831–880.

Jensen MC, Murphy KJ. 1990. CEO incentives: it’s nothow much you pay, but how. Harvard Business Review68(3): 138–149.

Knight FH. 1921. Risk, Uncertainty, and Profit . HoughtonMifflin: Boston, MA.

March JG, Simon HA. 1958. Organizations . Wiley: NewYork.

Marris R. 1964. The Economic Theory of ManagerialCapitalism . Free Press: New York.

Marris R. 1998. Managerial Capitalism in Retrospect . St.Martin’s Press: New York.

McClelland DC. 1953. The Achievement Motive.Appleton-Century-Crofts.: New York.

McClelland DC. 1961. The Achieving Society . D. VanNostrand: Princeton, NJ.

McClelland DC. 1965. Toward a theory of motiveacquisition. American Psychologist 20: 321–323.

McClelland DC. 1975. Power: The Inner Experience.Irvington: New York.

McClelland DC, Boyatzis RE. 1982. Leadership motivepattern and long-term success in management. Journalof Applied Psychology 67(6): 737–743.

McEachern WA. 1975. Managerial Control and Perfor-mance. D. C. Health: Lexington, MA.

McEachern WA. 1978. Corporate control and growth: analternative approach. Journal of Industrial Economics26: 257–266.

Messick DM, Sentis KP. 1983. Fairness, preference, andfairness biases. In Equity Theory: Psychological andSociological Perspectives , Messick DM, Cook KS(eds). Praeger: New York; 61–94.

Miller DJ. 1995. CEO salary increases may be rationalafter all: referents and contracts in CEO pay. Academyof Management Journal 38(5): 1361–1385.

Miller D, Chen M. 1994. Sources and consequences ofcompetitive inertia: a study of the US airline industry.Administrative Science Quarterly 39: 1–23.

Mueller DC. 1972. A life cycle theory of the firm. Journalof Industrial Economics 20(3): 199–219.

Murphy KJ. 1985. Corporate performance and managerialremuneration: an empirical analysis. Journal ofAccounting and Economics 7(1–3): 11–42.

O’Reilly CA III, Main BG, Crystal GS. 1988. CEOcompensation as tournament and social comparison: atale of two theories. Administrative Science Quarterly33(2): 257–274.

Parrino R. 1997. CEO turnover and outside succession:a cross-sectional analysis. Journal of FinancialEconomics 46: 165–197.

Peters SL, van den Bos K, Bobocel DR. 2004. Themoral superiority effect: self versus other differencesin satisfaction with being overpaid. Social JusticeResearch 17(3): 257–273.

Porac JF, Wade JB, Pollock TG. 1999. Industry cate-gories and the politics of the comparable firm inCEO compensation. Administrative Science Quarterly44(1): 112–144.

Raudenbush SW, Bryk AS. 2002. Hierarchical LinearModels: Applications and Data Analysis Methods (2ndedn). Sage: Thousand Oaks, CA.

Rediker KJ, Seth A. 1995. Boards of directors and sub-stitution effects of alternative governance mechanisms.Strategic Management Journal 16(2): 85–99.

Roberts DR. 1959. Executive Compensation . Free Press:Glencoe, IL.

Salancik GR, Pfeffer J. 1980. Effects of ownership andperformance on executive tenure in U.S. corporations.Academy of Management Journal 23(4): 653–664.

Schumpeter JA. 1942. Capitalism, Socialism, and Democ-racy . Harper and Row: New York.

Shen W, Cannella AA. 2002. Power dynamics withintop management and their impacts on CEOdismissal followed by inside succession. Academy ofManagement Journal 45(6): 1195–1206.

Simon HA. 1947. Administrative Behavior . Free Press:New York.

Tosi HL, Gomez-Mejia LR. 1989. The decoupling ofCEO pay and performance: an agency theoryperspective. Administrative Science Quarterly 34(2):169–189.

Tosi HL, Gomez-Mejia LR. 1994. Compensation moni-toring and firm performance. Academy of ManagementJournal 37: 1002–1016.

Tosi HL, Gomez-Mejia LR, Loughry M, Werner S, Ban-ning K, Katz JP, Harris R, Silva P. 1999. Managerialdiscretion, compensation strategy, and firm perfor-mance: the case for the ownership structure. Researchin Personnel and Human Resource Management 17:163–208.

Tosi HL, Misangyi VF, Fanelli A, Waldman DA, Yam-marino FJ. 2004. CEO charisma, compensation,and firm performance. Leadership Quarterly 15(3):405–420.



Wade JB, O’Reilly CA, Pollock TG. 2006. OverpaidCEOs and underpaid managers: fairness and executivecompensation. Organization Science 17(5): 527–544.

Walsh JP. 2008. CEO compensation and the responsibil-ities of the business scholar to society. Academy ofManagement Perspectives 22(2): 26–33.

Walster E, Walster GW, Berscheid E. 1978. Equity:Theory and Research . Allyn and Bacon: Boston, MA.

Watson R, Storey D, Wynarczyk P, Keasey K, Short H.1996. The relationship between job satisfaction andmanagerial remuneration in small and medium-sizedenterprises: an empirical test of ‘comparison income’and ‘equity theory’ hypotheses. Applied Economics28(5): 567–576.

Weinzimmer LG, Nystrom PC, Freeman SJ. 1998. Mea-suring organizational growth: issues, consequencesand guidelines. Journal of Management 24(2):235–262.

Williamson OE. 1964. The Economics of DiscretionaryBehavior: Managerial Objectives in a Theory of theFirm . Prentice-Hall: Englewood Cliffs, NJ.

Zenger T. 1992. Why do employees only rewardextreme performance? Examining the relationshipsamong performance, pay, and turnover. AdministrativeScience Quarterly , 37: 198–219.



APPENDIX: USING A CEO WAGEEQUATION TO ESTIMATE CEOUNDER- AND OVERPAYMENT

Similar to previous research (Wade et al., 2006;Watson et al., 1996), a CEO wage equation wasconstructed that regressed total CEO pay on sev-eral organizational, human capital, and structuralvariables that theory and evidence suggest aredeterminants of CEO compensation. Thus, theresiduals from the regression represented CEOunder- and overpayment relative to the labor mar-ket rate. Our sample included 6,076 observa-tions over time within 1,342 CEOs—we used theBayesian estimation afforded in the HLM analyt-ical methodology, which allows for the use of allthe available information from CEOs with miss-ing data to develop residual scores for CEOs withfull rank data (Raudenbush and Bryk, 2002). Asexpected with such a large dataset, there was miss-ing data. Also, because we estimated CEO residu-als in each year, we were able to use those CEOswith only one year of data in the estimation ofresiduals. In essence, we were able to incorpo-rate all of the information contained in the ini-tial sample of CEOs (i.e., the data we initiallypulled before removing those instances that did notallow for lagged analysis and those with missingdata—and thus the difference between the sam-ple size here and the ones used for the tests ofthe hypotheses) to determine relative pay for thoseCEOs without missing data; although we were ableto use the information from CEOs with missingdata, no residual score was created for such CEOs.

Model specification

A two-level ‘random coefficients’ model was runto estimate CEO under- and overpayment in eachyear, which consisted of a firm level and anindustry-level (grouped by four-digit SIC code).As explained in the analytical method section, thismodel specification accounts for any industry labormarket effects (Bloom and Milkovich, 1998), andproperly models firms’ compensation committeesuse of comparable firms within the industry to setCEO pay (Porac et al., 1999).

Firm-level model

ln(CEO total pay)ij = β0 j + β1 j (F irm Size)ij

+ β2 j (F irm Prof itability)ij+β3 j (Owner − managed)ij + β4 j

(Manager − controlled)ij + β5 j (Outsider

Ratio)ij + β6 j (Duality)ij + β7 j (CEO Age)ij

+ β8 j (CEO T enure)ij + β9 j (Inside hire)ij+β10 j (CEO Experience)ij

+ β11 j (Location)ij + rij (A1)

Industry-level model

β0 j = γ00 + µ0 j (A2)

β1 j = γ10 + µ1 j (A2a)

β2 j = γ20 + µ2 j (A2b)

β3 j = γ30 + µ3 j (A2c)

β4 j = γ40 + µ4 j (A2d)

β5 j = γ50 + µ5 j (A2e)

β6 j = γ60 + µ6 j (A2f)

β7 j = γ70 + µ7 j (A2g)

β8 j = γ80 + µ8 j (A2h)

β9 j = γ90 + µ9 j (A2i)

β10 j = γ100 + µ10 j (A2j)

β11 j = γ110 + µ11 j , (A2k)

where the indices i and j denote firms and indus-tries with i = 1, 2, . . ., nj firms within indus-try j ; and j = 1, 2, . . ., J industries; all vari-ables are measured in a manner consistent withthat described in the methods section; CEO totalpay is measured consistent with previous research(e.g., Gomez-Mejia et al, 1987; Jensen and Mur-phy, 1990): it consists of CEO cash compensationplus all other types of rewards including long-term incentive pay, stock options, restricted andunrestricted stock grants, and deferred compen-sation. Stock options are valued using a modi-fied version of the Black-Scholes (1973) method,



which allows for the inclusion of dividend pay-ments (Murphy, 1985). All long-term contingentpay including stock options are valued in the yearthey were granted. Data were obtained from boththe Execucomp database as well as annual proxystatements; inside hire was measured as a dummyvariable equal to 1 if the CEO was an inside hireand 0 otherwise (e.g., Buchholtz, Ribbens, andHoule, 2003); CEO experience accounts for theCEO’s previous position within any given organi-zation (chairman, president, etc.); and finally, loca-tion was accounted for in the estimation becauseCEOs in close proximity may be very salient topay issues and more likely to make comparisons(Galaskiewicz, 1985, 1997). Location was mea-sured as a dummy variable such that CEOs in the

dataset were clustered based on proximity usinga 50 mile radius. The location of any given CEOwas based on the physical address of the CEO’sorganizational headquarters. From this clustering,locations were given a dummy coding based onhaving at least 10 CEOs within a cluster. Approx-imately 65 percent of the CEOs were located inone of the 17 location dummies specified using thismethod. For example, there were at least 10 CEOslocated within a 50 mile radius of Washington,D.C. based on the physical address of the CEOs’organizational headquarters. Because dummy vari-ables were used, the 35 percent of CEOs not inany given cluster were represented in the intercept(i.e., they make up the reference category).


Copyright of Strategic Management Journal is the property of John Wiley & Sons, Inc. / Business and its

content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's

express written permission. However, users may print, download, or email articles for individual use.

Documents

Pay and Equity