institutional ownership and monitoring effectiveness: it's not just how much but what else you...

OrganizationScienceVol. 19, No. 3, May–June 2008, pp. 419–440issn 1047-7039 �eissn 1526-5455 �08 �1903 �0419

informs ®

doi 10.1287/orsc.1080.0359© 2008 INFORMS

Institutional Ownership and Monitoring Effectiveness:It’s Not Just How Much but What Else You Own

Ravi DharwadkarManagement Department, Syracuse University, Syracuse, New York 13244, [email protected]

Maria GoranovaOrganizations and Strategic Management, University of Wisconsin, Milwaukee, Wisconsin 53201, [email protected]

Pamela BrandesManagement Department, Syracuse University, Syracuse, New York 13244, [email protected]

Raihan KhanManagement Department, State University of New York, Oswego, New York 13126, [email protected]

Corporate governance research indicates that large owners provide effective monitoring. In this article, we expandfirm-level notions of monitoring to include large institutional owners’ investment portfolios and suggest that portfolio

characteristics affect owners’ motivation and capacity to monitor, which compromises the positive effects of monitoringat the firm level. Specifically, using data from 533 large firms over a 10-year period, we find that increases in the sizeof portfolio holdings, number of portfolio blockholdings, portfolio turnover, and the importance of a particular holdingreduce monitoring effectiveness in the context of executive compensation. Overall, we provide preliminary evidence that theportfolio characteristics of the largest institutional owners contradict firm-level monitoring effects; therefore, we stronglyrecommend that future studies consider both firm- and portfolio-level effects simultaneously to understand monitoringeffectiveness.

Key words : large owners; institutional investors; executive compensation

Ever since Berle and Means (1932) examined the im-plications of dispersed owners and self-interested man-agers using the separation of ownership and controlthesis, monitoring by large owners has been consid-ered an important governance solution to agency prob-lems (Shleifer and Vishny 1997). Large owners aremore likely to assume monitoring costs than free-ridingsmall investors, particularly when the benefits of moni-toring outweigh the costs of monitoring and allow largeowners to recoup their investments (Gillan and Starks2000, Shleifer and Vishny 1986). Research has exam-ined the benefits of large-owner monitoring at the firmlevel across several contexts including firm valuation(Thomsen and Pedersen 2000), productivity (Hill andSnell 1989), corporate strategy (Amihud and Lev 1981),and executive compensation (Hartzell and Starks 2003).In fact, Demsetz (1986) suggests that monitoring bylarge owners is more “robust and continuous” than is thedisciplining effect of the market for corporate control.

Prior research conceptualizes large-owner monitoringat the firm level in two ways. First, the “ownership con-centration” perspective considers the ownership stakes ofthe largest owner, the largest five owners, owners with1% of equity, and/or owners with 0.2% or more of thefirm’s equity (Pedersen and Thomsen 2003, Hartzell andStarks 2003, Tihanyi et al. 2003, Pollock et al. 2002).

For example, Hartzell and Starks (2003) find an associ-ation between the ownership of the top five institutionalowners and both reduced total executive compensationand increased pay for performance sensitivity, in sup-port of the effectiveness of firm-level monitoring. Sec-ond, the “blockholder” perspective suggests that ownersmust have at least a 5% stake (typically dummy coded)to be effective monitors who can influence compensation(Agrawal and Mandelker 1990, Core et al. 1999, Mehran1995). Along these lines, Beatty and Zajac suggest thatblockholders, considered better monitors than small hold-ers, are associated with a “lower proportion of total man-agerial compensation derived from non-cash incentives”(1994, p. 328). In summary, research on monitoring bylarge owners either explicitly or implicitly considers theimplications of ownership stakes at the firm level withoutacknowledging owners’ needs or capabilities to monitorthe many firms within their portfolios.

Over time, institutional ownership has largely sup-planted individual and family ownership and accounts forthe majority of all ownership (Daily et al. 2003, Healy andPalepu 2003). Reiterating this trend, Hoskisson and col-leagues (2002, p. 698) state: “Ownership concentrationhas been identified as an important tool to curtail a man-ager’s propensity to pursue inefficient strategies � � � � Inparticular, institutional investors have emerged as major

419

Dharwadkar et al.: Institutional Ownership and Monitoring Effectiveness420 Organization Science 19(3), pp. 419–440, © 2008 INFORMS

equity owners and thus key players in corporate gover-nance.” Despite this cognizance, research associates mon-itoring effectiveness with large institutional ownershipstakes at the firm level and ignores the implications ofowners’ large portfolios. That is, large institutional own-ers tend to be equated with family or other large individ-ual blockholders, who often maintain a long-term, con-centrated presence in few firms and whose fortune isclosely linked to their monitoring of those firms (Ander-son and Reeb 2003). As early as 1976, Drucker noted thatinstitutional investors are not typical “owners”; instead,they are investors who hold well diversified portfolioswith stock in thousands of companies simultaneously. Adecade later, Demsetz (1986) suggested that institutionalowners do not specialize their portfolios the way thatindividual and family owners do, casting doubt on theirmotivation to effectively monitor firms. Hendry and col-leagues (2006) recently reiterated this point by suggest-ing that both institutional investors and executives viewinstitutional investors as traders, whose main interest isto maximize portfolio profits while maintaining portfolioliquidity, rather than as owners who attempt to improvecorporate governance.

Although theoretical research recognizes the impor-tance of both firm and portfolio implications of institu-tional ownership on the need and capability to monitor(Ryan and Schneider 2002), no empirical research thatwe are aware of examines this issue. For example, con-sider the portfolio characteristics of three well knownmutual funds, Wellington Management, Fidelity Manage-ment (refer to Table 1), and Vanguard. In 2000, Welling-ton Management was the largest institutional owner inPenford Corp. (9.85%), but the average firm holdingof Wellington Management ($81 million) was approxi-mately eight times more important than their $9.8 millionfirm-level stake in Penford. Similarly, in 2001, Fidelitywas the largest owner in Tidewater Inc. (9.07%) but alsohad more than 2,500 firms in its portfolio, of which

Table 1 Firm Ownership and Portfolio-Level Characteristics: Examples

Company Penford Corp. Tidewater Inc. CPI CorporationYear 2000 2001 2000

Ownership stake of the largest 9.85 ($9.8 M) 9.07 ($172.1 M) 18.31 ($29.3 M)institutional owner (%)

Largest institutional owner Wellington Management Co. Fidelity Management & Newsouth CapitalResearch Company Management Inc.

Portfolio concentration $81 M $226 M $12 MPortfolio blockholding (%) 22 45 38Portfolio turnover (%) 10 18 9Firm significance in portfolio (%) 51 83 79Largest stake ($) in institutional Citigroup Inc. ($3.7 B) Microsoft ($16 B) Metris Cos ($111 M)owner’s portfolio

Largest stake (%) in institutional Pacific Union Bank Focal Communications Corp. CPI Corp.owner’s portfolio (42.6%) (21.4%) (18.31%)

Number of companies in 1,713 2,535 118institutional owner’s portfolio

Number of larger investments ($) 837 421 25in institutional owner’s portfolio thanthe focal company

421 investments were greater than the one in Tidewater.Finally, in the year 2000, Vanguard had 43 unique invest-ments with a value greater than $1 billion and 296 otherinvestments above $100 million within its portfolio. It isunlikely that these investors exert the same time and effortin monitoring all of their investments.

This article develops the monitoring construct furtherby expanding the notion of monitoring to include insti-tutional owners’ portfolio characteristics. For example, ifthe largest institutional owner’s portfolio consists of manylarge holdings and/or blockholdings, these holdings maydemand the monitoring attention of owners and detractfrom monitoring effectiveness at the firm level. Alterna-tively, if institutional owners have higher preferences forportfolio liquidity (i.e., portfolio turnover), their moni-toring effectiveness should decrease because they facedeclining incentives to monitor because of their propen-sity to use exit. Finally, if large institutional owners paymore attention to their larger holdings, they may com-promise their monitoring of smaller holdings. Hence, ourresearch question is as follows:How do large institutionalowners’ firm-level ownership and their portfolio charac-teristics influence monitoring effectiveness?

We might answer this question by considering a rangeof organizational issues (e.g., valuation, productivity,corporate strategy, research and development, corporaterestructuring, etc.), but instead we focus on executivecompensation, which clearly represents an area of activeinterest on the part of the institutional investors. Theseowners have become more vocal about the forms andlevels of executive compensation during the past twodecades through shareholder resolutions, proxy voting,and compensation guidelines advocacy (Karpoff et al.1996).1 Both anecdotal evidence and empirical researchfind that large institutional owners have taken the leadin mitigating agency conflicts with regard to executivecompensation (Bebchuk and Fried 2004). Despite theevidence related to the monitoring effectiveness of large

Dharwadkar et al.: Institutional Ownership and Monitoring EffectivenessOrganization Science 19(3), pp. 419–440, © 2008 INFORMS 421

owners in particular and institutional owners in general,many researchers find that agency predictions regardingexecutive compensation do not hold true or that com-pensation and performance relate only weakly (Bebchukand Fried 2004, Jensen and Murphy 2004, Yermack1995). To understand this complex phenomenon, webelieve that we must account for the portfolio char-acteristics of large owners. Specifically, whereas exist-ing research focuses on the ownership benefits at thefirm level (which are fairly obvious and well docu-mented), it ignores that institutional owners possessstakes in hundreds and even thousands of firms and thusmay face limited incentives to monitor. By assessingan owner’s portfolio characteristics, we may be able toshed more light on the ownership–executive compensa-tion relationship.

Research ContributionsOur research question is important from both theoreticaland practical perspectives. From a theoretical standpoint,we expand current notions of monitoring effectivenessby considering not only firm- but also portfolio-levelfacets, in line with the recommendations of Ryan andSchneider (2002) and Dalton and colleagues’ (2003)suggestion that we still know little about the implicationsof an institutional owner’s portfolio in terms of mitigat-ing agency problems, despite the increasingly dominantrole of institutional investors. Our framework differ-entiates the monitoring effectiveness of large investorson the basis of their ability to monitor a focal firmand their incentives to monitor, as reflected by theirportfolio structure. Thus, from a theoretical perspec-tive, we suggest that monitoring effectiveness dependson both firm- and portfolio-level variables and focus onignored aspects of ownership, namely, what else an insti-tutional owner owns in terms of its portfolio holdings.

For practice, the portfolio implications of monitor-ing are important not only from the point of view ofindividual firms, financial intermediaries, investors, andregulators but also with regard to the general demandfor stronger corporate governance. If portfolio charac-teristics offset firm-level monitoring effectiveness, largeinstitutional investors should consider how to developthe necessary monitoring capacity in their portfolios ordevelop market intermediaries to provide them with suchservices. Current securities laws generally discouragelarge ownership positions, as well as coordination among“voting groups” of owners (Coffee 1991), which mayhave encouraged the fragmentation of owners’ portfo-lios and reduced their “voice” related to monitoring.Because the number of stocks required for suitable diver-sification is much lower than that observed in manyinstitutional owners’ portfolios (Pozen 2002), it makessense for interested parties to pay attention to tradeoffsbetween diversification and monitoring.

HypothesesLarge Owners and Monitoring at the Firm LevelWe begin by revisiting the conventional logic sup-porting the effects of large owners at the firm level.Large owners can monitor agents effectively and reduceagency costs because of their higher stakes and rela-tively lower coordination costs compared with more dis-persed owners whose monitoring costs are prohibitivelyexpensive (Shleifer and Vishny 1986). Large owners canbear the costs of monitoring compared to small own-ers, because their potential returns from monitoring canexceed the costs (Gillan and Starks 2000). In the wordsof Conyon and Peck (1998, p. 150) who paraphraseHambrick and Finkelstein (1995), “widely dispersedshareholders � � �may have little incentive to monitormanagement � � � and shareholders may not optimallyexercise their corporate control responsibilities � � � � Thepresence of a large shareholder � � �may result in improvedmonitoring, vigilance, and corporate governance.” Largeowners may have to use “voice” and monitor manage-ment because they lack the ability to divest themselvesof their investments en masse because the sale of largeblocks of equity could lead to a substantial drop in thestock price (Coffee 1991, Pozen 2002). Finally, largeowners can exploit the power they gain from their own-ership stakes to monitor effectively (Kochhar and David1996). In summary, both theory and empirics suggestthat large owners are effective monitors (Shleifer andVishny 1997).

Large Institutional Ownership andExecutive CompensationThere are two approaches that consider the effects oflarge owners (including blockholders) on executive com-pensation. The first approach views large owners aseffective monitors who can act as a check on pay levelsand ensure that management does not expropriate rentsfrom shareholders in the form of greater compensation.This approach assumes that large owners can ensureex ante that gains in CEO wealth from incentive com-pensation will be linked to shareholder value creation.These studies link large-owner monitoring to (1) thelevel of compensation (e.g., Core et al. 1999, Davidet al. 1998, Hartzell and Starks 2003) and (2) pay forperformance sensitivity (e.g., Hartzell and Starks 2003,Yermack 1995). The second approach views large-ownermonitoring and the use of incentive alignment as sub-stitutes (Rediker and Seth 1995). In this approach, thelower use of incentive alignment necessitates monitoring(Beatty and Zajac 1994, Zajac and Westphal 1994) orthe presence of large owners reduces the need for highproportions of incentive compensation (Mehran 1995,David et al. 1998). These studies have demonstratedlinks between large owners (or blockholders) and themix of compensation (e.g., Bloom and Milkovich 1998,Beatty and Zajac 1994, Mehran 1995).


In introducing our first hypothesis, we recap existingtheory and empirical research in the context of largeinstitutional ownership. We expect that large institutionalowners will be associated with lower levels of compen-sation and higher levels of pay for performance sen-sitivity (Hartzell and Starks 2003). Previous researchsuggests that the presence of large owners is nega-tively related to pay levels but has not focused explic-itly on institutional ownership (Dyl 1988, Hambrick andFinkelstein 1995). Extending this logic, Hartzell andStarks (2003) find that large institutional owners areassociated with reduced levels of total compensation,consistent with the findings of large ownership in gen-eral. In addition, we consider another important aspectof executive compensation: the sensitivity of pay toshareholder wealth creation (Hambrick and Finkelstein1995, Hartzell and Starks 2003, Jensen and Murphy1990, Yermack 1995). Hambrick and Finkelstein (1995)find that externally controlled firms with large ownershave greater pay sensitivity. More recently, Hartzell andStarks (2003) indicated that large institutional owners,as more effective monitors, encourage greater pay forperformance sensitivity such that CEOs receive greaterrewards for improving shareholder value. Because stockoptions represent the greatest portion of incentive com-pensation, we argue that large owners will prefer tighterrelationships between stock option pay and shareholderwealth creation. Hence, the literature suggests that mon-itoring resulting from large institutional ownership isassociated with fewer agency conflicts, leading to lesstotal executive compensation and increased pay for per-formance sensitivity.

The second approach (i.e., the substitution perspec-tive) suggests that monitoring and incentive alignmentare intricately linked and that this has implications forpay mix (Roth and O’Donnell 1996). Under optimalcontracting, executives will receive more incentives untilthe incremental cost of these incentives exceeds the ben-efits derived from them (Bebchuk et al. 2002). How-ever, in the presence of monitoring by large owners, thisthreshold will be reached earlier than in the absence ofmonitoring by large owners. Holmstrom (1979) arguesthat even imperfect information gained from monitoringcan benefit principals and decrease their losses. If largeowners can effectively monitor agents, they can reducetheir reliance on incentives even with imperfect informa-tion. Because incentive compensation generally is morecostly because it has to compensate executives for addedrisk, monitoring by large owners should be adverselyrelated to the use of incentive compensation as part ofthe pay mix.2 Zajac and Westphal (1994), using the sub-stitution argument, state: “firms facing a more severeincentive problem (due to CEOs having lower incentivesfrom their compensation contracts and equity holdings)are more likely to have governance structures that pro-vide a higher level of monitoring of managerial behav-ior.” Other studies find that large owners are associated

with a lower use of incentives as a proportion of totalcompensation (David et al. 1998, Mehran 1995). Thesefindings are consistent with Eisenhardt’s (1989) asser-tion that principals capable of monitoring managementprefer to use more behavioral-based compensation (e.g.,greater proportion of salary in the pay mix) and lessoutcomes-based compensation (e.g., lower proportionof stock options and other allied incentives). Applyingthese arguments, monitoring resulting from large insti-tutional ownership will be associated with greater use ofbehavioral-based compensation (e.g., greater proportionof salary in the pay mix) and less outcomes-based com-pensation (e.g., lower proportions of incentives in thepay mix).

Hypothesis 1. Large institutional ownership at thefirm level associates positively with monitoring; specif-ically, the percentage ownership of the largest ownerrelates negatively to the level of total compensation, pos-itively to pay for performance sensitivity, positively tothe salary proportion of the pay mix, and negatively tothe contingent proportion of the pay mix, ceteris paribus.

Large Institutional Ownership and PortfolioConsiderationsLittle research investigates how institutional owners’portfolios may enable, or compromise, monitoring at thefirm level. Researchers implicitly assume that large insti-tutional owners can develop organizational structuresand capabilities to handle the complexity of managingtheir diverse portfolios. First, in terms of organizationalstructures, as the number of companies in an owner’sportfolio increases, the institutional owner could recruitmore fund managers and provide them with attractiveincentives to ensure the effective management of eachfund. Second, in terms of organizational capabilities,large institutional investors could increase their moni-toring expertise beyond that of the average investor byemploying professional portfolio managers and provid-ing them with resources and incentives to effectivelymanage investments (Agrawal and Mandelker 1992,Almazan et al. 2005). Large institutional investors withbetter investment performance should be more skill-ful monitors. In conclusion, large institutional ownersmay possess economies of scale related to monitor-ing because of their organizational structures and bettercapabilities as reflected in their portfolio performancecompared with small individual investors (cf., Ryan andSchneider 2002).

However, the average institutional owner in our studymaintains approximately 2,000 firms within its portfolioof investments. From a theoretical perspective, the orga-nizational structures and capabilities of these investorslead to an intriguing question: Are large institutionalowners able to supply adequate monitoring consideringthe demands made by such a multiplicity of investments?


In answering this question, we suggest that three factorsinfluence these demands from a portfolio perspective:(1) the size of portfolio investments and block invest-ments, (2) the propensity of the institutional investorto exit, and (3) the importance of a particular holdingwithin the investor’s portfolio.

Portfolio Holdings and Monitoring EffectivenessThe tendency has been to liken large institutional own-ers to family or other large individual blockholders whomaintain a longer-term concentrated presence in onlya few firms (Anderson and Reeb 2003). However, themultitude of investments in their portfolios may compro-mise large institutional owners’ monitoring at the firmlevel, in that they may not be able to provide the sameintensity of monitoring to each investment provided byless diversified individual or family investors. Both anec-dotal and empirical evidence indicates that institutionalowners invest in the truest sense and do not alwaysact as “owners” (Demsetz 1986, Coffee 1991, Hendryet al. 2006). For example, Coffee (1991, p. 1291) pointsout “indexed investors, who may hold securities in athousand or more corporations, have largely abandoned‘exit,’ but also find the exercise of ‘voice’ infeasiblebecause their extensive holdings exceed their capacity tomonitor.” Hendry and colleagues (2006, p. 1111) demon-strate this investor orientation of large institutional own-ers in their interviews with company managers and theirinstitutional investors, as the following summary of com-ments of a fund manager of a large British institutionalinvestor attests:

Even where the investor was a dominant shareholder itwas a trading model that was critical, “We had 10% ofthe company and when you got there, you just have toknow the company a hell of a lot better. I don’t thinkwe would be interested in meeting the company to try toand work out whether we thought the managing director’sany good. What we are really trying to do is, we want tobe absolutely sure about what we worked out from ournumbers � � �was effective, correct.” This interviewee wenton to talk of due diligence but this was not the due dili-gence of an owner. It was the due diligence of an investorwho had made a big bet with clients’ money to make surethe information underlying that bet was accurate [empha-sis added]. For other investors working out whether themanagement was any good was precisely what they wereinterested in, but only because, in their cases, that wasan important component of their trading models.

Given this investor orientation, several factors mightreduce monitoring effectiveness because of competingpriorities in the portfolio, as well as restricted monitor-ing resources: (1) goal incongruence within the portfolioamong different types of funds, (2) cost considerationsof different types of funds, (3) lack of incentives to fundmanagers to monitor, (4) legal ramifications that mayrestrict monitoring, and (5) an extensive focus on prob-lem firms within the portfolio. In other words, certain

portfolio characteristics may demand more monitoringby large institutional owners, but the monitoring supplymay be inadequate.

As far as competing priorities are concerned, port-folios consist of funds with varying objectives. Morn-ingstar, an independent firm, categorizes funds by threepossible levels of capitalization (large, mid, and small)and three investment objectives (value, blend, andgrowth), which yields nine possible combinations (Pozen2002). Different fund objectives within the portfoliocan conflict, leading to inefficient uses of monitoringresources, whereas less diversified blockholders shouldhave less conflict and more homogenous monitoring.

Second, fund fees and fund staffing may not coverthe costs of monitoring. The fee structure of some fundsguarantees little monitoring; for example, the advisoryfees of index mutual funds are approximately 20 basispoints (0.20%) per year, whereas fees for venture capitalfunds are around 1%–2% plus 10%–20% of all profits(Pozen 2002, p. 513). High fund fees prompt more activemonitoring whereas low fund fees require fund man-agers to bear the cost of monitoring. These cost implica-tions also occur in pension funds that often “have only askeleton in-house staff, which then hires several externalmoney managers, and the function of the in-house staffis basically to evaluate the performance of these fundmanagers” (Coffee 1991, p. 1342). Even institutionalactivism literature emphasizes cost considerations—theannual activism program budget for the top five pen-sion funds ranges from $50,000 to $1 million, amount-ing to less than 0.005% of fund assets (Del Guercio andHawkins 1999, p. 328).

A third factor contributing to the dissipation of effortat the portfolio level is the general lack of incentivesto fund managers to improve their monitoring efforts.Coffee (1991, p. 1326) states that investment managersare often compensated on the basis of an annual fee“equal to a declining percentage of the fund it manages”such as one-third of 1% of the first $500 million, onequarter of 1% of the next $250 million, and so forth.Such schemes (often used by mutual funds and pensionfunds) may provide insufficient incentives to encouragemonitoring. Woidtke (2002, p. 103) finds that private andpublic pension fund managers fare worse; in 1993 theaverage private pension fund manager who managed afund of at least $1 billion in assets received base pay of$101,700, only 60% of these funds paid bonuses, and theaverage bonus paid was $25,600. The figures for publicpension fund managers indicate a $75,200 base salaryfor funds over $1 billion, a paltry 7% of these funds paidbonuses, and an average bonus paid of $8,400. Unlessthere is clear link between monitoring and immediateimprovements in fund performance, fund managers mayhesitate to spend time on monitoring activities becauseincentive systems fail to encourage them to do so.


Fourth, the specific legal and regulatory environ-ments of institutional owners may limit monitoring andactivism further. Private pension funds are constrainedin their exercise of control over firms in their portfoliounder the Employee Retirement Income Security Act;attempts to influence management could lose plans theirtax-exempt status (Ryan and Schneider 2002). Similarly,insurance firms who play dual roles of equity holder andbondholder may compromise monitoring because suchactivity would affect their claims as bondholders (Ryanand Schneider 2002). Finally, institutional activism lit-erature indicates that institutional owners are active insome companies in their portfolios. The choice to beactive or to monitor actively may relate to poor per-formers in the portfolio or could be symbolic in nature(Carleton et al. 1998, Wahal 1996). Excessive focus onhigh-profile poor performers may distract attention fromthe rest of the portfolio.

These arguments suggest that monitoring supply maybe limited, in comparison with monitoring demands.Thus, portfolio characteristics may compromise moni-toring at the firm level for several reasons. Large insti-tutional owners may be less effective monitors whenthe average holdings in their portfolios become largerdue to the sheer size of the total investment portfolio ormore blockholdings. As the average stake increases, thedemand for monitoring these large investments increasesand therefore limits monitoring effectiveness based onthe ownership at the firm level. Returning to the caseof Penford Corp., Wellington Management’s stake of9.85% (i.e., $9.8 million), though large at the firmlevel, is minor compared with the average holding inWellington’s portfolio ($81 million), thereby diminish-ing the incentives of the institutional owner to moni-tor. Similar arguments exist regarding the large blocksheld in the institutional owner’s portfolio. Because theselarge owners are most likely to benefit from monitor-ing and are unlikely to resort to free-riding, they willhave to monitor a larger number of blocks in their port-folio, which will again stretch monitoring. As an illus-tration, consider an institutional investor holding 50%of its portfolio in blocks of 5% compared with anotherinstitutional investor that holds 10% of its portfolio inblocks of 5%. These investors have different monitoringdemands. Assuming that greater blockholdings requiremore monitoring, the benefits of firm-level monitoringget compromised as other portfolio firms vie for moni-toring attention. Thus, based on our arguments regard-ing the supply and demand for monitoring, the portfoliocharacteristics of large institutional owners will detractfrom their monitoring effectiveness as large owners andat the same time will offset the benefits of monitoringfrom the substitution perspective.

Hypothesis 2. Institutional ownership portfolio char-acteristics associate negatively with monitoring effec-tiveness; specifically, average portfolio holdings and the

proportion of blockholdings in the portfolio of the largestowner relates positively to the level of total compensation,negatively to pay for performance sensitivity, negativelyto the salary proportion of the pay mix, and positively tothe contingent proportion of the pay mix, ceteris paribus.

Thus far, our arguments about the implications of port-folio characteristics have been predicated on the assump-tion that institutional owners want to monitor the largerinvestments in their portfolios but that their concen-trated portfolios, containing many large holdings vyingfor monitoring attention, take a toll on their monitoringeffectiveness. However, there is another choice availableto these investors. As Hirschman (1970) elegantly statesin his book Exit, Voice, and Loyalty, exit provides aviable alternative for many institutional investors. Eakinset al. (1998) suggest that institutional investors preferfirms with greater stock turnovers, which implies thatexit (i.e., portfolio liquidity) becomes an important char-acteristic that investors seek in choosing their portfoliofirms. Institutional owners such as mutual funds, banks,and insurance companies often need greater liquidity,because their shareholder depositors or policyholderscan leave at a moment’s notice. In the early 1990s, theturnover rates for 73% of institutional investors over thecourse of a year ranged from 26% to 200% (Coffee1991, p. 1340). Although this rate has declined in recentyears, the decline is attributable more to the emergenceof index funds with their emphasis on lower fee struc-tures than to investors’ decisions to take on additionalmonitoring responsibilities (Coffee 1991, p. 1340).

Higher levels of portfolio turnover also indicate insti-tutional owners’ lower preferences for monitoring. Mon-itoring is expensive, requires a good deal of effort, hasuncertain outcomes, and provides decreasing marginalreturns (Tosi and Gomez-Mejia 1994). A comprehensiveresearch summary of the monitoring benefits associatedwith ownership activism suggests that such monitoringcreates few meaningful differences at the firm level interms of profitability or even shareholder value creation(Karpoff 2001). Given the uncertain short-term benefitsof monitoring, institutional investors with high portfo-lio turnovers may be reluctant to invest in monitoringcapacity at the portfolio level. Using exit strategies whendealing with some companies in their portfolio does notnecessarily mean that the institutional owner will not usevoice for other firms, but it does mean that their monitor-ing capacity is compromised by the tradeoffs they makewith respect to portfolio turnover. On one hand, a largeinstitutional owner (i.e., with a firm-level stake of 6%)with an annual portfolio turnover of 6% is more likelyto invest in monitoring because it should benefit from itsmonitoring investment. On the other hand, an investorwith the same firm-level stake that has an annual port-folio turnover of 60% is less sure about its continuity inthe firm and is less likely to benefit from the spilloverof its monitoring capabilities developed from monitor-


ing other firms in its portfolio. Therefore, it is likelythat investors with high portfolio turnover will not put inresources to develop monitoring capabilities at the port-folio level and thus compromise monitoring benefits atthe firm level and offset the benefits of monitoring fromthe substitution perspective.

Hypothesis 3. Institutional ownership turnover atthe portfolio level associates negatively with monitoringeffectiveness; specifically, the portfolio turnover of thelargest owner relates positively to the level of total com-pensation, negatively to pay for performance sensitivity,negatively to the salary proportion of the pay mix, andpositively to the contingent proportion of the pay mix,ceteris paribus.

We also turn our attention to the implications of thesignificance of the focal firm within the largest insti-tutional investor’s portfolio. Some portfolio structuresmay be consistent with large ownership monitoring atthe firm level, because the premise of large-owner mon-itoring depends on the benefits that accrue to the largestmonitor. Such benefits also arise when large institutionalowners pay more attention to larger firms in their portfo-lios. Therefore, portfolio- and firm-level interests shouldcoincide when institutional owners have both the “meansand the incentive to take action [i.e., they own a sub-stantial share of the target firm’s stock] � � � in that a sig-nificant portion of their portfolio is invested in the firm”(Ryan and Schneider 2002, p. 561, see also Roe 1994,Sundaramurthy and Lyon 1998).

As mentioned in the development of Hypothesis 1,large investors tend to monitor effectively because thebenefits of their monitoring outweigh the costs. The-ory also suggests that large institutional investors shouldfocus more on larger firms in their portfolio becausedoing so results in more obvious performance improve-ments. At the same time, even the largest institutionalowner may look askance at the firms in the lower rungof their portfolio holdings, even if, at the firm level, theyhold a significant stake. Although owners might facepressures to monitor their smaller holdings, the net ben-efits of successful monitoring combined with the lim-ited incentives to do so (explicated in Hypothesis 2)suggest that monitoring will be compromised for low-ranking investments. Previous research clearly identi-fies this conundrum. For example, Ryan and Schneider(2002) theorize that institutional owners are more likelyto be active in target firms as the percentage of their port-folio invested in the focal firm’s stock increases. Alter-natively, even if institutional owners own large stakes infirms, if those firms form an insignificant part of theirportfolio, they are unlikely to be interested in monitor-ing. By concentrating their efforts on larger holdings,these institutional owners are most likely to benefit theirprimary investors. Therefore, we posit that large insti-tutional owners focus their monitoring efforts on the

larger holdings in their portfolios, at the expense ofsmaller holdings, based on the rational calculus of effortexpended to benefits accrued. Thus, when the firm isimportant in the large institutional owner’s portfolio, weexpect this portfolio characteristic to improve monitor-ing effectiveness of these large owners and at the sametime complement the benefits of monitoring from thesubstitution perspective.

Hypothesis 4. Significance of the target firm withinthe portfolio of large institutional investors associatespositively with monitoring effectiveness; specifically, thesignificance of the target firm relates negatively to thelevel of total compensation, positively to pay for perfor-mance sensitivity, positively to the salary proportion ofthe pay mix, and negatively to the contingent proportionof the pay mix, ceteris paribus.

Whereas the main contribution of our paper has beento augment the monitoring construct by considering theportfolio characteristics of large institutional owners, ourfinal hypothesis explores interaction effects between thelarge institutional owner’s firm-level ownership and itsportfolio characteristics in terms of overall monitoringeffectiveness. We might expect that as (1) the number ofblocks held by a large owner, (2) the portfolio turnoverof the large owner, or (3) the significance of the firmin the portfolio increases, firm-level monitoring implica-tions vary as well. For example, if the large institutionalowner holds no other blocks in any other firms within itsportfolio, any decrease in monitoring effectiveness at thefirm level should be minimal. Alternatively, if the largeinstitutional owner holds numerous other blocks, mon-itoring effectiveness will be compromised. Similarly, ifthe large institutional owner has low portfolio turnover,indicating that it seldom uses exit, the decrease in moni-toring effectiveness at the firm level should be minimal.However, a high level of portfolio turnover suggests adecrease in monitoring effectiveness. Finally, large insti-tutional owners likely will be more vigilant monitors ofthe larger, more significant firms in their portfolios, sofirm-level concentration and portfolio significance willinteract in a positive manner.

Hypothesis 5A. Large institutional ownership at thefirm level interacts negatively with portfolio concentra-tion, portfolio blockholdings, and portfolio turnover withrespect to monitoring effectiveness.

Hypothesis 5B. Large institutional ownership at thefirm level interacts positively with significance of the firmin the portfolio with respect to monitoring effectiveness.

MethodStudy SampleWe draw our sample from the S&P 500, Mid-Cap 400,and Small-Cap 600 companies. First, we collected annualCEO compensation and CEO ownership data from S&P’sExecuComp database for all firms that appeared con-


tinuously during the period 1993–2002,3 which pro-vided us with data on 623 firms. Second, we extractedquarterly and annual end-of-year institutional ownershipdata for the years 1992–2001 from the CDA/SpectrumThomson Financial’s 13F database. Third, we collectedfinancial information from Compustat for 1992–2001.Missing variables from the different data sets brought oursample to 4,520 firm-year observations for 533 firms. Toensure that our independent variables predate the depen-dent variables, we lagged them one year. The samplerepresents 241 industries at the four-digit standard indus-trial classification (SIC) level and 50 industries at thetwo-digit SIC level. Because we imposed no restrictionsduring the data extraction, the sample represents a goodcross section of industries; furthermore, the 10-year timeperiod for the study is important because it encompassesa variety of economic conditions. Unlike prior research,which considers the level of ownership in the focal firm(e.g., Agrawal and Mandelker 1990, Hartzell and Starks2003), we consider the entire portfolio of the institutionalinvestor and its behavior over time to estimate its incen-tives and ability to monitor at the firm level.

Dependent Variables

Total Compensation. This variable represents thevalue of all components of the CEO’s compensationfor a specific year, including salary, bonuses, long-termincentives, and the value of options granted in that year.

Proportion of Salary. To measure the salary portionof the CEO’s compensation that is not directly linked toorganizational outcomes, we use the ratio of the CEO’sbase salary to the total compensation he or she earnedfor the year.

Proportion of Contingent Compensation. Whereasoptions-based compensation is linked to long-termoutcomes, annual bonuses could be tied to variousshort-term performance measures, such as accountingearnings (Healy 1985). Furthermore, contingent com-pensation such as bonuses and long-term incentivesexposes managers to more risk and income instability(Daily et al. 1998, Tosi and Gomez-Mejia 1994). There-fore, we measure the proportion of compensation that iscontingent on firm-level outcomes both in the short termand in the long term. This variable represents the ratioof options and bonus to the total compensation the CEOearned during the year.

Pay for Performance Sensitivity. Although influen-tial institutional owners can restrain excessive execu-tive compensation, they may be more concerned aboutpay for performance (e.g. Jensen and Murphy 1990,Hambrick and Finkelstein 1995). Therefore, we measurethe pay for performance sensitivity as the sensitivityof the options grant per $1,000 change in shareholderwealth. Following Yermack (1995), we calculate this

variable using the partial derivative of the Black–Scholesvalue of options grants. Stock options represent thelargest component of executive compensation, and insti-tutional owners have expressed increased interest in thisform of compensation (Hartzell and Starks 2003). Wenote that this variable provides an ex ante measure ofperformance sensitivity in contrast to ex post measures,such as changes in CEO pay. Thus,

Pay for Performance Sensitivity

≈��shares represented by option award�

�shares outstanding at start of year�

�= e−dt�ln�P/E�+ T �r −d+�2/2�/��√T ��4

For example, a sensitivity value of 1.13 indicates that thevalue of stock option compensation increased by $1.13per $1,000 increase in shareholder value.

Independent Variables

Largest Institutional Owner Stake. Following Thom-sen and Pedersen (2000), we measure this variable as thepercentage of shares owned by the largest institutionalowner (top institutional owner’s shares divided by firm’stotal number of outstanding shares) in the focal firm,using data from the CDA/Spectrum Thomson Finan-cial 13F database. As the percentage of shares ownedincreases, owners can exercise more influence over thefirm and monitor more effectively. However, this mea-sure fails to account for owners’ incentives to monitorfrom the portfolio perspective or the importance of thefocal firm to the institutional investor.

Portfolio-Level Measures. Portfolio-based measuresof ownership address incentives to monitor, as embed-ded in the portfolio-holding characteristics of the largestinstitutional investor, and contribute additional infor-mation to more traditional measures of monitoringeffectiveness (Bushee 1998). To calculate the portfo-lio measures, we first identified the largest institutionalowner using Thomson Financial’s 13F database for eachyear in the study. In the second step, we extracted thecomplete portfolio holdings for these institutional own-ers to calculate the portfolio-based ownership variables.We provide numerical examples of portfolio measuresfor Wellington Management Company, Fidelity Manage-ment & Research Co., and Newsouth Capital Manage-ment, Inc. in Table 1.

Portfolio Concentration. Following Bushee (1998) wemeasure this variable as the average value of the invest-ments in the institutional owner’s portfolio. Thus, it mea-sures the tendency of the top institutional investor tohold large stakes in its portfolio. If the largest institu-tional owner holds significant investments in many otherfirms, its monitoring efforts at the focal firm may becompromised because it has become too busy to moni-tor effectively. The following calculation (Bushee 1998)


pertains to a skewed and kurtotic variable, so we took anatural log transformation:

Portfolio Concentrationt

=∑�Shares Heldkt × Stock Pricekt�/Number of

Stocks Ownedt�

where concentration is the sum of portfolio weights(shares held times stock price) in firm k at the end ofthe year divided by the total number of stocks owned bythe institution (Bushee 1998).

Portfolio Blockholding. This variable measures thetendency of the top institutional blockholder to investin large blocks. Traditional agency literature posits thatblockholders that own 5% or more of the company’sstock have sufficient incentives and ability to monitoreffectively (Bethel et al. 1998). Therefore, this variableprovides a more conservative measure of the propensityof the largest institutional investor to monitor firms inits portfolio other than the focal firm. We use a modifiedversion of Bushee’s (1998) measure of the proportion ofthe actual blockholdings in the portfolio. Because we areinterested in the competing effect of investments vyingfor monitoring attention, we multiply Bushee’s measureby 1 if the institutional owner is a blockholder in thefocal firm and 0 otherwise.

Portfolio Blockownershipt

= �∑�Shares Heldkt×Stock Pricekt�×Blockkt�×Targett

�∑�Shares Heldkt×Stock Pricekt��

�

where Blockkt = 1 if the percentage holding in firm k inthe portfolio is greater than or equal to 5% and 0 oth-erwise, and Targett = 1 if the percentage holding in thefocal firm is greater than or equal to 5% and 0 otherwise.

Portfolio Turnover. This variable measures the tradingbehavior of the largest institutional investor. Investorswith high portfolio turnover could be less prone to holdstocks over longer time periods and thus less likely toengage in costly monitoring efforts. Following Bushee(1998) this measure is the average absolute change in aninstitution’s ownership position over a quarter (denotedby q). Because increased turnover may be fueled by thefund’s growth and availability, we scale the shares tradedby the fund’s equity holdings at both the beginning andthe end of the quarter. The following calculation is basedon Thomson Financial 13F data; we take a natural logtransformation to correct for skewness

Portfolio Turnoverq

= (∑��Shares Heldkq × Stock Pricekq�

)

· (∑�Shares Heldkq × Stock Pricekq�

+∑�Shares Heldkq−1 × Stock Pricekq−1��

−1�

Portfolio Firm Significance. This variable reflects thepercentile ranking of the focal firm in the portfolio ofits largest owner, based on the dollar value of the invest-ments. The measure is inversed, so a higher numberindicates that the focal firm is of greater importance tothe institutional owner; for example, the 99th percentilemeans that the investment in the focal firm exceeds invalue 99% of the other investments in the portfolio.Thus, the variable indicates the proportion of the insti-tutional investor’s portfolio that consists of lower invest-ment stakes than those in the focal firm. The followingcalculation is based on Thomson Financial 13F data:

Portfolio Firm Significancet

= 1− �Position of the Focal Firm in the Portfoliot/Number of Firms in Portfoliot��

Control VariablesTo isolate the effects of the ownership variables onCEO compensation, we control for several factors thathave been shown to influence compensation in previousstudies.

Size. Previous studies link firm size and CEO com-pensation (Gomez-Mejia 1994, Lambert et al. 1991). Wecontrol for firm size by using the natural logarithm ofthe firm’s total assets.

Growth. Prior research argues that CEOs in high-growth firms may be higher marginal contributors andthus receive greater compensation (Finkelstein and Boyd1998, Gaver and Gaver 1993, Wright et al. 1996).Therefore, we control for growth opportunities with theTobin’s Q proxy of market value to the book valueof equity.

Firm Performance. We used both accounting- andmarket-based measures of performance. FollowingDavid and colleagues (1998), we include return on assets(ROA) and Jensen’s alpha as measures of market-basedperformance.

CEO Ownership. The level of ownership by CEOshas been the subject of many studies that relate it to risktaking, firm strategy, and CEO compensation (Bergeret al. 1997, Finklestein and Hambrick 1989). In agencytheory, CEO ownership serves as an indicator of incen-tive alignment and thus affects the potential for oppor-tunistic behavior, such as compensation maximization(Finkelstein and Boyd 1998). We measure CEO equityas the percentage of outstanding shares owned by theCEO at the end of each fiscal year.

Change in CEO. Changes in CEO appointments mayinfluence firm strategy and CEO compensation (Bergeret al. 1997, Finklestein and Hambrick 1989). We mea-sure this variable with a dummy code equal to 1 if thefirm experienced executive succession during the year.


CEO Duality. Executives serving simultaneously aschair of the board of directors may be better ableto influence their compensation (David et al. 1998,Westphal and Zajac 1994). Therefore, we included anindicator variable equal to 1 if the CEO is also the chairand 0 otherwise.

CEO Tenure. Suggested by prior literature as a proxyfor entrenchment and ability of the CEOs to exercisepower in setting their own compensation packages (Hilland Phan 1991), CEO tenure appears in this study as thenumber of years the individual had been CEO of a givencompany.

Liquidity. Firms experiencing cash shortages may pre-fer to grant options rather than use cash compensation(Shin 2005); therefore, we control for liquidity usingcurrent assets scaled by current liabilities.

Financial Leverage. We measure financial leverage asfirm indebtedness, which may limit the extent of risktaking and use of incentive plans (Yermack 1995). Wescaled the firm’s debt by the company’s assets.

CEO Salary Exceeding $1 Million. The OmnibusBudget Reconciliation Act of 1993 provides a limiton the deductibility of non-performance-based execu-tive compensation in excess of $1 million (Tolia 1997).Although companies may choose to pay higher taxesand pay higher salaries, we control for instances whenthe salary exceeds this limit. The indicator variable isequal to 1 if the CEO salary exceeds $1 million and 0otherwise.

Firm Risk. If compensation is commensurate with thecompany’s risk, executives in high-risk environmentsmay receive higher compensation (David et al. 1998,Hill and Phan 1991). Accordingly, we control for afirm’s systematic risk, as measured by the variance offirm stock price relative to that of the stock market.

Lagged Compensation. Prior compensation mayaffect current executive compensation, so we introducedlagged compensation variables in the models. To theextent that prior compensation determines subsequentcompensation, this variable also implicitly controls foromitted variables (Daily et al. 1998).

Institutional Ownership. Prior research indicates thataggregate institutional ownership affects compensation(e.g., Khan et al. 2005). We control for total ownershipof other institutional investors.

Industry. Previous studies link industry affiliation andCEO compensation (Gomez-Mejia 1994, Lambert et al.1991). We control for industry by using dummy codesthat represent the SIC two-digit classification of the firm.

AnalysesIt is more appropriate to study the relationship betweenownership and compensation in a longitudinal man-ner, because cross-sectional specification can lead tobiased or misleading parameter estimates (Murphy 1985,Finkelstein and Boyd 1998). Therefore, we used cross-sectional time-series analyses, which can isolate theeffects of specific actions and treatments over time andcross sections. Furthermore, with panel data analysis,which uses the stratification of individual entities andgrouping around time to provide better control on theeffects of missing or unobserved variables (Hsiao 1986),we can analyze both the idiosyncratic and intertempo-ral dynamics of the entities (Hill and Phan 1991). Oura priori choice for the analysis was a random-effectsmodel (REM) because we intend to make inferencesbeyond the sample used in this study, and REM allowsus to model the error specific to cross-sectional andtemporal units (Maddala 2002). This model employsgeneralized least squares (GLS) and thereby preservesmore of the information in the data and tends to bemore efficient. Furthermore, unlike fixed-effects mod-els (FEM), REMs let us include time-invariant variablessuch as industry controls.5 Finally, we applied modelsthat accommodated both cross-sectional and temporalrandom effects to control for changing economic condi-tions over the 10-year period that our data encompass.

ResultsThe descriptive statistics, including means, standarddeviations, and correlations, for all study variables areprovided in Table 2. The majority of the correlationsare significant at the 0.001 level and are of magnitudeand direction similar to those found in previous stud-ies (e.g., David et al. 1998). In the years represented bythe panel, on average total CEO compensation is $4.35million (standard deviation [s.d.] $5.36 million), salaryproportion constitutes 31.2%, contingent compensationproportion is 53.7%, and overall pay for performancesensitivity is 1.13. The average CEO ownership is 2%(s.d. of 5.6%), in line with previous studies (McGuireand Matta 2003), and average ownership by the largestinstitutional owner is 8.6%, (s.d. of 4.9%). The portfolioconcentration measure reveals that the average invest-ment in the largest institutional owner’s portfolio has amarket value of $85.4 million (s.d. of $101.1 million)and that 24.9% (s.d. of 21.7%) of the investor’s portfoliois held in blocks of 5% or greater. The portfolio turnoverstatistics suggest that institutional investors on averagechange ownership in 12.5% (s.d. of 6%) of their hold-ings from quarter to quarter. Furthermore, the averagefirm exceeds in value 88% (s.d. of 14%) of the portfo-lio holdings of its largest institutional owner. In addi-tion to these descriptive statistics, the correlation tableprovides some interesting insights. In line with previ-ous research, we find that firm size, growth, profitability,


Table2

Descriptive

StatisticsandCorrelations

No.

Variable

Mean

S.D.

12

34

56

78

910

1112

13

1Total

4�35

5�36

1compe

nsation($M)

2Sa

lary/total(%)

31�20

21�50

−0�828

∗∗∗

13

Continge

nt(options

+53

�70

24�70

0�552∗

∗∗−0

�592

∗∗∗

1bo

nus)/total(%)

4Firm

size

(assetsin$B

)6�00

11�50

0�583∗

∗∗−0

�345

∗∗∗

0�078∗

∗∗1

5Growth

(Tob

in’sQ)

1�74

1�45

0�196∗

∗∗−0

�242

∗∗∗

0�248∗

∗∗−0

�098

∗∗∗

16

Profitability(ROA)

0�05

0�11

0�11

∗∗∗

−0�098

∗∗∗

0�086∗

∗∗0�103∗

∗∗0�287∗

∗∗1

7CEO

ownership(%

)2�00

5�60

−0�169

∗∗∗

0�135∗

∗∗−0

�108

∗∗∗−0

�215

∗∗∗

0�08

∗∗∗

0�065∗

∗∗1

8Chang

eof

CEO

0�08

0�28

0�055∗

∗∗−0

�115

∗∗∗

0�069∗

∗∗0�039∗

∗−0

�01

−0�046

∗∗−0

�043

∗∗1

9CEO

duality

0�81

0�39

0�113∗

∗∗−0

�041

∗∗0�007

0�15

∗∗∗

−0�035

∗0�065∗

∗∗0�061∗

∗∗−0

�025

†1

10CEO

tenure

(years)

7�20

7�60

−0�031

∗0�033∗

−0�002

−0�096

∗∗∗

0�077∗

∗∗0�068∗

∗∗0�426∗

∗∗0�055∗

∗∗0�129∗

∗∗1

11Liqu

idity

(CA/CL)

1�95

1�90

−0�178

∗∗∗

0�053∗

∗∗0�067∗

∗∗−0

�474

∗∗∗

0�267∗

∗∗0�023

0�179∗

∗∗−0

�01

−0�104

∗∗∗

0�135∗

∗∗1

12Financialleverag

e0�20

0�15

−0�023

0�078∗

∗∗−0

�139

∗∗∗

0�226∗

∗∗−0

�301

∗∗∗−0

�234

∗∗∗−0

�129

∗∗∗

0�003

−0�002

−0�079

∗∗∗−0

�28∗

∗∗1

(deb

t/assets)

13Sa

lary

>$1

million

0�15

0�36

0�425∗

∗∗−0

�198

∗∗∗

0�075∗

∗∗0�41

∗∗∗

0�018

0�043∗

∗−0

�017

−0�065

∗∗∗

0�075∗

∗∗0�056∗

∗∗−0

�124

∗∗∗

0�039∗

∗1

14Jensen’salph

a0�01

0�01

0�09

∗∗∗

−0�17∗

∗∗0�226∗

∗∗−0

�18∗

∗∗0�33

∗∗∗

−0�022

0�03

∗−0

�014

−0�077

∗∗∗

0�083∗

∗∗0�314∗

∗∗−0

�206

∗∗∗−0

�068

∗∗∗

15Firm

risk(Beta)

0�99

0�64

0�089∗

∗∗−0

�148

∗∗∗

0�195∗

∗∗−0

�152

∗∗∗

0�238∗

∗∗−0

�118

∗∗∗

0�033∗

−0�01

−0�11∗

∗∗0�058∗

∗∗0�327∗

∗∗−0

�175

∗∗∗−0

�031

∗

16Lagg

edsalary

32�00

22�30

−0�342

∗∗∗

0�44

∗∗∗

0�11

∗∗∗

−0�206

∗∗∗−0

�171

∗∗∗−0

�074

∗∗∗

0�09

∗∗∗

−0�012

−0�019

0�054∗

∗∗0�014

0�046∗

∗−0

�124

∗∗∗

prop

ortion(%

)17

Lagg

edop

tions

34�10

27�60

0�35

∗∗∗

−0�315

∗∗∗−0

�36∗

∗∗0�145∗

∗∗0�176∗

∗∗−0

�021

−0�147

∗∗∗−0

�062

∗∗∗−0

�032

∗−0

�089

∗∗∗

0�044∗

∗−0

�06∗

∗∗0�109∗

∗∗

prop

ortion(%

)18

Lagg

edtotal

4�66

12�84

0�624∗

∗∗−0

�439

∗∗∗

0�391∗

∗∗0�493∗

∗∗0�183∗

∗∗0�098∗

∗∗−0

�15∗

∗∗−0

�003

0�097∗

∗∗−0

�04∗

∗−0

�14∗

∗∗−0

�031

∗0�365∗

∗∗

compe

nsation($M)

19Lagg

edcontinge

nt53

�20

25�10

0�378∗

∗∗−0

�353

∗∗∗

0�473∗

∗∗0�092∗

∗∗0�238∗

∗∗0�112∗

∗∗−0

�128

∗∗∗−0

�071

∗∗∗

0�002

−0�039

∗0�068∗

∗∗−0

�124

∗∗∗

0�101∗

∗∗

compprop

ortion(%

)20

Institutional

50�20

16�10

0�361∗

∗∗−0

�27∗

∗∗0�253∗

∗∗0�199∗

∗∗0�069∗

∗∗0�155∗

∗∗−0

�227

∗∗∗

0�007

0�081∗

∗∗−0

�052

∗∗∗−0

�01

−0�03∗

0�152∗

∗∗

ownership(%

)21

Largestinstitutional

8�60

4�90

−0�047

∗∗0�039∗

0�022

−0�164

∗∗∗−0

�149

∗∗∗−0

�035

∗−0

�038

∗0�011

−0�019

−0�018

0�096∗

∗∗0�045∗

∗−0

�042

∗∗

owner(%

)22

Portfolio

85�40

101�10

0�382∗

∗∗−0

�29∗

∗∗0�22

∗∗∗

0�35

∗∗∗

0�142∗

∗∗0�087∗

∗∗−0

�134

∗∗∗

0�031∗

0�024

−0�031

∗−0

�108

∗∗∗

0�003

0�198∗

∗∗

concentra

tion($M)

23Po

rtfolio

24�90

21�72

0�057∗

∗∗−0

�048

∗∗0�074∗

∗∗−0

�049

∗∗∗−0

�088

∗∗∗−0

�021

−0�049

∗∗0�029∗

−0�043

∗∗−0

�001

0�021

0�039∗

∗0�024

blockholding

(%)

24Po

rtfolio

turnover

(%)

12�49

5�74

0�03

∗−0

�047

∗∗0�068∗

∗∗−0

�073

∗∗∗

0�07

∗∗∗

−0�009

−0�01

−0�003

−0�014

−0�005

0�061∗

∗∗−0

�029

∗0�004

25Firm

sign

ificance

0�88

0�14

0�38

∗∗∗

−0�273

∗∗∗

0�191∗

∗∗0�471∗

∗∗0�127∗

∗∗0�239∗

∗∗−0

�084

∗∗∗−0

�017

0�119∗

∗∗−0

�003

−0�155

∗∗∗−0

�003

0�184∗

∗∗

inpo

rtfolio

26Pa

yforpe

rformance

1�13

2�15

0�119∗

∗∗−0

�212

∗∗∗

0�318∗

∗∗−0

�15∗

∗∗−0

�02

−0�224

∗∗∗

0�048∗

∗0�081∗

∗∗−0

�095

∗∗∗

0�012

0�136∗

∗∗−0

�031

∗−0

�069

∗∗∗

sensitivity

27Chang

einshareholde

r0�53

8�95

0�056∗

∗∗−0

�062

∗∗∗

0�03

†−0

�015

0�288∗

∗∗0�08

∗∗∗

0�026†

0�04

∗∗0�037∗

0�035∗

0�006

−0�046

∗∗

wealth

($B)

28Chang

einshareholde

r0�81

8�33

0�092∗

∗∗−0

�083

∗∗∗

0�042∗

∗0�146∗

∗∗0�255∗

∗∗0�103∗

∗∗0�012

0�017

−0�005

0�00

−0�002

−0�067

∗∗∗

0�051∗

∗

wealth

t−1($B)

29Market

8�10

23�77

0�309∗

∗∗−0

�234

∗∗∗

0�128∗

∗∗0�552∗

∗∗0�38

∗∗∗

0�138∗

∗∗−0

�02

0�04

∗∗0�00

−0�024

−0�069

∗∗∗−0

�1∗∗

∗0�281∗

∗∗

capitalization($B)


Table2

(cont’d.)

No.

Variable

1415

1617

1819

2021

2223

2425

2627

28

14Jensen’salph

a1

15Firm

risk(Beta)

0�804∗

∗∗1

16Lagg

edsalary

−0�099

∗∗∗−0

�12∗

∗∗1

prop

ortion(%

)17

Lagg

edop

tions

0�186∗

∗∗0�218∗

∗∗−0

�434

∗∗∗

1prop

ortion(%

)18

Lagg

edtotal

0�062∗

∗∗0�102∗

∗∗−0

�575

∗∗∗

0�574∗

∗∗1

compe

nsation($M)

19Lagg

edcontinge

ntcomp

0�212∗

∗∗0�207∗

∗∗0�16

∗∗∗

−0�708

∗∗∗

0�839∗

∗∗1

prop

ortion(%

)20

Institutional

0�139∗

∗∗0�15

∗∗∗

−0�169

∗∗∗

0�213∗

∗∗0�289∗

∗∗0�251∗

∗∗1

ownership(%

)21

Largestinstitutional

0�065∗

∗∗0�095∗

∗∗0�017

0�025†

−0�042

∗∗0�012

0�236∗

∗∗1

owner(%

)22

Portfolio

0�044∗

∗0�029†

−0�193

∗∗∗

0�227∗

∗∗0�329∗

∗∗0�232∗

∗∗0�327∗

∗∗0�081∗

∗∗1

concentra

tion($M)

23Po

rtfolio

0�081∗

∗∗0�075∗

∗∗−0

�026

†0�049∗

∗∗0�041∗

∗0�054∗

∗∗0�245∗

∗∗0�533∗

∗∗0�308∗

∗∗1

blockholding

(%)

24Po

rtfolio

turnover

(%)

0�144∗

∗∗0�044∗

∗−0

�052

∗∗∗

0�076∗

∗∗0�047∗

∗0�064∗

∗∗0�12

∗∗∗

0�07

∗∗∗

0�026†

0�122∗

∗∗1

25Firm

sign

ificance

0�012

−0�007

−0�171

∗∗∗

0�121∗

∗∗0�31

∗∗∗

0�168∗

∗∗0�238∗

∗∗0�059∗

∗∗0�05

∗∗∗

−0�017

0�021

1inpo

rtfolio

26Pa

yforpe

rformance

0�104∗

∗∗0�139∗

∗∗0�128∗

∗∗0�028†

0�058∗

∗∗−0

�021

0�032∗

−0�021

0�078∗

∗∗−0

�091

∗∗∗

0�049∗

∗∗0�05

∗∗∗

−0�181

∗∗∗1

sensitivity

27Chang

einshareholde

r0�009

−0�021

−0�042

∗∗−0

�011

0�009

0�038∗

0�015

−0�081

∗∗∗

0�024

−0�04∗

∗0�039∗

0�049∗

∗−0

�020

1wealth

($B)

28Chang

einshareholde

r0�085∗

∗∗0�014

−0�052

∗∗0�021

0�096∗

∗∗0�039∗

−0�014

−0�105

∗∗∗

0�046∗

∗−0

�073

∗∗∗

0�012

0�075∗

∗∗−0

�037

∗0�103∗

∗∗1

wealth

t−1($B)

29Market

0�026†

0�008

−0�167

∗∗∗

0�111∗

∗∗0�171∗

∗∗0�131∗

∗∗−0

�006

−0�232

∗∗∗

0�158∗

∗∗−0

�151

∗∗∗−0

�032

∗0�208∗

∗∗−0

�113

∗∗∗0�35

∗∗∗

0�443∗

∗∗

capitalization($B)

Notes.N

=4�520�533

firms×10

years�.†

p<0�10,∗p<0�05,∗

∗ p<0�01,∗

∗∗p<0�001.


and CEO duality correlate positively with total compen-sation (r = 0�583, 0.196, 0.11, and 0.113, respectively),and CEO ownership, typically considered a solution toagency problems, correlates negatively �r = −0�169�.Finally, our firm- and portfolio-level measures correlatein an opposite manner. For example, though ownershipstake by the largest institutional owner correlates nega-tively with total compensation �r = −0�047�, portfolioconcentration, portfolio blockholding, portfolio turnover,and firm significance in the portfolio correlate positivelywith total compensation (r = 0�382, 0.057, 0.03, 0.38),which offers some preliminary univariate support for ourhypotheses.

From Table 3, we gain some interesting insights intothe implications of portfolio characteristics for gover-nance and executive compensation. Our analyses pro-ceeded in three steps; we note any changes in the BuseR-square. Consistent with previous studies (e.g., Gomez-Mejia and Welbourne 1989), firm size relates positivelyto total compensation (� = 0�52, p < 0�001) and tothe proportion of contingent compensation (�= 0�02,p < 0�01), but it relates negatively to the proportionof salary (� = −0�26, p < 0�001). Growing firmstend to provide higher levels of total compensation(�= 0�11, p < 0�001) and rely less on salary in the paymix (�=−0�10, p < 0�001) and more on outcomes-based compensation (� = 0�03, p < 0�001), whichagain is consistent with previous research. Account-ing profitability relates positively to total compensa-tion (�= 0�02, p < 0�05), as does a change in CEO(�= 0�14, p < 0�001). Change in CEO also relates neg-atively to the proportion of salary in the pay mix(�=−0�33, p < 0�001) and positively to the propor-tion of contingent compensation (�= 0�07, p < 0�001).CEO ownership relates negatively to the proportionof contingent compensation (� = −0�01, p < 0�05).Increased ownership causes the CEO to bear morerisk, which can be exacerbated by increased incen-tive pay and therefore lead to greater risk aversion(Wiseman and Gomez-Mejia 1998). Finally, financialleverage is associated with reduced total compensa-tion (�=−0�09, p < 0�001), an increased proportion ofsalary (�= 0�08, p < 0�001), and the proportion of con-tingent compensation (�=−0�01, p < 0�01). Consistentwith prior research (e.g., Hartzell and Starks 2003) firmsize relates negatively to pay for performance sensi-tivity (�=−0�33, p < 0�001). Perhaps because pow-erful CEOs can extract higher compensation regardlessof firm performance, CEO duality negatively related topay for performance sensitivity (�=−0�29, p < 0�001),whereas the link is positive for new CEOs (� = 0�72,p < 0�001).

In Hypothesis 1, we suggested that greater ownershipby the top institutional owner would increase its abilityto influence the firm and thus facilitate monitoring andis associated with less total compensation, greater pay

for performance sensitivity, a higher proportion of salary,and a lower proportion of contingent compensation (seeCore et al. 1999, David et al. 1998, Hartzell and Starks2003, Mehran 1995). Consistent with previous findings,large institutional owners, by the very nature of theirownership stakes and monitoring abilities in comparisonwith small owners, have the ability to influence CEOcompensation. Our results also indicate that the largestowners are able to monitor; the coefficient for largestinstitutional owner is negative for total compensation(� = −0�03, p < 0�05) but is not significantly relatedto pay for performance sensitivity. Additional analysesindicate that large institutional ownership is associatedwith pay for performance sensitivity in the absence ofportfolio considerations (see Appendix 1). In addition,large institutional owners are positively associated withsalary as a proportion of total compensation (�= 0�03,p < 0�05) even after we control for most of the standardvariables in such research. In addition, by using paneldata analysis techniques, we intentionally controlled forthe effects of missing and unobserved variables and iso-lated effects both cross-sectionally and over time. Anylimitations in the choice of variables or endogeneity arenot specific to our sample, and we replicate previousresults for this particular hypothesis.

In strong support of Hypothesis 2, we find that theowner’s portfolio concentration is negatively associatedwith monitoring effectiveness. Specifically, portfolioconcentration in terms of increases in average hold-ings relates positively to the level of total compen-sation (� = 0�03 p < 0�05), negatively to pay forperformance sensitivity (� = −0�23, p < 0�05), andnegatively to salary as a percentage of total compensa-tion (�=−0�02, p < 0�05). Similarly, greater portfolioblockholding (i.e., the number of blocks in the portfolio)by the largest institutional owner compromises its firm-level monitoring. Portfolio blockholding is associatedwith greater total compensation (�= 0�03, p < 0�05) anda higher proportion of contingent compensation (� =0�01, p < 0�001). The portfolio-level effects for thelargest institutional owner are in the opposite directionof the firm-level ownership effect.

In weak support of Hypothesis 3, portfolio turnoveris associated with less reliance on salary as a propor-tion of pay (�=−0�02, p < 0�05), which suggests thatthe turnover in the largest owner’s portfolio relates tocomplacency in monitoring. Finally, contrary to Hypoth-esis 4, we find that firm significance in the portfoliorelates positively to the level of total compensation (�=0�06, p < 0�001), negatively to pay for performancesensitivity (�=−0�07, p < 0�05), negatively to the pro-portion of salary (�=−0�06, p < 0�001), and positivelyto the proportion of contingent compensation (�= 0�02,p < 0�001). Overall, our results indicate that portfolio-level influences are significantly associated with facetsof compensation, and researchers and practitioners need


Table 3 Monitoring of CEO Compensation by Largest Institutional Owner (Time-Series Cross-Sectional Regression with Firm andYear Random Effects)

Total executive Pay for performance Proportion ofcompensation sensitivity Proportion of salary contingent pay

Variables Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 13 Model 14 Model 15

Step 1Constanta 7�86∗∗∗ 7�83∗∗∗ 7�85∗∗∗ 1�07∗ 1�00∗ 1�02∗ −1�36∗∗∗ −1�35∗∗∗ −1�36∗∗∗ 0�45∗∗∗ 0�43∗∗∗ 0�44∗∗∗Firm size 0�52∗∗∗ 0�49∗∗∗ 0�49∗∗∗ −0�33∗∗∗ −0�29∗∗∗ −0�28∗∗∗ −0�26∗∗∗ −0�23∗∗∗ −0�23∗∗∗ 0�02∗∗ 0�01 0�01∗Growth 0�11∗∗∗ 0�10∗∗∗ 0�10∗∗∗ −0�05 −0�04 −0�04 −0�10∗∗∗ −0�09∗∗∗ −0�09∗∗∗ 0�03∗∗∗ 0�03∗∗∗ 0�03∗∗∗Profitability 0�02∗ 0�02∗ 0�02∗ −0�25∗∗∗ −0�24∗∗∗ −0�23∗∗∗ −0�02 −0�02 −0�02 0�00 0�00 0�00CEO ownership −0�02 −0�02 −0�02 0�09∗ 0�08 0�08 0�02 0�02 0�02 −0�01∗ −0�01∗ −0�01∗Change of CEO 0�14∗∗∗ 0�15∗∗∗ 0�15∗∗∗ 0�72∗∗∗ 0�71∗∗∗ 0�71∗∗∗ −0�33∗∗∗ −0�34∗∗∗ −0�34∗∗∗ 0�07∗∗∗ 0�07∗∗∗ 0�07∗∗∗CEO duality 0�02 0�03 0�03 −0�29∗∗ −0�28∗∗ −0�26∗∗ −0�01 −0�01 −0�01 0�00 0�01 0�01CEO tenure 0�02∗ 0�02 0�02 −0�03 −0�03 −0�03 0�01 0�01 0�01 −0�01 −0�01 −0�01Liquidity −0�03 −0�03 −0�03 0�02 0�02 0�02 −0�02 −0�02 −0�02 0�00 0�00 0�00Financial leverage −0�09∗∗∗ −0�09∗∗∗ −0�09∗∗∗ −0�09∗ −0�09∗ −0�09∗ 0�08∗∗∗ 0�07∗∗∗ 0�07∗∗∗ −0�01∗∗ −0�01∗∗ −0�01∗∗Salary> $1 million 0�31∗∗∗ 0�31∗∗∗ 0�30∗∗∗ 0�08 0�08 0�08 −0�08∗ −0�08∗ −0�08∗ −0�01 −0�01 −0�01Jensen’s alpha 0�17∗∗∗ 0�16∗∗∗ 0�15∗∗∗ 0�04 0�03 0�02 −0�12∗∗ −0�10∗∗ −0�10∗∗ 0�00 0�00 0�00Firm risk −0�15∗∗∗ −0�14∗∗∗ −0�14∗∗∗ 0�01 0�01 0�02 0�09∗ 0�08∗ 0�08∗ 0�00 0�01 0�01Lagged total 0�01 0�01 0�01compensation

Lagged salary 0�08∗∗∗ 0�07∗∗∗ 0�07∗∗∗proportion

Lagged contingent 0�02∗∗∗ 0�02∗∗∗ 0�02∗∗∗compensationproportion

Institutional ownership 0�08∗∗∗ 0�06∗∗∗ 0�06∗∗∗ −0�05 −0�05 −0�04 −0�05∗∗∗ −0�04∗∗ −0�04∗∗ 0�02∗∗∗ 0�02∗∗∗ 0�02∗∗∗

Step 2Largest institutional −0�03∗ −0�03∗ 0�03 0�01 0�03∗ 0�03∗ −0�01 −0�01owner

Portfolio concentration 0�02∗ 0�03∗ −0�23∗ −0�23∗ −0�02∗ −0�02∗ 0�01 0�01Portfolio blockholding 0�03∗ 0�03∗ 0�05 0�06 −0�01 −0�02 0�01∗∗∗ 0�01∗∗∗Portfolio turnover 0�01 0�01 0�01 0�01 −0�02∗ −0�02∗ 0�00 0�00Firm significance 0�06∗∗∗ 0�06∗∗∗ −0�08∗ −0�07∗ −0�06∗∗∗ −0�06∗∗∗ 0�02∗∗∗ 0�02∗∗∗in portfolio

Step 3Largest institutional 0�01 −0�05 0�00 0�00owner×portfolioconcentration

Largest institutional −0�01 −0�05∗ 0�02 0�00owner×portfolioblockholding

Largest institutional −0�01 0�06∗ 0�03∗∗ −0�01∗owner×portfolioturnover

Largest institutional 0�03∗∗∗ 0�11∗∗∗ −0�02∗ 0�01∗owner× firmsignificancein portfolio

Buse R-square (%) 23�1 24�1 24�4 5�1 5�3 5�8 14�1 14�8 15�1 7�9 9�0 9�2Change in R-square 1∗∗∗ 0�3∗∗∗ 0�2 0�5∗∗∗ 0�7∗∗∗ 0�3∗∗ 1�1∗∗∗ 0�2

Notes. N = 4�520 (533 firms×10 years). ∗p < 0�05, ∗∗p < 0�01, ∗∗∗p < 0�001.a49 dummy codes controlling for SIC two-digit industry level are not reported here for brevity.

to consider these effects to obtain a clearer picture ofmonitoring by institutional investors.

Finally, to test our exploratory-interaction hypoth-esis, we used the standard Aiken and West (1991)methodology to create the interaction terms and foundthat the variance inflation factors for the variables in themodel were all within an acceptable range �<10�. Wediscuss only interactions when the change in R-squareover the previous model is significant. We find that own-ership interacts with firm significance in the portfolio toincrease total compensation (� = 0�03, p < 0�001) andpay for performance sensitivity (� = 0�11, p < 0�001)

while reducing the proportion of salary (� = −0�02,p < 0�05). In addition, firm-level ownership interactswith portfolio turnover to influence pay for performancesensitivity (� = 0�06, p < 0�05) and the proportion ofsalary (� = 0�03, p < 0�01). Finally, firm-level owner-ship interacts with portfolio blockholdings to influencepay for performance sensitivity (�=−0�05, p < 0�05).Given the limited theory development and empiricalresearch regarding portfolio significance, we discusssome possible explanations subsequently. Although mosteffects were in the hypothesized direction, the results forfirm significance indicate the need for future research.


To understand this further, we plotted the interaction ofthe largest institutional owner and firm significance withrespect to total compensation in Figure 1 (� = 0�03,p < 0�001), which reveals that, on one hand, greater sig-nificance is associated with higher levels of total com-pensation, irrespective of ownership. On the other hand,monitoring by the larger institutional owner suppressestotal compensation in low-significance firms in the port-folio. We discuss this point more subsequently.

Supplementary AnalysesAlthough our results indicate that the presence of power-ful institutional owners constrains the ability of agents toextract excessive compensation, we do not find a signif-icant effect on pay for performance sensitivity. BecauseHartzell and Starks (2003) found that ownership bythe top five institutional investors relates positively, weconducted additional analyses using their model speci-fication as a starting point, but we replaced their own-ership construct “ownership by the top five institutionalinvestors” with our “ownership of the largest institu-tional owner” (see Appendix 1). Similar to their results,we find that market capitalization relates negatively tooptions performance sensitivity (�=−0�17, p < 0�001),whereas ownership by the largest institutional ownerrelates positively to it (� = 0�04, p < 0�05). However,when we extend the model by including our portfolio-level explanatory variables, the coefficient for the largestinstitutional owner no longer is significant. In addition,consistent with the prior model specification, portfo-lio concentration and firm significance in the portfoliorelate negatively to pay for performance sensitivity (�=−0�09, p < 0�001; �=−0�16, p < 0�001, respectively).

Although we direct attention to the underresearchedarea of how the portfolio characteristics of institutionalinvestors affect their incentives to monitor, institutionalinvestors can develop the organizational capabilities tohandle the complexity of monitoring. For example, com-petent institutional investors may make more invest-ments and still manage them effectively or recruit morefund managers to deal with the resulting complexity.6

Thus, an institutional owner’s organizational capability

Figure 1 Interaction Effect of Largest Institutional OwnerStake and Firm Significance in Portfolio on TotalExecutive Compensation

low high

Largest institutional owner stake

To

tal c

om

pen

sati

on

Low

Hig

h

High firm significanceLow firm significance

in managing diversified investments merits considera-tion in the context of monitoring at the firm level. Wetherefore obtained data on mutual fund performance andthe number of funds per institutional investor from theCRSP Mutual Fund database, a comprehensive data setthat provides information about open-end mutual funds.The CRSP database identifies fund managers by nameor simply as “team managed”; therefore, the number offunds per institutional investor is a proxy for the numberof managers employed by the fund and thus the expertiseavailability for the investor. We measured prior perfor-mance of each fund as the increase (decrease) in netassets value per share (NAV). Because different fundsmanage different amounts of assets, we scaled the mea-sure by the total net assets (TNA) managed by each fundand then aggregated them to reach a performance mea-sure for the institutional investor. Unfortunately, suchdata are not available for all institutional owners, whichsignificantly reduces our data set �n= 2�428�. Becauseof the large number of observations missing from ourpanel, we conduct an ordinary least squares (OLS) anal-ysis of the pooled data (see Appendix 2). In the presenceof autocorrelation, OLS is consistent but inefficient, sowe use the Newey–West correction (Greene 2003).

Despite the added controls and sample size differ-ences, the tenets of our inferences remain unchanged.Similar to the results we discussed previously, the coef-ficient for the largest institutional owner is negativefor total compensation (� = −0�10, p < 0�001), pos-itive for salary as a proportion of total compensation(� = 0�08, p < 0�001), and negative for the propor-tion of contingent compensation (�=−0�02, p < 0�01).We find that the owner’s portfolio concentration relatespositively to the level of total compensation (� = 0�13p < 0�001), negatively to pay for performance sensitiv-ity (�=−0�34, p < 0�001) and to salary as a percentageof total compensation (�=−0�06, p < 0�01), and pos-itively to the proportion of contingent pay (� = 0�01,p < 0�05). Again, in contrast with Hypothesis 4, we findthat firm significance in the owner’s portfolio relatespositively to the level of total compensation (�= 0�22,p < 0�001), negatively to pay for performance sensitiv-ity (�=−0�24, p < 0�001), negatively to the proportionof salary (�=−0�11, p < 0�001), and positively to theproportion of contingent compensation (� = 0�02, p <0�001). Although the effect of portfolio blockholding onpay for performance sensitivity is insignificant in the fullsample, when we controlled for fund performance andthe number of funds per institutional investor, we foundthat portfolio blockholding is related positively to payfor performance sensitivity (� = 0�17, p < 0�01). Con-sistent with our reasoning that competent institutionalinvestors tend to make more investments and managethem more effectively, the number of fund managersrelates positively to pay for performance sensitivity (�=0�07, p < 0�05). Prior fund performance is not signif-icantly related to executive compensation, irrespective


of the measure of compensation used. Because fundperformance is a conjoint indicator of the institutionalinvestor’s ability to both select investments and exercisemonitoring, the use of a finer-grained measure of orga-nizational capability seems well warranted.

Finally, we must consider the issue of endogeneity,especially because increased institutional ownership isaccompanied by the increased use of stock optioncompensation during our study period. These trendscould cause spurious relationships between the twovariables (Hartzell and Starks 2003). To address thisissue, we conducted two-stage least squares (2SLS) (seeAppendix 3). In the first stage, we estimated institutionalownership using prior changes in shareholder wealth,firm size, industry, year, and turnover (Bennett et al.2003) as instrumental variables. In the second stage,we estimated pay for performance sensitivity using thefitted values from the first-stage regression and intro-ducing portfolio-level variables in the model. Similarto previously reported results, portfolio concentrationand firm significance are adversely related to pay forperformance sensitivity (� = −0�002, p < 0�001; � =−2�61 p < 0�001). The coefficients for these variablesdo not change significantly if we control for institutionalinvestor performance or the number of funds managed(�=−0�002, p < 0�001; �=−2�74 p < 0�001), thoughthe coefficient for portfolio blockholding is significantin the extended model (�= 0�70, p < 0�05).

Discussion and ConclusionWe have argued that, to comprehend the implications oflarge institutional owners in terms of monitoring effec-tiveness, researchers must focus not only on firm-levelownership in a focal firm but also on the portfolio-levelcharacteristics of its largest institutional investor. Con-sistent with previous literature, we use various controls(e.g., industry, size, growth, profitability, CEO owner-ship, change in CEO, CEO duality, leverage, liquid-ity, risk, etc.) and find that they behave as anticipated.We therefore test a traditional hypothesis relating largeinstitutional ownership to aspects of CEO compensation(level, sensitivity, and mix) and then supplement it withhypotheses pertaining to the implications of four port-folio characteristics of large owners, namely, averageholdings, blockholdings, turnover, and significance ofthe focal firm. Our findings clearly indicate that portfo-lio characteristics matter. Larger average holdings, moreblockholdings, portfolio turnover, and even the portfo-lio significance of firms potentially offset the monitoringbenefits of large institutional owners.

As we noted previously, large owners appear to offeran important solution to the agency problems in the faceof dispersed ownership. Large institutional owners at thefirm level are associated with reduced levels of totalcompensation, increased pay for performance sensitivity,and influence on the pay mix. Do these large institutionalowners’ portfolios hinder overall monitoring effective-

ness? In other words, if large institutional owners (e.g.,Fidelity with approximately 3,000 firms in its portfolio)have numerous stakes in many firms, are firm-level mon-itoring benefits offset by portfolio-level characteristics?

To answer this question, we assess the impact ofthe portfolio and conclude that it offsets overall mon-itoring effectiveness at the firm level. All else beingequal, larger average holdings, more blockholdings,more turnover, and even firm significance in the port-folio reduced monitoring effectiveness. In our supple-mentary analyses, we include two additional measuresof the institutional investors’ capacity to monitor (num-ber of funds as a proxy for the number of fund managersand overall portfolio profitability as a measure of capa-bility) and still find support for our portfolio variables.Second, we replicate findings from Hartzell and Starks(2003) for large owners’ positive effects on pay for per-formance sensitivity and demonstrate that our portfoliovariables counteract the firm-level effect. This indicatesthe utility of the portfolio variables even when usingdifferent techniques and with different control variables.Third, in response to Demsetz’s (1983) admonition aboutfirm ownership’s endogeneity implications and the welldocumented need among institutional owners for liquid-ity (Bennett et al. 2003), we use firm share turnover asan instrumental variable for total institutional ownershipand show that portfolio effects still matter. Overall, ourresults suggest that the beneficial effect of monitoringby large institutional owners gets constrained by theirportfolio characteristics.

From a theoretical perspective, institutional owners’portfolio characteristics and investor orientation limitthe development of sufficient monitoring capacity asdemanded by their highly fragmented portfolios. Wehave suggested various reasons (e.g., multiple fundobjectives, costs, incentives, etc.) for this state of affairs.Institutional investors may focus on risk and cost reduc-tion, as well as on reducing the “annoyance of monitor-ing management” (Demsetz 1983). The average value ofthe equity stake held by the largest institutional investorin our sample was $86 million. With many such invest-ments, owners may become overburdened and fracturedand therefore resort to focusing their limited monitoringcapacity on either a few very large holdings or highlyvisible problematic holdings. At the bottom line, the sup-pression effect generated at the firm level may be off-set by demands at the portfolio level, as indicated byour empirical findings. Higher average holdings demandmore monitoring and are associated with higher totalcompensation, lower salary proportion, higher contin-gent proportion in the pay mix, and lower pay for perfor-mance sensitivity. If larger institutional investors actuallyhave greater monitoring capacity, and if having largeraverage portfolio holdings is a proxy for that capacity,we should have found opposite portfolio effects.


Furthermore, our findings about other portfolio char-acteristics are similar and, more important, consistentwith our preceding theorizing. When the number ofblockholdings in the institutional owner’s portfolio isgreater, the owner’s monitoring effectiveness is compro-mised. Portfolio blockholding relates positively to thelevel of total compensation and pay mix that favors con-tingent pay. In other words, the number of blockhold-ings appears to compromise the benefit of firm-levelblockholding. We thus repeat our preceding arguments:With more blockholdings, institutional investors gainmore experience and economies of scale and scope withrespect to monitoring. However, everything else beingconstant, maintaining a large number of blocks offsetsthe benefits of firm-level effects by straining the institu-tional investors’ monitoring capacity.

Our findings pertaining to portfolio turnover are some-what weaker, in that it relates negatively to only thepercentage of salary in the pay mix. We hypothesizedthat, as institutional owners become more short-term-oriented and increase their use of exit as an alternative,they compromise their monitoring. Two possible expla-nations may explain our contrary finding. First, portfolioturnover decreased over the time period of our study asmore institutional owners began to offer “indexed funds”to investors (Pozen 2002). As fund managers began tochoose broad indices, such as the S&P 500 or the Russell3,000, they de facto became passive investors whoseactive trading volume was limited to adjustments in theindex, which decreased turnover. Although we do nothave access to information about the volume of index-ing activity, anecdotal evidence suggests that it is on therise (Pozen 2002) and may be associated with reducedvariance in turnover. Second, researchers should con-sider owner turnover at the firm level, because as institu-tional investors increase their stakes in the firm they mayincrease their monitoring. In contrast, if they decreasetheir stake, it may signal reduced monitoring and anincreased desire to exit.

Finally, the portfolio significance of the target firmrelates positively to the level of total compensationbut negatively to pay for performance sensitivity andproportion of salary of total compensation. This inter-esting finding contrasts with our hypothesized effect,which favored monitoring-based explanations for exec-utive compensation. Conventional theorizing might sug-gest that large owners are more likely to monitor thelarger firms in their portfolios because of the obvi-ous benefits associated with such monitoring (Ryan andSchneider 2002), but our finding necessitates an exami-nation of alternative explanations. One such explanationrelies on the power imbalance between the interestedparties. As David and colleagues (1998) argue, CEOcompensation could be an outcome of a political process,and large owners may be able to exercise more power infurthering their interests. Although prior research does

not differentiate explicitly among monitoring, power,and deterrence by large owners, our portfolio signifi-cance findings may shed new light on this area of CEOcompensation. That is, we find that large owners areless able to influence larger firms in their portfolios.Instead, they are better able to monitor smaller firms inwhich they have large ownership positions. However, asthe firms get larger, the owners’ monitoring effective-ness appears to be negated (perhaps due to their lesserpower compared with huge firms like General Electricor ExxonMobil). Perhaps large investors tread carefullyin such cases, because the downside risks of fallout fromineffective monitoring may be greater for larger hold-ings. Alternatively, institutions may get locked into thecompanies in which they have invested the most (Shortand Keasey 1997), which makes changes in compensa-tion and/or governance more difficult.

From our investigation of interaction effects (Fig-ure 1), we find that large owners make no difference forsignificant firms in their portfolio with respect to totalcompensation. Consistent with our findings pertainingto Hypothesis 4, for firms that are more important inthe portfolio, the suppression effect of firm concentra-tion vanished. It exists for firms with low significancewhere owners have higher stakes and the firm is nothighly significant in the portfolio. Although our portfolioconcentration and turnover effects are in the appropri-ate direction, our findings regarding firm significance inthe portfolio raise interesting questions that need furtherinvestigation and have important practical significance.In general, portfolio effects are in the opposite directionof the firm-level influences, so both firm and portfolioeffects matter. Paying singular attention to either onewill result in an incomplete understanding of the phe-nomenon of monitoring effectiveness.

In addition to our empirical findings, our study haspractical implications. First, there is a possibility of cre-ating market intermediaries who could offer monitor-ing services more economically (Coffee 1991, Latham1998). Because these services would be targeted at insti-tutional owners, investors would have to incur greatermonitoring fees to acquire them, which means greatercosts to individual investors interested in encourag-ing greater monitoring or incentivizing fund managersto monitor. Second, in perhaps a more obvious solu-tion, portfolio diversification and monitoring capacityshould be considered in conjunction. Basic finance text-books indicate that 99% of market risk can be elimi-nated by using a portfolio of 100 stocks (Brealey andMeyers 2003). If so, more work needs to be done toclarify the diversification–monitoring tradeoff, especiallyas portfolios become larger. Third, because portfoliodiversification may weaken governance with respect tocompensation, it may encourage other alternate gover-nance mechanisms such as the market for corporate con-trol (Bhide 1994, Rediker and Seth 1995). Researchers


and practitioners need to be aware of both the benefitsand costs of large institutional owners’ portfolio diversi-fication in the context of corporate governance.

We believe that our findings have strong moorings,in that our sample consists of many large firms over along time period, and the firms in our sample have fairlyhigh levels of institutional ownership, a necessary con-dition for testing our hypotheses. However, our studyhas several limitations. For the sake of parsimony wefocused only on the largest institutional owner’s portfo-lio. Researchers indicate that, because of the free-riderproblem, only a large shareholder has the incentive toengage in costly monitoring (Gillan and Starks 2000).Consideration of several of the larger owners might pro-vide more insight, though it would also have to takeinto consideration that different owners could have con-flicting interests (Hoskisson et al. 2002). In addition, thenature of the available data limited our focus on largefirms, so it would be interesting to determine whether thefindings generalize to smaller firms. Finally, we focusedon executive compensation. Additional studies shouldconsider other issues such as valuation, productivity,research and development, and so forth.

To extend previous research (e.g., David et al. 1998,Graves 1988, Hartzell and Starks 2003), we investigateinstitutional ownership and portfolio characteristics atthe aggregate institutional level. Although focusing oninstitutional ownership at the broadest level makes sensein most cases, mutual funds and investment companiesconstitute the largest subset of owners within this broadcategory of ownership (Roe 1994), and the ownershipstakes of such institutions (e.g., Vanguard, Fidelity) maynot easily aggregate into a single category. For example,institutional owners may manage various mutual fundswith different investment objectives and different fundadvisers and managers dedicated to meeting their vari-ous clients’ objectives. Therefore, the monitoring needsof growth fund portfolios might not coincide with theneeds of income fund portfolios that belong to the samemutual fund family of a particular institutional investor.

Given these limitations, further research could proceedin two broad directions. First, a finer structural measure-ment of the institutional owners’ portfolios is warranted.For example, by considering Fidelity, a large institu-tional owner, we achieved simplicity in understanding itseffect, even though Fidelity represents the additive sumof different fund managers and advisers. By addressingdifferences within large owners (i.e., at the fund level),researchers may get a more accurate picture of thisphenomenon. Second, further research should exploresome other antecedents that could shed light on whylarge institutional investors monitor particular firms intheir portfolios, choose to be active with respect to cer-tain shareholdings, or neglect certain firms altogether.Although we know much about monitoring and share-holder activism, we have limited knowledge about why

large institutional investors choose to be active monitorsin some firms and not others.

Finally, though this study cannot provide conclu-sive evidence about the importance of the portfoliocharacteristics of institutional owners, it offers a meansby which to understand the possible negation of the own-ership influences at the firm level. Overall, we providepreliminary evidence that the portfolio characteristicsof the largest institutional owner tend to contradict thefirm-level effects; therefore, we firmly recommend thatfuture studies consider both firm- and portfolio-leveleffects simultaneously to explain executive compensa-tion and other organizational issues influenced by cor-porate governance.

AcknowledgmentsThe authors thank Eric Powers, Ed Zajac, and the anonymousreviewers for their feedback on this manuscript. A previousversion of this paper was presented at the Academy of Man-agement Meeting in 2005.

Appendix 1. Pay for Performance Sensitivity andMonitoring by Largest Institutional Owner—Replicating and Extending Hartzell and Starks (2003)(Ordinary Least Squares with Newey–West Correction)

Pay for performance sensitivity

Model 1 Model 2Variables std. coefficients std. coefficients

Step 1Constanta 0�000∗∗∗ 0�000∗∗∗

Change in 0�042∗ 0�027∗∗

shareholder wealtht

Change in 0�027∗ 0�017∗

shareholderwealtht−1

Tobin’s qt−1 −0�015 0�006Market −0�166∗∗∗ −0�112∗∗∗

capitalizationt−1Total institutional −0�115∗∗ −0�051∗

ownershipt−1Largest institutional 0�040∗ 0�034ownert−1

Step 2Portfolio concentration −0�092∗∗∗

Portfolio blockholding 0�026Portfolio turnover 0�021Firm significance −0�163∗∗∗

in portfolioNumber of 3,744observations

R-square (%) 7�3 9�8Change in R-square 2�5∗∗∗

Notes. a49 dummy codes controlling for SIC two-digit industry leveland eight dummy codes controlling for year effects are not reportedhere for brevity. The sample is reduced by one year due to the useof double-lagged independent variable (i.e., change in shareholderwealtht−1).

∗p < 0�05, ∗∗p < 0�01, ∗∗∗p < 0�001.


App

endi

x2.

Mon

itor

ing

ofCEO

Com

pens

ation

byth

eLar

gest

Institut

iona

lOwne

rW

hen

theLar

gest

Institut

iona

lOwne

rIs

aM

utua

lFun

d(O

rdin

ary

Lea

stSq

uare

swith

New

ey–W

estCor

rection)

CEO

totalcom

pensation

Payforpe

rformance

sensitivity

Prop

ortionof

salary

Prop

ortionof

continge

ntpa

y

Variables

Mod

el1

Mod

el2

Mod

el3

Mod

el4

Mod

el5

Mod

el6

Mod

el7

Mod

el8

Mod

el9

Mod

el13

Mod

el14

Mod

el15

Step

1Constanta

7�37

∗∗∗

7�63

∗∗∗

7�66

∗∗∗

0�62

∗∗0�24

0�33

−1�17∗

∗∗−1

�31∗

∗∗−1

�32∗

∗∗0�22

∗∗∗

0�25

∗∗∗

0�25

∗∗∗

Firm

size

0�21

∗∗∗

0�14

∗∗∗

0�14

∗∗∗

−0�25∗

∗∗−0

�13∗

∗∗−0

�10∗

∗−0

�11∗

∗∗−0

�07∗

∗−0

�07∗

∗0�01

∗0�00

0�01

Growth

0�13

∗∗∗

0�07

∗∗0�07

∗∗−0

�14∗

−0�07

−0�05

−0�12∗

∗∗−0

�09∗

∗∗−0

�09∗

∗∗0�02

∗∗∗

0�02

∗∗∗

0�02

∗∗∗

Profitability

0�03

−0�01

−0�01

−0�27∗

∗−0

�21∗

∗−0

�20∗

∗0�00

0�02

0�02

0�00

0�00

0�00

CEO

ownership

−0�07∗

∗∗−0

�06∗

∗−0

�06∗

∗0�12

0�11

0�11

0�06

∗∗∗

0�05

∗∗0�05

∗∗−0

�01∗

−0�01∗

−0�01∗

Chang

eof

CEO

0�06

∗∗0�06

∗∗∗

0�06

∗∗∗

0�07

∗0�06

0�06

∗−0

�09∗

∗∗−0

�09∗

∗∗−0

�09∗

∗∗0�01

∗∗0�01

∗∗0�01

∗∗CEO

duality

0�08

∗∗∗

0�05

∗∗0�05

∗∗−0

�17∗

∗∗−0

�13∗

∗−0

�14∗

∗−0

�03∗

−0�02

−0�02

0�01

0�00

0�00

CEO

tenure

0�03

0�03

0�03

−0�01

0�00

0�01

−0�03

−0�03

−0�03

0�00

0�00

0�00

Liqu

idity

−0�25∗

∗∗−0

�20∗

∗∗−0

�20∗

∗∗0�12

∗0�07

0�06

0�06

∗∗0�04

0�04

∗−0

�01

0�00

0�00

Financialleverag

e−0

�04∗

−0�03

−0�03

−0�05

−0�07

−0�07

0�06

∗∗∗

0�06

∗∗0�05

∗∗−0

�01∗

∗−0

�01∗

∗−0

�02∗

∗Sa

lary

>$1

million

0�20

∗∗∗

0�18

∗∗∗

0�18

∗∗∗

0�01

0�03

0�02

−0�03∗

−0�02

−0�02

0�00

−0�01

−0�01

Jensen’salph

a0�08

0�10

0�10

0�18

0�06

0�06

−0�04

−0�05

−0�06

−0�01

−0�01

−0�01

Firm

risk

0�02

0�00

0�00

0�05

0�16

0�14

−0�03

−0�03

−0�02

0�03

∗0�03

∗0�03

∗Lagg

edtotal

0�13

∗∗∗

0�10

∗∗∗

0�10

∗∗∗

compe

nsation

Lagg

edsalary

b0�25

∗∗∗

0�22

∗∗∗

0�22

∗∗∗

Lagg

edcontinge

nt0�32

∗∗∗

0�31

∗∗∗

0�31

∗∗∗

compe

nsation

prop

ortion

Institutionalownership

0�23

∗∗∗

0�20

∗∗∗

0�20

∗∗∗

0�02

0�14

∗∗0�13

∗∗−0

�09∗

∗∗−0

�09∗

∗∗−0

�09∗

∗∗0�03

∗∗∗

0�03

∗∗∗

0�03

∗∗∗

Fund

performance

−0�02

−0�02

−0�02

−0�02

−0�02

−0�03

0�00

0�00

0�00

0�00

0�00

0�00

Num

berof

fund

smanag

ed0�05

∗∗∗

0�04

∗∗0�04

∗∗0�03

0�08

∗0�07

∗−0

�03∗

−0�03∗

−0�03

0�00

∗0�00

0�00

Step

2Largestinstitutionalowner

−0�10∗

∗∗−0

�10∗

∗∗−0

�05

−0�02

0�08

∗∗∗

0�08

∗∗∗

−0�01∗

−0�02∗

∗Po

rtfolio

concentra

tion

0�13

∗∗∗

0�13

∗∗∗

−0�35∗

∗∗−0

�34∗

∗∗−0

�06∗

∗−0

�06∗

∗0�01

∗0�01

∗Po

rtfolio

blockholding

−0�02

−0�02

0�19

∗∗0�17

∗∗−0

�01

−0�01

0�01

0�01

Portfolio

turnover

0�00

0�00

0�00

0�00

−0�02

−0�02

0�01

0�01

Firm

sign

ificanceinpo

rtfolio

0�21

∗∗∗

0�22

∗∗∗

−0�26∗

∗∗−0

�24∗

∗∗−0

�11∗

∗∗−0

�11∗

∗∗0�02

∗∗∗

0�02

∗∗∗

Step

3Largestinstitutional

0�01

0�10

∗∗0�01

0�00

owner×

portfolio

concentra

tion

Largestinstitutionalowner

−0�03∗

−0�08∗

0�02

−0�01∗

×po

rtfolio

blockholding


−0�01

0�00

−0�01

−0�01∗

∗×po

rtfolio

turnover


0�01

0�07

0�00

0�00

×firm

sign

ificanceinpo

rtfolio

Adj.R

-squ

are(%

)45

�750

�550

�513

�816

�817

�234

�636

�436

�431

�031

�832

�0Chang

ein

R-squ

are

4�8∗

∗∗0�0

3�0∗

∗∗0�4∗

1�8∗

∗∗0�0

0�8∗

∗∗0�2

Notes.N

=2�428.

∗ p<0�05,∗

∗ p<0�01,∗

∗∗p<0�001.

a49

dummycode

scontrolling

forSICtwo-digitind

ustryleveland

nine

dummycode

scontrolling

foryear

effectsarenotrep

ortedhere

forbrevity.


Appendix 3. Pay for Performance Sensitivity and Institutional OwnershipTwo-Stage Least Squares

Largest institutionalFull sample owner is mutual fund

Inst. own PPS Inst. own PPS(1st stage) (2nd stage) (1st stage) (2nd stage)Model 1 Model 2 Model 3 Model 4

Stage 1Constanta 0�704∗∗∗ −3�250∗ 0�665∗∗∗ −3�990∗Share turnover 0�021∗∗∗ 0�022∗∗∗Firm size −0�000∗∗∗ −0�000∗∗∗

Stage 2Change in shareholder wealtht 0�000 0�000Change in shareholder wealtht−1 0�000 0�000Total institutional ownershipt−1 8�772∗∗∗ 9�559∗∗∗Tobin’s qt−1 −0�030 −0�018Market capitalizationt−1 −0�000 −0�000Largest institutional ownert−1 0�969 −0�070Portfolio concentrationt−1 −0�002∗∗∗ −0�002∗∗∗Portfolio blockholdingt−1 0�201 0�704∗Portfolio turnovert−1 0�302 −0�994Firm significance in portfoliot−1 −2�608∗∗∗ −2�744∗∗∗Fund performance 0�006Number of funds managed 0�000

Number of observations 3,744 3,744 2,230 2,230Adj. R-square (%) 22�4 6�5 26�5 8�4

a49 dummy codes controlling for SIC two-digit industry level and eight dummy codes controlling for year effects are notreported here for brevity. The sample is reduced by one year due to the inclusion of a double-lagged independent variable(change in shareholder wealtht−1).

∗p < 0�05, ∗∗p < 0�01, ∗∗∗p < 0�001.

Endnotes1Consider the following recent examples: 776 corporate gov-ernance proposals submitted by shareholders during the 2004proxy season, 40% were related to executive compensation(Georgeson Shareholder Report 2004). Institutional investorsseek to shape compensation through proxy voting and byformulating voting guidelines; for example, the institutionalowner CalPERS (California Personnel Employee RetirementSystem) not only voted against 43% of executive stock plans in1999–2000 but also withheld votes from members of compen-sation committees who had authorized “outrageous” compen-sation for CEOs and other top executives (Frederic W. Cookand Co. 2001).2As an anonymous reviewer pointed out, this issue is morecomplicated because of risk considerations. As research byZajac and colleagues (Beatty and Zajac 1994, Zajac andWestphal 1994) suggests, monitoring relates inversely toincentives, but risky firms (e.g., initial public offering firms)must rely more on monitoring than on incentives.3Although some data on CEO compensation are availablethrough ExecuComp from 1992, the majority of the data pointsare missing for that year.4Where P = market price of the stock on the date the optionwas granted; E = exercise price of the option grant; d= ln�1+dividend per share�; r = ln�1 + interest rate�, where interestrate is the interest on 10-year U.S. treasury bonds during thelast month of the year; and T = life of options, equal to the

time period until the expiration date of the option grant. Whenthe expiration date is not reported, we set the option’s lifeequal to 10 years, the period used by an overwhelming major-ity of the awards (Yermack 1995); and � = standard Black–Scholes volatility calculated over 60 months.5Our conclusions are essentially the same if we use FEM; theresults are available from the authors on request.6We are indebted to an anonymous reviewer for this insight.

ReferencesAgrawal, A., G. Mandelker. 1990. Large shareholders and the moni-

toring of managers. J. Financial Quant. Anal. 25(2) 143–161.

Agrawal, A., G. Mandelker. 1992. Shark repellents and the roleof institutional investors in corporate governance. ManagerialDecision Econom. 13(1) 15–22.

Aiken, L., S. West. 1991. Multiple Regression� Testing and Interpret-ing Interactions. Sage, Newbury Park, CA.

Almazan, A., J. C. Hartzell, L. Starks. 2005. Active institutionalshareholders and costs of monitoring: Evidence from executivecompensation. Financial Management 34(4) 5–34.

Amihud, Y., B. Lev. 1981. Risk reduction as a managerial motive forconglomerate mergers. Bell J. Econom. 12(2) 605–617.

Anderson, R. C., D. M. Reeb. 2003. Founding-family ownership andfirm performance: Evidence from the S&P 500. J. Finance 58(3)1301–1327.


Beatty, R. P., E. J. Zajac. 1994. Managerial incentives, monitoringand risk bearing: A study of executive compensation, ownership,and board structure in initial public offerings. Admin. Sci. Quart.39(2) 313–335.

Bebchuk, L., J. Fried. 2004. Pay without Performance� The UnfulfilledPromise of Executive Compensation. Harvard University Press,Cambridge, MA.

Bebchuk, L. A., J. M. Fried, D. I. Walker. 2002. Managerial powerand rent extraction in the design of executive compensation.Univ. Chicago Law Rev. 69(3) 751–846.

Bennett, J. A., R. W. Sias, L. T. Starks. 2003. Greener pastures andthe impact of dynamic institutional preferences. Rev. FinancialStud. 16(4) 1203–1238.

Berger, P. G., E. Ofek, D. L. Yermack. 1997. Managerial en-trenchment and capital structure decisions. J. Finance 52(4)1411–1438.

Berle, A. A., G. C. Means. 1932. The Modern Corporation and Pri-vate Property. Macmillan, New York.

Bethel, J. E., J. P. Liebeskind, T. Opler. 1998. Block share purchasesand corporate performance. J. Finance 53(2) 605–634.

Bhide, A. 1994. Efficient markets, deficient governance: US securitiesregulations protect investors and enhance market liquidity, butdo they alienate managers and shareholders? Harvard Bus. Rev.72(6) 129–139.

Bloom, M., G. T. Milkovich. 1998. Relationships among risk, incen-tive pay, and organizational performance. Acad. Management J.41(3) 283–297.

Brealey, R. A., S. Myers. 2003. Principles of Corporate Finance.Irwin/McGraw-Hill, New York.

Bushee, B. J. 1998. The influence of institutional investors on myopicR&D investment behavior. Accounting Rev. 73(3) 305–333.

Carleton, W. T., J. M. Nelson, M. S. Weisbach. 1998. The influenceof institutions on corporate governance through private negotia-tions: Evidence from TIAA-CREF. J. Finance 53 1335–1362.

Coffee, J. C. 1991. Liquidity versus control: The institutional investoras corporate monitor. Columbia Law Rev. 91(6) 1277–1368.

Conyon, M. J., S. I. Peck. 1998. Board control, remuneration com-mittees, and top management compensation. Acad. ManagementJ. 41(2) 146–157.

Core, J. E., R. W. Holthausen, D. F. Larcker. 1999. Corporate gov-ernance, chief executive officer compensation, and firm perfor-mance J. Financial Econom. 51 371–406.

Daily, C. M., J. L. Johnson, A. E. Ellstrand, D. R. Dalton. 1998.Compensation committee composition as a determinant of CEOcompensation. Acad. Management J. 41(2) 209–220.

Daily, M. C., D. R. Dalton, N. Rajagopalan. 2003. Governancethrough ownership: Centuries of practice, decades of research.Acad. Management J. 46(2) 151–158.

Dalton, D. R., C. M. Daily, S. T. Certo, R. Roengpitya. 2003. Meta-analyses of financial performance and equity: Fusion or confu-sion? Acad. Management J. 46(1) 13–26.

David, P., R. Kochhar, E. Levitas. 1998. The effect of institutionalinvestors on the level and mix of CEO compensation. Acad.Management J. 41(2) 200–208.

Del Guercio, D., J. Hawkins. 1999. The motivation and impact ofpension fund activism. J. Financial Econom. 52(3) 293–340.

Demsetz, H. 1983. The structure of corporate ownership and the the-ory of the firm. J. Law Econom. 26 375–390.

Demsetz, H. 1986. Corporate control, insider trading, and rates ofreturn. Amer. Econom. Rev. 76(2) 313–316.

Drucker, P. E. 1976. The Unseen Revolution� How Pension FundSocialism Came to America. Heinemann, London.

Dyl, E. A. 1988. Corporate control and management compensation:Evidence on the agency problem. Managerial Decision Econom.9 21–25.

Eakins, S. G., S. R. Stansell, P. E. Wertheim. 1998. Institutional port-folio composition: An examination of the prudent investmenthypothesis. Quart. Rev. Econom. Finance 38(1) 93–109.

Eisenhardt, K. 1989. Agency theory: An assessment and review. Acad.Management Rev. 14(1) 57–74.

Finkelstein, S., B. K. Boyd. 1998. How much does the CEO mat-ter? The role of managerial discretion in the setting of CEOcompensation. Acad. Management J. 41(2) 179–199.

Finkelstein, S., D. C. Hambrick. 1989. Chief executive compensation:A study of the intersection of markets and political processes.Strategic Management J. 10(2) 121–135.

Frederic W. Cook & Co. 2001. CalPERS Proxy Voting. http://www.fwcook.com.

Gaver, J. J., K. M. Gaver. 1993. Additional evidence on the asso-ciation between the investment opportunity set and corporatefinancing, dividend, and compensation policies. J. AccountingEconom. 16 125–160.

Georgeson Shareholder. 2004. Annual Corporate Governance Review�Annual Meetings, Shareholder Initiatives, Proxy Contests. http://www.georgesun.com/usa/download/acgr.2004.pdf.

Gillan, S. L., L. T. Starks. 2000. Corporate governance proposalsand shareholder activism: The role of institutional investors.J. Financial Econom. 57(2) 275–305.

Gomez-Mejia, L. R. 1994. Executive compensation: A reassessmentand a future research agenda. G. Ferris, ed. Research in Per-sonnel and Human Resources Management, Vol. 12. JAI Press,Greenwich, CT, 161–222.

Gomez-Mejia, L. R., T. M. Welbourne. 1989. Strategic design of exec-utive compensation programs. Gomez-Mejia, ed. Compensationand Benefits. Bureau of National Affairs, Inc., Washington, D.C.,216–269.

Graves, S. B. 1988. Institutional ownership and corporate R&D in thecomputer industry. Acad. Management J. 31(2) 417–428.

Greene, W. H. 2003. Econometric Analysis, 5th ed. Pearson Education(Singapore) Pte Ltd., Delhi, India.

Hambrick, D. C., S. Finkelstein. 1995. The effects of ownership struc-ture on conditions at the top: The case of CEO pay raises.Strategic Management J. 16 175–193.

Hartzell, J. C., L. T. Starks. 2003. Institutional investors and executivecompensation. J. Finance 58(6) 2351–2374.

Healy, P. M. 1985. The effect of bonus schemes on accounting deci-sions. J. Accounting Econom. 7 85–107.

Healy, P. M., K. G. Palepu. 2003. How the quest for efficiency cor-roded the market. Harvard Bus. Rev. (July) 76–85.

Hendry, J., P. Sanderson, R. Barker, J. Roberts. 2006. Owners ortraders? Conceptualizations of institutional investors and theirrelationship with corporate managers. Human Relations 59(8)1101–1132.

Hill, C. W., P. Phan. 1991. CEO tenure as a determinant of CEO pay.Acad. Management J. 34(3) 707–717.

Hill, C. W. L., S. A. Snell. 1989. Effects of ownership structure andcontrol on corporate productivity. Acad. Management J. 32(1)25–46.


Hirschman, A. O. 1970. Exit, Voice, and Loyalty� Responses toDecline in Firms, Organizations and States. Harvard UniversityPress, Cambridge, MA.

Holmstrom, B. 1979. Moral hazard and observability. Bell J. Econom.14 74–91.

Hoskisson, R. E., M. Q. Hitt, R. A. Johnson, W. Grossman. 2002.Conflicting voices: The effects of institutional ownership hetero-geneity and internal governance on corporate innovation strate-gies. Acad. Management J. 45(4) 697–716.

Hsiao, C. 1986. Analysis of Panel Data. Cambridge University Press,Boston.

Jensen, M. C., K. J. Murphy. 1990. CEO incentives—It’s not howmuch you pay, but how. Harvard Bus. Rev. 68(3) 138–153.

Jensen, M. C., K. J. Murphy. 2004. Remuneration: Where we’ve been,how we got to here, what are the problems, and how to fix them.Finance Working Paper 44/2004, ECGI and Harvard BusinessSchool, Boston.

Karpoff, J. M. 2001. The impact of shareholder activism on targetcompanies: A survey of empirical findings. Working paper, Uni-versity of Washington, Seattle.

Karpoff, J. M., P. H. Malatesta, R. A. Walkling. 1996. Corpo-rate governance and shareholder initiatives: Empirical evidence.J. Financial Econom. 42(3) 365–395.

Khan, R., R. Dharwadkar, P. Brandes. 2005. Institutional ownershipand CEO compensation: A longitudinal examination. J. Bus.Res. 58(8) 1078–1088.

Kochhar, R., P. David. 1996. Institutional investors and firm innova-tion: A test of competing hypotheses. Strategic Management J.17(1) 73–84.

Lambert, R. A., D. F. Larcker, K. Weigelt. 1991. How sensitive isexecutive compensation to organizational size? Strategic Man-agement J. 12(5) 395–402.

Latham, M. 1998. Corporate monitoring: New shareholder power tool.Financial Analysts J. 8(54) 9–15.

Maddala, G. S. 2002. Introduction to Econometrics, 3rd ed. JohnWiley & Sons, Ltd., Singapore.

McGuire, J., E. Matta. 2003. CEO stock options: The silent dimen-sions of ownership. Acad. Management J. 46(3) 255–265.

Mehran, H. 1995. Executive compensation structure, ownership, andfirm performance. J. Financial Econom. 38(2) 163–184.

Murphy, K. J. 1985. Corporate performance and managerial remuner-ation: An empirical analysis. J. Accounting Econom. 7 11–42.

Pedersen, T., S. Thomsen. 2003. Ownership structure and value ofthe largest European firms: The importance of owner identity.J. Management Governance 7(1) 27.

Pollock, T. G., H. M. Fischer, J. B. Wade. 2002. The role of power andpolitics in the repricing of executive options. Acad. ManagementJ. 45(6) 1172–1182.

Pozen, R. C. 2002. The Mutual Fund Business, 2nd ed. HoughtonMifflin, Boston.

Rediker, K. J., A. Seth. 1995. Boards of directors and substitutioneffects of alternate governance mechanisms. Strategic Manage-ment J. 16 85–99.

Roe, M. J. 1994. Strong Managers, Weak Owners� The Political Rootsof American Corporate Finance. Princeton University Press,Princeton, NJ.

Roth, K., S. O’Donnell. 1996. Foreign subsidiary compensation strat-egy: An agency theory perspective. Acad. Management J. 39(3)678–703.

Ryan, L. V., M. Schneider. 2002. The antecedents of institutionalinvestor activism. Acad. Management J. 27(4) 554–573.

Shin, J. Y. 2005. The composition of institutional ownership and thestructure of CEO compensation. Working paper, University ofWisconsin, Madison. http://ssrn.com/abstract=639797.

Shleifer, A., R. Vishny. 1986. Large shareholders and corporate con-trol. J. Political Econom. 94 461.

Shleifer, A., R. W. Vishny. 1997. A survey of corporate governance.J. Finance 52(2) 737–783.

Short, H., K. Keasey. 1997. Institutional shareholders and corpo-rate governance in the United Kingdom. K. Keasey, S. Thomp-son, M. Wright, eds. Corporate Governance� Economic,Management, and Financial Issues, Vol. 2. Oxford UniversityPress, West Sussex, England, UK, 18–53.

Sundaramurthy, C., D. W. Lyon. 1998. Shareholder governance pro-posals and conflict of interests between inside and outside share-holders. J. Management Issues 10 30–44.

Thomsen, S., T. Pedersen. 2000. Ownership structure and economicperformance in the largest European companies. Strategic Man-agement J. 21(6) 689–705.

Tihanyi, L., R. A. Johnson, R. E. Hoskisson, M. A. Hitt. 2003. Insti-tutional ownership differences and international diversification:The effects of boards of directors and technological opportunity.Acad. Management J. 46(2) 195–237.

Tolia, B. 1997. The CEO compensation puzzle: Theory and prac-tice. Unpublished doctoral dissertation, University of Wisconsin,Milwaukee.

Tosi, H. L., L. R. Gomez-Mejia. 1994. CEO compensation monitoringand firm performance. Acad. Management J. 37(4) 1002–1016.

Wahal, S. 1996. Pension fund activism and firm performance.J. Financial Quant. Anal. 31 1–23.

Westphal, J. D., E. J. Zajac. 1994. Substance and symbolism in CEOs’long-term incentive plans. Admin. Sci. Quart. 39 367–390.

Wiseman, R., L. R. Gomez-Mejia. 1998. A behavioral agency modelof risk taking. Acad. Management Rev. 23 133–153.

Woidtke, T. 2002. Agents watching agents?: Evidence from pensionfund ownership and firm value. J. Financial Econom. 63(2)99–132.

Wright, P., S. P. Ferris, A. Sarin, V. Awasthi. 1996. Impact of corpo-rate insider, blockholder, and institutional equity ownership onfirm risk taking. Acad. Management J. 39(2) 441–463.

Yermack, D. 1995. Do corporations award CEO stock options effec-tively? J. Financial Econom. 39(2–3) 237–269.

Zajac, E. J., J. D. Westphal. 1994. The costs and benefits of manage-rial incentives and monitoring in large U.S. corporations: Whenis more not better? Strategic Management J. 15 121–142.

institutional ownership and monitoring effectiveness: it's not just how much but what else you...

Documents