are ceos rewarded for luck? the ones without...

ARE CEOS REWARDED FOR LUCK THE ONESWITHOUT PRINCIPALS ARE

MARIANNE BERTRAND AND SENDHIL MULLAINATHAN

The contracting view of CEO pay assumes that pay is used by shareholders tosolve an agency problem Simple models of the contracting view predict that payshould not be tied to luck where luck is dened as observable shocks to perfor-mance beyond the CEOrsquos control Using several measures of luck we nd thatCEO pay in fact responds as much to a lucky dollar as to a general dollar Askimming model where the CEO has captured the pay-setting process is consis-tent with this fact Because some complications to the contracting view could alsogenerate pay for luck we test for skimming directly by examining the effect ofgovernance Consistent with skimming we nd that better governed rms paytheir CEO less for luck

I INTRODUCTION

CEO pay is usually viewed through the lens of principalagent models Under this contracting view pay is used to reducethe moral hazard problem that arises because CEOs often ownvery little of the rms they control Shareholders (perhaps actingthrough the board or the compensation committee) optimallydesign the pay package in order to increase the CEOrsquos incentive tomaximize rm value1 Simple models of the contracting viewgenerate one important prediction Shareholders will not rewardCEOs for observable luck By luck we mean changes in rmperformance that are beyond the CEOrsquos control Tying pay to

The results in this paper were previously circulated as part of a largerworking paper entitled ldquoDo CEOs Set Their Own Pay The Ones Without Princi-pals Dordquo We are extremely grateful to Daron Acemoglu Rajesh Aggarwal GeorgeBaker Patrick Bolton Peter Diamond Robert Gibbons Denis Gromb Brian HallBengt Holmstrom Caroline Hoxby Glenn Hubbard Lawrence Katz Jorn-SteffenPischke Nancy Rose David Scharfstein Robert Shimer Andrei Shleifer RichardThaler and seminar participants at the University of California at BerkeleyColumbia University the University of Chicago Harvard University the Mas-sachusetts Institute of Technology Princeton University and the National Bu-reau of Economic Research Corporate Finance Summer Institute 1999 for veryhelpful comments We thank Kenneth Ayotte and Michael Mitton for excellentresearch assistance Michael Haid for giving us access to his data set of oilcompanies and David Yermack for giving us access to his data on executivecompensation Financial support was provided by the Russell Sage Foundationthe Princeton Industrial Relations Section and the Princeton Center for EconomicPolicy Studies e-mail mbertranprincetonedu mullainmitedu

1 Murphy [1985 1986] is a forerunner of the vast empirical literature on thecontracting view Murphy [1999] and Abowd and Kaplan [1999] summarize theCEO pay literature Formal tests of the contracting view can be found in Gibbonsand Murphy [1990 1992] Garen [1994] Hubbard and Palia [1994] Bertrand andMullainathan [1999] and Aggarwal and Samwick [1999a 1999b]

copy 2001 by the President and Fellows of Harvard College and the Massachusetts Institute ofTechnologyThe Quarterly Journal of Economics August 2001

901

luck therefore cannot provide better incentives and will onlymake the contract riskier [Holmstrom 1979]2

This paper starts by examining whether or not CEOs are infact paid for luck using three measures of luck3 First we performa case study of the oil industry where large movements in oilprices tend to affect rm performance on a regular basis Secondwe use changes in industry-specic exchange rate for rms in thetraded goods sector Third we use year-to-year differences inmean industry performance to proxy for the overall economicfortune of a sector4 For all three measures we nd that CEO payresponds signicantly to luck In fact we nd that CEO pay is assensitive to a lucky dollar as to a general dollar Moreover theseresults hold as well for discretionary components of paymdashsalaryand bonusmdashas they do for options grants

These results are inconsistent with a simple contractingview Motivated by practitioners such as Crystal [1991] we pro-pose an alternative skimming which can explain these results[Bertrand and Mullainathan 2000a] The skimming view alsobegins with the separation of ownership and control but it arguesthat this separation allows CEOs to gain effective control of thepay-setting process itself Both because of entrenchment such aspacking the board with supporters and because of the complexityof the pay process many CEOs de facto set their own pay withlittle oversight from shareholders Their pay level then becomesconstrained by an unwillingness to draw shareholdersrsquo attentionPay for performance arises in the skimming view because goodperformance may ease these constraints in essence creating

2 Note our emphasis on observable luck In any model given the random-ness of the world CEOs (and almost everybody else) will end up being rewardedfor unobservable luck Note also our emphasis on the fact that this predictionholds in simple agency models As we will discuss shortly complications to theagency model can in principle alter this result

3 Blanchard Lopez-de-Silanes and Shleifer [1994] present suggestive evi-dence on pay for luck by showing that windfall gains from court rulings raise thepay of CEOs It is only suggestive since court rulings may not be luck but rathera result of the CEOrsquos work In other domains Shea [1999] independently performsan exercise similar to ours for baseball players

4 This last test very much resembles the approach followed in the relativeperformance evaluation (RPE) literature [Gibbons and Murphy 1990 Janakira-man Lambert and Larcker 1992 Aggarwal and Samwick 1999a] Problems canarise with RPE as a special case of luck Filtering this specic kind of luck may notbe optimal from an agency theoretical point of view As Gibbons and Murphy[1990] note relative performance evaluation can distort CEO incentives if theycan ldquotake actions that affect the average output of the reference grouprdquo Aggarwaland Samwick [1999b] develop a formal model along these lines By using othershocks to performance that are even more objectively beyond managerial inu-ence we circumvent these problems

902 QUARTERLY JOURNAL OF ECONOMICS

slack for the CEO In other words when the rm is doing wellshareholders are less likely to notice a large pay package To theextent that lucky dollars create slack as readily as general dol-lars pay for luck arises

Finding pay for luck however does not necessarily single outthe skimming model Complications to the agency model canmake it such that paying for luck is in fact optimal For examplesuppose that the value of a CEOrsquos human capital rises and fallswith industry fortunes One would then nd that pay correlateswith luck because the CEOrsquos outside wage moves with luckAnother possibility is that boards may tie pay to luck in order tomotivate CEOs to forecast or respond to luck shocks SubsectionIID discusses whether arguments such as these can really ex-plain the pay for luck relationship

To further differentiate skimming from these explanationswe empirically examine a direct implication of the skimmingmodel Skimming should be less prevalent in better governedrms Well-governed rms such as those with a large share-holder present on the board limit the CEOrsquos ability to capture thepay process We test this hypothesis using several measures ofgovernance presence of large shareholders (on the board andoverall) CEO tenure (interacted with the presence of large share-holders to better proxy for entrenchment) board size and frac-tion of directors that are insiders Consistent with skimming wegenerally nd that the better governed rms pay less for luck5

These effects are strongest for the presence of large shareholderson the board An additional large shareholder on the board re-duces pay for luck by between 23 and 33 percent Large share-holders are especially important as CEO tenure increases con-sistent with the idea that unchecked CEOs can entrenchthemselves over time If pay for luck were optimal we would haveexpected well-governed rms to pay for luck as much as (if notmore than) poorly governed rms For example whether or not alarge shareholder is present the CEO would have to be rewardedfor a rise in the value of his human capital These ndingssuggest that at least some of the pay for luck in poorly governedrms is due to skimming by CEOs

5 Whenever we refer to ldquoless pay for luckrdquo we mean that there is less pay forluck relative to the amount of pay for performance Thus these results would notbe driven by well-governed rms simply giving less overall pay for performanceIn fact we nd that governance correlates very little with pay for performanceonly with pay for luck

903ARE CEOS REWARDED FOR LUCK

II PAY FOR LUCK TEST

IIA Theoretical Background

A simple theoretical model will make more precise whatagency theory says about the reward for observable luck Con-sider a standard agency setup where risk-neutral shareholderstry to induce a risk-averse top manager to maximize rm perfor-mance Since the actions of the CEO can be hard to observeshareholders will be unable to sign a contract that species theseactions Instead shareholders will offer a contract to the CEOwhere her compensation level is made to depend on the rmrsquosperformance Let p represent rm performance and a the CEOrsquosactions which by assumption are unobservable to the sharehold-ers Firm performance depends on the actions of the CEO and onrandom factors We split the random factors into two componentsthose that can be observed by shareholders and those that cannotFor an oil rm the price of crude oil would be an observablerandom factor Letting o be the observable factor and u be theunobservable noise term we assume that performance can bewritten as p = a + d o + u

Under some technical conditions (CARA utility and Brown-ian motion for the performance process) Holmstrom and Milgrom[1987] calculate the optimal incentive scheme for this model Lets denote this incentive scheme Since shareholders can only ob-serve two variables p and o the incentive scheme could at mostdepend on these two variables In fact shareholders will onlyreward CEOs for performance net of the observable factor

(1) s 5 a 1 b ( p 2 d o) 5 a 1 b (a 1 u)

In other words the optimal incentive scheme lters the ob-servable luck from performance This is because leaving o in theincentive scheme provides no added benet to the principal as bydenition the agent has no control over o Motivating her on ohas no incentive effects Beyond providing no benet tying pay toluck actually costs the principal because the variance of theincentive scheme is higher and the principal must increase meanpay to compensate the risk-averse CEO

In practice explicit incentive contracts such as optionsrarely lter For example options are rarely if ever indexedagainst market performance This need not be inconsistent witha lack of ltering however It may be that the discretionarycomponents of pay such as salary and bonus are the ones used to


lter In theory these other components could adjust enough toundo the effect of the options value uctuating with luck Suchadjustment may happen if a board were to monitor luck and altereach yearrsquos salary bonus and number of new options granted sothat the CEOrsquos overall pay package remains free of luck

IIB Empirical Methodology

Within the agency framework most of the empirical litera-ture on CEO pay estimates an equation of the form

(2) y it 5 b p perfit 1 g i 1 x t 1 a X p Xit 1 e itwhere yit is total CEO compensation in rm i at time t perfit isa performance measure g i are rm xed effects x t are time xedeffects and Xit are rm- and CEO-specic variables such as rmsize and tenure The coefcient b captures the strength of the payfor performance relationship

Performance is typically measured either as changes in ac-counting returns or stock market returns and we will use bothmeasures6 In measuring compensation yit much of the literaturefocuses on the ow of new compensation Ideally the compensa-tion in a given year would also include changes in the value ofunexercised options granted in previous years [Hall and Liebman1998] Such a calculation requires data on the accumulated stockof options held by the CEO each year whereas existing data setsincluding ours contain only information on new options grantedeach year Consequently our compensation measure excludesthis component of the change in wealth For our purposes how-ever this exclusion does not pose much of a problem The changein wealth due to changing option values is mechanically tied toluck since options are not indexed Thus even if these data wereavailable focusing on the subjective components of pay wouldstill be a natural strategy We discuss this issue at greater lengthin subsection IID

To estimate the general sensitivity of pay to performance wewill follow the literature and estimate equation (2) using a stan-dard Ordinary Least Squares (OLS) model To estimate the sen-sitivity of pay to luck we need to use a two-stage procedure Inthe rst stage we will predict performance using luck in order toisolate changes in performance that are caused by luck In the

6 These are ow measures In practice given the rm xed effects we willuse market value and level of accounting returns as measures of perf i t


second stage we will see how sensitive pay is to these predictablechanges in performance This two-stage procedure is essentiallyan Instrumental Variables (IV) estimation where the luck vari-able is the instrument for performance7

Letting o be luck the rst equation we estimate is

perfit 5 b p o it 1 gi 1 c t 1 aX p Xit 1 e itwhere oit represents the luck measure (oil price for example)From this equation we predict a rmrsquos performance using only

information about luck Call this predicted value perfit We thenask how pay responds to these predictable changes in perfor-mance due to luck

y it 5 b Luck p perfˆ it 1 g i 1 x t 1 a X p Xit 1 e itThe estimated coefcient b L uck indicates how sensitive pay is tochanges in performance that come from luck Since such changesshould be ltered basic agency theory predicts that b Lu ck shouldequal 0

IIC Oil Industry Study

We now turn to the oil industry as a case study of pay forluck As Figures I and II show the price of crude oil has uctu-ated dramatically over the last 25 years These large uctuationshave caused large movements in industry prots Moreover theselarge uctuations in crude oil prices are likely to have beenbeyond the control of a single American CEO For example thesharp decline in crude oil price at the end of 1985 was caused bySaudi Arabiarsquos decision to reform its petroleum policy and toincrease production an action hardly attributable (and neverattributed) to the CEOs of American oil rms Similarly the largeoil price increase between 1979 and 1981 is usually attributed toan internal policy change by OPEC Oil price movements there-fore provide an ideal place to test for pay for luck they affect

7 One might wonder why we should use this procedure rather than simplyinclude o directly into the pay for performance equation (2) and run OLS toestimate

yit 5 b p perf it 1 f p o 1 g i 1 x t 1 a x p Xit 1 e it

This equation is hard to interpret however Even if there is no pay for luck thecoefcient f will not equal 2 b but rather 2 b d as we can see from equation (1)Since we do not estimate d the estimated coefcient f can be small either becausethere is pay for luck or simply because d is small The rst equation in the IVprocedure circumvents this problem by scaling the effect of luck on performance


performance are measurable and are plausibly beyond the con-trol of the CEOs

We use a data set on the pay and performance for the 51largest American oil companies between 1977 and 1994 to imple-

FIGURE IReal Price of a Barrel of Crude Oil

FIGURE IIMean Rate of Accounting Return


ment the methodology of the previous section8 Before moving toregression analysis it is useful to look directly at how pay uc-tuates compared with the movements in Figures I and II InFigure III we have graphed changes in oil prices for each year andchanges in mean log pay in the industry Two striking factsemerge First pay changes and oil price changes correlate quitewell In twelve of the seventeen years they are of the same signboth are up or both are down This is suggestive of pay for luckSecond the remaining ve years where pay and oil prices move inopposite directions are all years in which the oil price drops Thishints at an asymmetry while CEOs are always rewarded for goodluck they may not always be punished for bad luck

This graphical analysis does not quantify pay for luck Onecannot compare it with the size of the overall pay for perfor-mance It also does not control for other rm-specic variablesthat might be changing over time Table I follows the empirical

8 We are extremely grateful to Michael Haid for making the data set used inthis section available to us See Haid [1997] for further details about the construc-tion of the data set Appendix 1 provides mean and standard deviations for themain variables of interest While the original data set covers 51 companies overthe period 1977 to 1994 information on CEO pay is available for only 50 of theseoriginal 51 companies Moreover CEO pay is available from 1977 on for only 34of these 50 companies The nal data set we use covers 827 companyyearobservations

FIGURE IIIOil Industry CEO Pay and Crude Oil Price


methodology presented in subsection IIB which permits a moresystematic analysis All regressions use log (total compensation)as dependent variable and include rm xed effects age andtenure quadratics and a performance measure as dependentvariables9 We also include a year quadratic to allow for the factthat CEO pay has been trending up during this period Column(1) estimates the sensitivity of pay to a general change in account-ing performance The coefcient of 82 suggests that if an oil rmincreases its accounting return by one percentage point totalcompensation rises by 82 p 01 = 0082 log point Roughlya one percentage point increase in accounting returns leads to a8 percent increase in pay Note that the sign and magnitude of allthe other covariates in the regression seem sensible Pay in-

9 Total compensation in this table and all other tables includes salarybonus other incentive payments and value of options granted in that year

TABLE IPAY FOR LUCK FOR OIL CEOS (LUCK MEASURE IS LOG PRICE OF CRUDE OIL)

DEPENDENT VARIABLE ln (TOTAL COMPENSATION)

Specication General Luck General Luck(1) (2) (3) (4)

Accounting rateof return

82 215 mdash mdash(16) (104)

Log (shareholderwealth)

mdash mdash 38 35(03) (17)

Age 05 07 05 05(02) (03) (02) (02)

Age2p 100 2 04 2 05 2 04 2 04

(02) (02) (02) (02)Tenure 01 01 01 01

(01) (01) (01) (01)Tenure2

p 100 2 03 2 03 2 03 2 03(02) (01) (02) (02)

Firm xed effects Yes Yes Yes YesYear quadratic Yes Yes Yes YesSample size 827 827 827 827Adjusted R2 70 mdash 75 mdash

a Dependentvariable is the logarithm of total compensation Performance measure is accounting rate ofreturn in columns (1) and (2) and the logarithm of shareholder wealth in columns (3) and (4) All nominalvariables are expressed in 1977 dollars

b Summary statistics for the sample of oil rms are available in Appendix 1c The luck regression (columns (2) and (4)) instrument for performance with the logarithm of the price

of a barrel of crude oil in that year expressed in 1977 dollarsd Each regression includes rm xed effects and a quadratic in yeare Standard errors are in parentheses


creases with age and to a lesser extent with tenure Both the ageand tenure prole are concave (the negative coefcient on thequadratic term)

Column (2) estimates the sensitivity of pay to luck As de-scribed above we instrument for performance with the log of oilprice10 The coefcient in column (2) now rises to 215 Thissuggests that a one percentage point rise in accounting returnsdue to luck raises pay by 215 percent Given the large standarderrors one cannot reject that the pay for luck coefcient and payfor general performance coefcient are the same One can how-ever strictly reject the hypothesis of complete ltering oil CEOsare paid for luck that comes from oil price movements

Columns (3) and (4) perform the same exercise for a marketmeasure of performance shareholder wealth The coefcient of38 on column (3) suggests that a 1 percent increase in share-holder wealth leads to roughly a 38 percent increase in CEO payIn column (4) we nd that a 1 percent increase in shareholderwealth due to luck leads to 35 percent increase in CEO payAgain pay for luck matches pay for general performance

IID More General Tests

The oil industry case study while instructive raises thequestion of how generalizable these results are In this subsectionwe will examine luck shocks that affect a broader set of rms Wefocus on two measures of luck movements in exchange rates andmean industry performance By affecting the extent of importpenetration and hence foreign competition exchange rate move-ments can strongly affect a rmrsquos protability11 Movements inmean industry performance also proxy for luck to the extent thata CEO does not inuence how the rest of her industry performsAs we mentioned before this last instrument is more question-able In practice however we nd that mean industry move-ments operate exactly like exchange rate or oil price movements

To implement these tests we use compensation data on 792large corporations over the 1984 ndash1991 period The data set wasgraciously made available to us by David Yermack and AndreiShleifer It is extensively described in Yermack [1995] Compen-

10 This table does not report the rst-stage regressions of performance on oilprice But as one would expect from Figures I and II these regressions show verysignicant coefcients on oil price ( p lt 001)

11 Revenga [1992] uses exchange rates as an instrument for import pene-tration Bertrand [1999] shows its effects on rm protability


sation data were collected from the corporationsrsquo SEC Proxy10-K and 8-K llings Other data were transcribed from theForbes magazine annual survey of CEO compensation as well asfrom SEC Registration statements rmsrsquo Annual Reports directcorrespondence with rms press reports of CEO hires and depar-tures and stock prices published by Standard amp Poorrsquos Firmswere selected into the sample on the basis of their Forbes rank-ings Forbes magazine publishes annual rankings of the top 500rms on four dimensions sales prots assets and market valueTo qualify for the sample a corporation must appear in one ofthese Forbes 500 rankings at least four times between 1984 and1991 In addition the corporation must have been publicly tradedfor four consecutive years between 1984 and 1991

Yermackrsquos data are attractive in that they provide both gov-ernance variables and information on options granted not justinformation on options exercised But they do not include changesin the value of options held which we must therefore excludefrom our compensation measure If anything this biases us to-ward understanding the amount of pay for luck Since options arenot indexed changes in the value of options held will covaryperfectly with luck Including these changes in the compensationmeasure would only increase the measured pay for luck Thisdata limitation therefore is less of a concern for our purposes

Table II presents summary statistics for the main variablesof interest in the full Yermack data12 All nominal variables areexpressed in 1991 dollars The average CEO earns $900000 insalary and bonus His total compensation is nearly twice thatamount at $1600000 The difference indicates the large fractionof a CEOrsquos pay that is due to options grants The average CEO isroughly 57 years old and has been CEO of the rm for nine yearsAs far as governance goes the average rm in our sample has112 large shareholders of which less than a fourth are sitting onthe board There are on average thirteen directors on a boardForty-two percent of them are insiders13

12 In practice depending on the required regressors the various tests in thefollowing sections will be performed on various subsamples of the original dataNone of these main variables of interest signicantly differ in any of thesesubsamples

13 Technically we dene insiders to be both inside and gray directors Aninside director is dened as a director who is a current or former ofcer of thecompany A gray director is a relative of a corporate ofcer or someone who hassubstantial business relationships with the company outside the course of regularbusiness


Our rst general measure of luck focuses on exchange ratemovements We exploit the fact that exchange rates between theU S dollar and other country currencies uctuate greatly overtime We also exploit the fact that different industries are affectedby different countriesrsquo exchange rates For example since the toyindustry may be more affected by Japanese imports while thelumber industry may be more affected by Bolivia these twoindustries may experience very different shocks in the same yearThis allows us to construct industry-specic exchange rate move-ments which are arguably beyond a specic CEOrsquos control sincethey are primarily determined by macroeconomic variables Theexchange rate shock measure is based on the weighted average ofthe log real exchange rates for importing countries by industryThe weights are the share of each foreign countryrsquos import in totalindustry imports in a base year (1981ndash1982) Real exchange ratesare nominal exchange rates (expressed in foreign currency perdollar) multiplied by U S CPI and divided by the foreign countryCPI Nominal exchange rates and foreign CPIs are from theInternational Financial Statistics of the International MonetaryFund

TABLE IISUMMARY STATISTICS FULL YERMACK CEO SAMPLE

Mean S D

Age of CEO 5742 684Tenure of CEO 910 808Salary and bonus 90169 79515ln (Salary and bonus) 662 60Total compensation 159585 348832ln (Total compensation) 698 81Number of large shareholders (All) 112 142Number of large shareholders on board 24 74Board size 1345 454Fraction of insiders on board 42 19

a Sample period is 1984ndash1991b All nominal variables are expressed in thousands of 1991 dollarsc Column 1 is the mean for each variable while column 2 is the standard deviationd A large shareholder is dened as someone who owns more than 5 percent of the common shares in

the company excluding the CEO Insiders here denote directors who are current or former ofcers of thecompany relatives of corporate ofcers or anyone who has a substantial business relationship withthe company Relationships arising in the normal course of business would not be called an insider Ourinsider denition corresponds to insider + gray in the original Yermack data

e Total compensation equals salary plus bonus plus total value of options grants plus othercompensation


Panel A of Table III examines this luck measure Note thatsince the exchange rate measure can only be constructed forindustries where we have imports data the sample size is muchsmaller here than for our full sample All regressions control forrm and year xed effects as well as for quadratics in tenure andage14 Column 1 uses as dependent variable the level of cash

14 We do not report the coefcients on age and tenure to save space butthey resemble the oil industry ndings positive but diminishing effects of tenureand age

TABLE IIIPAY FOR LUCK

Dep var Cash comp ln (cash)ln (totalcomp) ln (cash)

ln (totalcomp)

Specication General Luck General Luck General Luck General Luck General Luck

Panel A Luck Measure is Exchange Rate Shock

Income 17 35 mdash mdash mdash mdash mdash mdash mdash mdash(02) (16)

Incomeassets mdash mdash 213 294 236 439 mdash mdash mdash mdash(16) (128) (28) (217)

ln (shareholderwealth)

mdash mdash mdash mdash mdash mdash 22 32 31 57(02) (13) (03) (23)

Sample size 1737 1737 1729 1729 1722 1722 1713 1713 1706 1706Adjusted R2 75 mdash 75 mdash 58 mdash 75 mdash 59 mdash

Panel B Luck Measure is Mean Industry Perfo rmance






a Dependent variable is the level of salary and bonus in columns 1 and 2 the logarithm of salary andbonus in columns 3 4 7 and 8 and the logarithm of total compensation in columns 5 6 9 and 10Performance measure is operating income before extraordinary items in columns 1 and 2 (in millions)operating income to total assets in columns 3 to 6 and the logarithm of shareholder wealth in columns 7 to10 All nominal variables are expressed in real dollars

b In the luck regressions in Panel A the performance measure is instrumented with current and laggedappreciation and depreciation dummies and current and lagged exchange rate index growth First-stageregressions are presented in Appendix 2

c In the luck regressions in Panel B the performance measure is instrumented with the total assets-weighted average performance measure in the rmrsquos two-digit industry (the rm itself is excluded from themean calculation)

d Each regression includes rm xed effects year xed effect and demographic controls (quadratics inage and tenure)

e Standard errors are in parentheses


compensation Thus relative to our standard specication we donot run this regression in logs and do not include value of optionsgranted Since prots are reported in millions and pay is reportedin thousands the coefcient of 17 in column 1 suggests that$1000 increase in prots leads to a 17 cent increase in perfor-mance Column 2 performs the same exercise for pay for luck weinstrument for performance using the exchange rate shocks15 Asin the oil case we nd a pay for luck coefcient that is of the sameorder of magnitude as the pay for general performance coefcient

Columns 3 through 6 run the more standard regressionwhere we use the logarithm of pay and an accounting measure ofperformance (operating income divided by total assets) In col-umns 3 and 4 we use only cash compensation while in columns 5and 6 we use total compensation In both cases we nd thesensitivity of pay to luck to be about the same as the sensitivity ofpay to general performance When accounting performance risesby one percentage point compensation (either total or cash) risesby about 2 percent whether that rise was due to luckmdashexchangerate movementsmdashor not

Columns 7 through 9 replicate these four columns for marketmeasures of performance Again we nd pay for luck thatmatches the pay sensitivity to a general shock A 1 percentincrease in shareholder wealth raises pay (again either total orcash) of about 3 percent irrespective of whether this rise wascaused by luck or not

Two important points should be taken away from this panelFirst the average rm rewards its CEO as much for luck as itdoes for a general movement in performance There seems to bevery little if any ltering at all Since we use a totally differentshock these ndings address theoretical concerns about the useof mean industry shocks (such as those raised in Gibbons andMurphy [1990] and Aggarwal and Samwick [1999b]) and showthat the lack of ltering observed in RPE ndings generalizes toother sources of luck

Second there is as much pay for luck on discretionary com-ponents of pay (salary and bonus) as there is on other components

15 First-stage regressions are reported in Appendix 2 In practice we use asinstruments continuous variables for exchange rate appreciation (current andlagged) as well as dummies to allow for nonlinear effects of appreciation (alsocurrent and lagged) The dummies are formed at the 2 and 4 percent cutoffs bothfor appreciation and depreciation The instruments are jointly highly signicantwith rst-stage p-values of less than 1 percent in all cases


such as options granted This rules out the notion that pay forluck mechanically arises because rms commit (implicitly or ex-plicitly) to multiyear stock option plans where the number ofoptions grants is xed ahead of time Under such plans as rmvalue rises so does the value of precommitted options grants[Hall 1999] Because salary and bonus are the most subjectivecomponents of pay nding pay for luck on these variables is verysuggestive Boards are rewarding CEOs for luck even when theycould lter it

In Panel B of Table III we replicate Panel A except that ourmeasure of luck becomes mean performance of the industrywhich is meant to capture external shocks that are experiencedby all the rms in the industry More specically as an instru-ment for rm-level rate of accounting return in a given year weuse the weighted average rate of accounting return in that year inthe two-digit industry that rm belongs to excluding the rmitself from the calculation16 The weight of a given rm in a givenyear is the share of its total assets in the aggregate ldquototal assetsrdquoof the two-digit industry the rm belongs to Similarly as aninstrument for rm-level logarithm of shareholder wealth in agiven year we use the weighted average of the log values ofshareholder wealth in the two-digit industry in that year againexcluding the rm itself from the calculation and using totalassets to weight each individual rm17

As in Panel A all regressions include rm xed effects andyear xed effects We also control for a quadratic in CEO age anda quadratic in CEO tenure The regressions include more thantwice the data points of Panel A because we can now use all rmsnot only those in the traded goods sector Panel B shows a patternquite similar to Panel A The pay for luck relationship in allspecications again roughly matches the pay for general perfor-mance Besides reinforcing the ndings of Panel A these latestndings suggest that previous RPE results arose probably notbecause of mismeasurement of the reference industry or of theindustry shock but because of true pay for luck

16 We also investigated the use of one-digit and three-digit industry meansas instruments and found qualitatively similar results

17 These mean industry performance measures are constructed fromCOMPUSTAT To maximize consistency of the performance measures betweenYermackrsquos rms and the rest of industry we also compute shareholder wealth andincome to assets ratios from COMPUSTAT for the rms in Yermackrsquos Becausenot all rms in Yermackrsquos data are present in COMPUSTAT in every year we loseabout 800 rm-year observations


III WHY IS THERE PAY FOR LUCK

The results so far clearly establish pay for luck There areseveral reactions possible to this evidence First one could take itas evidence of skimming To understand how the skimming modelpredicts pay for luck consider a CEO who has captured the payprocess His primary worry in setting pay will be that outrageousskimming may cause otherwise passive investors to stand up andnotice Good performance however provides the CEO with extraslack For example shareholders may scrutinize a rm moreclosely during bad times This allows higher pay when perfor-mance is good and produces a positive link between pay andperformance but for different reasons than in the contractingview If good performance creates slack irrespective of whether itwas lucky pay for luck will result18

Alternatively one could argue that pay for luck is in factoptimal and that the evidence so far is consistent with the con-tracting view One reason why pay for luck might be optimal isthat the CEOrsquos outside option may in fact depend on luck Whenthe oil industry enjoys a good fortune the human capital of oilCEOs may simply become more valuable Firms then pay their oilCEOs more simply to match their increased outside optionsThus pay for luck is optimal here not as an incentive device butmerely because the optimal level of pay increases with luck19

Objections can be raised against this view First our sugges-tive evidence of asymmetry in pay for luck may be hard to recon-cile with this view Average CEO compensation in the oil industryalways goes up when the price of crude oil goes up but does notalways go down when the price of crude oil goes down In our fullsample we performed a similar test using the industry luckshock For accounting measures we again found that pay re-sponds more to positive industry shocks than to negative ones(with no asymmetry on general pay for performance) For market

18 The scrutiny of otherwise passive investors may be triggered by absoluteperformance for several reasons The very nature of deciding where to pay atten-tion requires focusing on variables immediately at hand Passive investors mayuse accounting returns earnings growth or stock return in deciding when to acton a rm and all these variables are clearly not preltered The idea that outra-geous pay actually produces political intervention of some form has been pointedout in Jensen and Murphy [1990] Empirical work on this can be found in JoskowRose and Wolfram [1996]

19 While the argument made here is rather imprecise Himmelberg andHubbard [2000] provide a formal model as well as empirical results that interpretpay for luck in this light


measures we could not reject symmetry Second it is unclearwhy a CEOrsquos human capital should become more valuable asindustry fortunes rise In fact it may be exactly in bad times thathaving the right CEO is most valuable A priori either relation-ship seems plausible To test this assumption we examined turn-over in the CEO market We found no statistically signicantrelationship between a CEOrsquos turnover and industry returns(after controlling for the rmrsquos returns) and a point estimate thatwas negative This suggests that if anything turnover is coun-tercyclical Third we tested the effect of the industryrsquos averageCEO turnover rate on pay for luck If pay for luck were caused bymarket competition for CEOs then industries with higher turn-over should exhibit the greatest pay for luck For accountingmeasures we found that industries with the highest turnover infact showed the least pay for luck For market measures of per-formance we found no relationship between industry turnoverand pay for luck Of course for the last two ndings one couldalways argue that competitive pressures operate through thethreat of turnover rather than through actual turnover As awhole though we have been unable to nd positive evidence thatoutside bidding up of CEO wages could explain our results

Another reason why pay for luck may be optimal is that onemay want to provide incentives to the CEO to forecast or respondto luck20 This kind of argument can be most readily evaluated inour oil industry application Suppose that a particularly talentedCEO in the oil industry understood the political subtleties of theArab countries and forecast the coming of the positive oil shock atthe beginning of the 1980s By increasing output from existing oilwells increasing inventories or intensifying search for new wellshe could have increased his rmrsquos prots when the shock didcome Shouldnrsquot shareholders reward this farsighted CEO Theimportant point here is that those CEOs who were exceptional inhaving forecast should indeed be rewarded But this is not whatwe test for We use none of the between rm variation in responseto the oil shock We merely test whether the average rm expe-riences a rise (or fall for the negative shocks) in pay Put another

20 An argument similar to hedging has been made by Diamond [1998]Tying pay to luck may generate incentives for the CEO to change his correlationwith the luck variable In practice diversication seems to be more in the interestof management than shareholders Tufano [1996] for example demonstrates thatmanagerial characteristics such as share or option ownership are quite predic-tive of risk management in a sample of gold rms


way our results suggest that a CEO who responds to the shockexactly the same way as every other oil CEO is rewarded Thiscannot be a reward for having forecast well Again one maywant to reward CEOs for exceptional responsiveness to shocksbut there is little reason to reward them for just averageresponsiveness

A nal set of responses to pay for luck would be to abandonthe literal contracting view and argue that ltering out luck issimply impossible This might be because of cognitive complexityin understanding what is luck and what is not luck Part of thiscognitive complexity may be a pure information issue if there arenot enough data available to gure out the appropriate effect ofluck For example estimating the coefcient d in equation (1) maysimply not be possible Part of the cognitive complexity may bepsychological as in the evidence on the fundamental attributionerror [Durell 1999]21 None of the evidence so far directly refutesthis argument

IIIA The Effect of Governance

While we have argued against some of the various extensionsof the simple agency model in the end we still believe that theymerit serious consideration They suggest to us that the pay forluck nding does not per se rule out agency models The resultsare also consistent with the idea that ltering out luck is just notfeasible Therefore we now turn to testing a specic prediction ofthe skimming view rather than arguing against the other viewsSince the skimming view emphasizes the CEOrsquos ability to gaincontrol of the pay process corporate governance should play animportant role in skimming It is exactly in the poorly governedrms where we expect CEOs to most easily gain control of the payprocess This suggests that we should expect more pay for luck inthe poorly governed rms22

21 Another possibility is that luck is not contractible In practice we do notbelieve this is important for most of our ndings First it is difcult to believe thatnoncontracting issues can explain our results in the oil industry case study theprice of crude oil can easily be measured and written into a contract Second evenin the presence of noncontractibility subjective performance evaluation shouldeffectively lter (see Baker Gibbons and Murphy [1994])

22 Would the alternative explanations above predict an effect of governanceFirst one might argue that better governance increases the efcacy of monitoringmechanisms and hence reduces the need to pay for performance This argumentwould predict a change in the overall level of pay for performance not only in payfor luck We circumvent this problem by looking at the change in pay for luckrelative to the change in general pay for performance As an aside general pay for


To examine how pay for luck differs between well-governedand poorly governed rms we estimate two equations First inorder to provide a baseline we ask how pay for general perfor-mance (not luck) differs between well-governed and poorly gov-erned rms We estimate an OLS equation similar to equation (2)except that we allow the pay for performance coefcient to dependon governance

(3) y it 5 b p perfit 1 u p (Gov it p perfit)

1 g i 1 x t 1 a X p Xit 1 a G p Govit 1 e itwhere Govit is a measure of governance To understand thisequation differentiate both sides with respect to performance toget shy yit shy perfit = b + u p Govit In words this specicationallows the pay for performance sensitivity to be a function of thegovernance variable A positive value for u would imply thatbetter governed rms show greater pay for performance

Equation (3) of course tells us nothing about pay for luckmerely about pay for performance To get at pay for luck wereestimate this equation using our two-stage instrumental vari-ables procedure23 We then compute an estimate of the effect ofgovernance on pay for general performance uˆ and an estimate ofthe effect of governance on pay for luck uˆ Lu ck

Our test then consists in comparing uˆ and uˆ L uck We willspeak of more pay for luck in poorly governed rms when poorlygoverned rms display more pay for luck relative to pay forgeneral performance If poorly governed rms simply gave morepay for performance and pay for luck rose as a consequence wewould not refer to this as more pay for luck In practice we willsee that it is pay for luck that changes with governance while payfor performance hardly changes We will also verify that these

performance does not systematically correlate with governance Second for the-ories that rely on changes in the value of the CEOrsquos human capital it is unclearwhy these changes would happen more in the poorly governed rms Finally thepresumption that ltering is somehow cognitively impossible would clearly berefuted if some rms could lter

23 An extremely important caveat here our approach allows for the possi-bility that better governed rms may have a different responsiveness of perfor-mance to luck So for example if well-governed rms were merely less responsivein their performance (not their pay) to luck this would not create a spurious payfor luck difference between well-governed and poorly governed rms Technicallyperformance perf it the endogenous variable we need to instrument appears bothdirectly and indirectly (the term Govi t p perf it) in this equation When weinstrument we perform two rst stages one for the direct effect perf i t and one forthe interaction term Govi t p perf it This procedure is crucial because it allows theeffect of luck on performance to depend on governance


results are robust to allowing pay for luck to vary by rm sizeOtherwise one might simply worry that governance is a proxy forsize24

In Table IV we implement this framework for the case oflarge shareholders We ask whether the presence of large share-holders affects pay for luck Shleifer and Vishny [1986] amongothers argue that large shareholders improve governance in arm A single investor who holds a large block of shares in a rm

24 One must also be careful in interpreting the results from this exerciseThey are merely suggestive of the cross-sectional relationship between gover-nance and the extent of skimming They do not necessarily imply that a policy ofchanging a specic governance variable will necessarily lead to a change in theextent of skimming To make strong policy suggestions such as this one wouldneed more exogenous variation in governance and see its effects on CEO pay suchas in Bertrand and Mullainathan [1999] The governance results here are usefulbecause they demonstrate the relevance of the skimming view and not becausethey isolate policy mechanisms to reduce skimming

TABLE IVLARGE SHAREHOLDERS AND PAY FOR LUCK (LUCK MEASURE IS MEAN INDUSTRY

PERFORMANCE) DEPENDENT VARIABLE ln (Total Compensation)

Governance measure Large shareholders Large shareholders on board

SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)

Income Assets 218 459 mdash mdash 214 449 mdash mdash(238) (912) (217) (882)

Governanc e 2 094 2 416 mdash mdash 2 181 2 148 mdash mdashIncomeassets (094) (204) (176) (396)ln (shareholder

wealth)mdash mdash 249 383 mdash mdash 258 318

(018) (219) (017) (199)Governanc e mdash mdash 001 2 066 mdash mdash 2 019 2 076Log (shareho lder

wealth)(007) (036) (016) (053)

Governance 2 009 018 2 017 411 2 006 084 100 480(011) (018) (049) (240) (021) (033) (108) (356)

Sample size 4610 4610 4570 4570 4621 4621 4581 4581Adjusted R2 695 706 694 706

a Dependent variable is the logarithm of total compensation performance measure is operating incometo total assets All nominal variables are expressed in real dollars

b In all the luck regressions both the performance measure and the interaction of the performancemeasure with the governance measure are instrumented The instruments are the asset-weighted averageperformance in the two-digit industry and the interactions of the industry performance with that governancemeasure

c ldquoLarge shareholdersrdquo indicates the number of blocks of at least 5 percent of the rmrsquos common shareswhether the block holder is or is not a director ldquoLarge shareholders on boardrdquo indicates the number of blocksof at least 5 percent of the rmrsquos common shares that are held by directors of the board

d Each regression includes rm xed effects year xed effects a quadratic in age and a quadratic intenure



will have greater incentives to watch over the rm than a dis-persed group of small shareholders25 In our context the idea oflarge shareholders ts most naturally as this matches the intui-tion of ldquohaving a principal aroundrdquo Yermack data contain avariable that counts the number of individuals who own blocks ofat least 5 percent of the rmrsquos common shares When the CEOhappens to own such a block we exclude this block from thecount We further know whether these large shareholders are onthe board or not A priori one might expect that large sharehold-ers on the board have the greatest impact They can exert theircontrol not just through implicit pressure or voting but also witha direct voice on the board Since the information is available wewill consider the effect both of all large shareholders and of onlythose on the board

The rst four columns of Table IV use all large shareholdersas our measure of governance All regressions include the usualcontrols Column (1) estimates how the sensitivity of pay to per-formance depends on governance for accounting measures of per-formance The rst row tells us that a rm with no large share-holders shows a sensitivity of log compensation to accountingreturn of 218 An increase in accounting return of one percentagepoint leads to an increase in pay of about 2 percent The secondrow tells us that adding a large shareholder only weakly de-creases the sensitivity of pay to general performance and thiseffect is not statistically signicant For example a one percent-age point increase in accounting return now leads to a 209percent increase in pay when the rm has one large shareholder(compared with 218 in the absence of any large shareholder)Column (2) estimates how large shareholders affect pay forluck26 As before the rst row tells us that there is signicant payfor luck The second row here however tells us that this pay forluck diminishes signicantly in the presence of a large share-holder A one percentage point increase in accounting returns dueto luck leads to roughly a 46 percent increase in pay when thereis no large shareholder but only a 42 percent increase in pay

25 They also point out a possible opposing effect very large shareholdersmay have a greater ability to expropriate rents for themselves This effect is likelyto be greatest in other countries where investor protection is weakest Empiricalevidence on the efcacy of large shareholders can be found in Zeckhauser andPound [1990] Shivdasani [1993] and Denis and Serrano [1996]

26 In all that follows we will use mean industry performance as our mea-sure of luck since this produces the most powerful rst stages in the IVframework


when there is one more large shareholder Each additional largeshareholder decreases this effect by 4 percent This is a 10percent drop in the pay for luck coefcient for each additionallarge shareholder27

Columns (3) and (4) estimate the same regressions usingmarket measures of performance In this case the pay for generalperformance does not depend at all on the existence of a largeshareholder (a coefcient of 001 with a standard error of 007)We again nd however that pay for luck diminishes with thepresence of a large shareholder While the result is only signi-cant at the 10 percent level the economic magnitude is largerThe pay for luck coefcient now drops 066383 rsquo 17 percent foreach large shareholder

In columns (5) through (8) we repeat the above exercise butalter the governance measure We now focus only on large share-holders on the board Comparing columns (6) and (2) we see thatthe governance effect strengthens signicantly with respect tothe ltering of accounting performance We see that the pay forluck drops by 33 percent for each additional large shareholderThe results are very statistically signicant On market perfor-mance measures we nd the effect also rises but less dramati-cally In column (8) the pay for luck drops 23 percent with eachlarge shareholder on the board Moreover this last result isinsignicant In summary our ndings in Table IV highlight howlarge shareholders (especially those on the board) affect the ex-tent of pay for luck Firms with more large shareholders show forless pay for luck

The results in Table IV simply compare rms with largeshareholders with rms without This ignores the effects of CEOtenure another important determinant of governance A commonbelief is that CEOs who have been with the rm longer have hada chance to become entrenched perhaps by appointing friends onthe board In this case we would expect high tenure CEOs to

27 One may recall from subsection IID that we are excluding the cumulatedoptions from our measure of compensation as these will mechanically covary withstock price and hence luck However if rms with large shareholders providedmore options they may effectively provide more pay for luck because a biggerfraction of compensation (and hence for pay for luck) comes from the mechanicalportion To test at least this presumption we compared the fraction of totalcompensation that were options grants between well-governed and poorly gov-erned rms We found no consistent economically or statistically signicant dif-ference for our governance measures


show the greatest pay for luck Moreover we would expect thiseffect to be strongest in those rms where governance is weak andthere is no large shareholder present to limit the increased en-trenchment Hence in the absence of large shareholders weexpect fairly strong governance early in a CEOrsquos tenure but thisgovernance should weaken over time as he entrenches himself Inthe presence of large shareholders we not only expect strongergovernance but also that this stronger governance should lastthroughout the CEOrsquos tenure It is harder for a CEO to beginstacking the board when there is a large shareholder aroundThus we expect a rise in pay for luck with tenure in the absenceof a large shareholder but less of a rise (or even no rise) in thepresence of a large shareholder

Table V tests this idea We rst sort rms into two groupsbased on whether they have a large shareholder present on the

TABLE VTENURE LARGE SHAREHOLDERS AND PAY FOR LUCK (LUCK MEASURE IS MEAN

INDUSTRY PERFORMANCE) DEPENDENT VARIABLE ln (TOTAL COMPENSATION)

Any large shareholder on the board

No Yes No Yes

SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)

Income Assets 214 335 128 247 mdash mdash mdash mdash(30) (96) (65) (260)

CEO tenure p 00 13 063 2 006 mdash mdash mdash mdashIncomeassets (02) (05) (045) (131)Log (shareholder

wealth)mdash mdash mdash mdash 24 26 27 53

(02) (24) (05) (32)CEO tenure p mdash mdash mdash mdash 003 009 2 005 2 013Log (sh wealth) (001) (005) (003) (010)CEO tenure 01 2 00 010 016 2 002 2 045 044 084

(00) (01) (011) (016) (01) (04) (020) (059)Sample size 3884 3884 740 740 3841 3841 743 743Adjusted R2 7030 757 715 700

a Dependentvariable is the logarithm of total compensation All nominal variables are expressed in realdollars

b In all the luck regressions both the performance measure and the interaction of the performancemeasure with the CEO tenure are instrumented The instruments are the asset-weighted average perfor-mance in the two-digit industry and the interactions of the industry performance with the CEO tenure

c Sample in columns (1) (2) (5) and (6) is the set of rm-year observations for which there is no largeshareholder sitting on the board of directors sample in columns (3) (4) (7) and (8) is the set of rm-yearobservations for which there is at least one large shareholder sitting on the board of directors




board28 This produces 740 or so data points for rms with largeshareholders and 3880 or so data points for rms without largeshareholders For each set we now separately estimate regression3 for these two groups with tenure as our governance measure

Columns (1) and (2) focus on accounting measures of perfor-mance in rms without a large shareholder The second row tellsus that while tenure does not affect pay for performance itgreatly increases pay for luck In fact a CEO with (roughly) themedian tenure of nine years shows about 13 p 9335 rsquo 35percent greater pay for luck than one who just began at the rmLet us contrast this with columns (3) and (4) which estimate thesame effect for rms with a large shareholder present Here wend that tenure does not affect pay for luck at all while ifanything it seems to raise pay for performance slightly [Gibbonsand Murphy 1992] Thus pay for luck increases with tenure inthe absence of a large shareholder but does not change withtenure in the presence of a large shareholder

Columns (5) through (8) repeat the tests of columns (1)through (4) for market measures of performance Here the resultsare less stark but still very suggestive Comparing columns (6)and (5) we see that both pay for performance and pay for luck risewith tenure but pay for luck rises three times as fast (003 versus009) The coefcient on the pay for luck however is only signi-cant at the 10 percent level The economic signicance howeverstays large as a CEO with a tenure of nine years shows anincrease in pay for luck of 009 p 926 rsquo 31 percent but a rise inpay for performance of only 10 percent In columns (8) and (7) wesee that if anything pay for luck and pay for performance bothdiminish with tenure

While large shareholders correspond most closely to the ideaof a principal other governance measures could also be used Ourdata contain two variables that were shown to be importantgovernance measures in the past the size of the board and thefraction of board members who are insiders in the rm Smallboards are thought to be more effective at governing rms Yer-mack [1996] for example shows that smaller boards correlatewith larger q values for rms Insiders on the board are generally

28 We focus only on large shareholders on the board because these providedthe strongest results in Table IV We have used all large shareholders and foundsimilar although statistically weaker results


thought to weaken governance29 The rst four columns in TableVI estimate the effect of board size on pay for luck Columns (1) and(2) show that for accounting measures the direction of the effect isthe opposite of what we postulated but the coefcient is statisticallyinsignicant Note that the actual size of the coefcient is tiny evena huge increase in board size of ten board members leads only to a2 percent drop in pay for luck Columns (3) and (4) however showthat there are signicant effects for market measures of perfor-mance and these are of the expected sign Consider the differencebetween two boards one of which has ten board members and one ofwhich has six The big board rm shows a pay for performancecoefcient of 240 and a pay for luck coefcient of 229 The smallerboard continues to show a pay for performance coefcient of 228But it shows a pay for luck coefcient of 177 This represents a dropof (229 2 177)177 rsquo 30 percent in the pay for luck coefcient

The othermeasurewe examine is a measureof insider presenceonthe board This variable is measured as fraction of board members whoare rm insiders or gray directors Columns (5) and (6) show that onaccounting measures insider presence dramatically increases the payfor luck coefcient (signicant at the 10 percent level) In a board withten directors turning one of the outside directors from an outsider to aninsider increases pay for luck by 451 p 1227 rsquo 20 percent The effecton pay for performance is negative and small Columns (7) and (8) showthat on market performance measures insider board presence againincreases pay for luck but while the coefcient continues to be eco-nomically large it is statistically quite insignicant

We turn to our last governance measure in columns (9)through (12) where we construct an index that aggregates all thegovernance measures used so far number of large shareholdersnumber of large shareholders on the board board size and in-sider presence on board30 To form the index we demean each of

29 Empirical evidence on the effect of insiders on the board can be found inBaysinger and Butler [1985] Weisbach [1988] Rosenstein and Wyatt [1990]Hermalin and Weisbach [1991] Byrd and Hickman [1992] and Brickley Colesand Terry [1994] We have also examined CEO ownership and whether thefounder is present We do not report these for space reasons but both producegenerally signicant effects Founders and CEOs with high insider ownershipboth show greater pay for luck

30 Market valuation of a rm may provide another index of governance Wehave examined how a market-to-book measure correlates with the extent ofskimming on accounting returns We found using a procedure identical to the oneused for the governance variables that rms with higher market-to-book showedlower levels of pay for luck on accounting returns Clearly it would be conceptuallyawkward to do a similar exercise for market returns


the four governance variables divide it by its standard deviationand then take the sum of these standardized variables For boardsize we use negative of board size in this procedure For fraction

TABLE VICORPORATE GOVERNANCE AND PAY FOR LUCK (LUCK MEASURE IS MEAN INDUSTRY

PERFORMANCE) DEPENDENT VARIABLE ln (TOTAL COMPENSATION)

Governancemeasure Board size Fraction insiders

SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)


Governanc e 2 045 2 093 mdash mdash 2 482 451 mdash mdashIncomeassets (043) (094) (853) (269)Log (shareholder


(034) (210) (029) (215)Governanc e mdash mdash 002 013 mdash mdash 027 126Log (sh wealth) (002) (006) (05) (190)Governance 012 015 2 013 2 080 158 2 315 2 066 2 742


Governancemeasure Governance index

SpecicationGeneral

(9)Luck(10)

General(11)

Luck(12)

Income Assets 207 423 mdash mdash(210) (865)

Governanc e 007 2 216 mdash mdashIncomeassets (057) (134)Log (shareholder wealth) mdash mdash 249 252

(016) (232)Governanc e mdash mdash 2 003 2 033Log (sh wealth) (004) (015)Governance 2 016 000 010 210

(007) (011) (027) (103)Sample size 4610 4610 4551 4551Adjusted R2 695 705

a Dependent variable is the logarithm of total compensation All nominal variables are deated Eachregression includes rm xed effects year xed effects a quadratic in age and a quadratic in tenureStandard errors are in parentheses

b ldquoBoard sizerdquo indicates the number of members of the board of directors as listed in the proxy statementnear the start of the scal year ldquoFraction insidersrdquo is the fraction of inside and ldquograyrdquo directors on the boardof directors ldquoGovernance indexrdquo is the unweighted average of four standardized governance variables(number of large shareholders number of large shareholders on board minus board size and one minusfraction insiders)

c In all the luck regressions both the performance measure and the interaction of the performance measurewith the governancemeasure are instrumented The instruments are the asset-weighted average performance inthe two-digit industry and the interactions of the industry performance with that governance measure


of insiders on the board we use one minus that fraction Thisguarantees that the resulting governance index has larger valueswhenever the rm is better governed31

For accounting measures of performance (columns (9) and(10)) we again nd that pay for luck diminishes with the gover-nance while pay for performance does not change The coef-cient however is only signicant at the 10 percent level Togauge the magnitude of these effects consider a one-standard-deviation increase in the governance index about 2 Such anincrease leads to a 2 216 p 2423 rsquo 10 percent fall in the pay forluck coefcient When we use market measures (columns (11) and(12)) increases in the governance index greatly reduce pay forluck but hardly affect pay for performance In this case thecoefcients are signicant at the 5 percent level Moreover theirmagnitude is bigger A one-standard-deviation increase in gover-nance decreases pay for luck by 26 percent

Finally we investigate the robustness of our ndings Theprimary concern one might have is that we have not adequatelycontrolled for rm size One might worry that large rms havequite different pay for performance sensitivities than small rms[Baker and Hall 1999] If this also translates into different pay forluck sensitivities the estimates above might confuse this sizeeffect for a governance ldquoeffectrdquo

In Table VII we address this problem by controlling for sizeinteracted with performance We reestimate equation (3) but thistime include a term Size p perfit Our measure of size in theseregressions is average log real assets of the rm over the periodWe report the results for two governance measures large share-holders on the board and the governance index although we havereestimated all the previous tables with these controls and foundsimilar results Columns (1) through (4) are to be compared withcolumns (5) through (8) of Table IV We see that the effect ofgovernance on the ltering of accounting rates of return in factstrengthens when these controls are added ( 2 223 versus 2 148)

31 This particular way of proceeding will tend to count large shareholders onthe board twice once as on the board and once as general large shareholders Thisis a crude way of incorporating our prior belief (supported by the ndings in TableIV) that large shareholders on the board matter more When we use eithermeasure in the index alone we nd qualitatively similar results We have alsoestimated a regression in which we include all four governance measures (andtheir interactions) together These regressions showed all the governance mea-sures entering with the same sign and only the large shareholder variables beingstatistically signicant


The effects on market performance measures however weakenslightly (059 versus 076) Columns (5) through (8) use the gov-ernance index and are to be compared with columns (9) through(12) in Table VI This comparison shows that the results weakenvery slightly (2 197 versus 2 216) for accounting measures aswell as for market measures ( 2 027 versus 2 033) For bothgovernance measures however the results remain economicallysignicant In two of the cases (column (4) and to some extentcolumn (6)) they remain statistically insignicant In the othertwo they remain statistically signicant

We have also attempted other robustness checks We checkedwhether ltering may happen over longer time horizons by ag-gregating our data over several years as well as looking at lagsWe also allowed for interactions between performance and year in

TABLE VIICORPORATE GOVERNANCE AND PAY FOR LUCK ROBUSTNESS CHECKS

(LUCK MEASURE IS MEAN INDUSTRY PERFORMANCE)DEPENDENT VARIABLE ln (TOTAL COMPENSATION)

Governancemeasure Large shareholders on board Governance index

SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)

Income Assets 288 369 mdash mdash 2 570 123 mdash mdash(114) (146) (131) (617)


wealth)mdash mdash 223 2 136 mdash mdash 251 2 194

(086) (334) (098) (345)Governanc e mdash mdash 2 018 2 059 mdash mdash 2 003 2 027Log (shareho lder

wealth)(016) (053) (004) (012)

Governance 2 010 127 094 365 2 019 2 002 007 170(021) (038) (109) (357) (007) (010) (029) (085)

Firm Size p

PerformanceYes Yes Yes Yes Yes Yes Yes Yes

Sample size 4621 4621 4581 4581 4610 4610 4551 4551Adjusted R2 695 mdash 706 mdash 695 mdash 705 mdash

a Dependentvariable is the logarithm of total compensation Performance measure is operating incometo total assets All nominal variables are expressed in real dollars Each regression includes rm xed effectsyear xed effects a quadratic in age and a quadratic in tenure

b In all the luck regressions both the performancemeasure and the interaction of the performancemeasurewith the governancemeasure are instrumented The instruments are the asset-weighted average performance inthe two-digit industry and the interactions of the industry performance with that governance measure

c ldquoLarge shareholders on boardrdquo indicates the number of blocks of at least 5 percent of the rmrsquos commonshares that are held by directors of the board ldquoGovernanceindexrdquo is the unweightedaverage of four standardizedgovernance variables (number of large shareholders number of large shareholders on board minus board sizeand one minus fraction insiders) ldquoFirm sizerdquo indicates average log assets over the same period

d Standard errors are in parentheses


our regressions since there are known to be changes in the pay forperformance sensitivity over this time period These modica-tions did not alter our qualitative ndings

IV CONCLUSION

CEOs are rewarded for luck Moreover pay for luck is as largeas pay for general pay for performance Pay for luck also appears onthe most discretionary components of compensation salary andbonus Looking closer we found that pay for luck is strongest amongpoorly governed rms Adding a large shareholder on the board forexample decreased the pay for luck by 23 to 33 percent This ndingweakens two prominent explanations of pay for luck ldquoPaying forluck is optimalrdquo and ldquoFiltering out luck is impossiblerdquo

More broadly these results encourage a revision of our viewson CEO pay Poorly governed rms t the predictions of theskimming view Well-governed rms t the predictions of thecontracting view better They are to remove some luck in settingpay This suggests that both views hold some sway Other empiri-cal facts support this idea In Bertrand and Mullainathan [2000b]we showed that well-governed rms charged CEOs more for theoptions they were granted Options contain a gift componentbecause even if the CEO does nothing they have value from theintrinsic volatility of the stock (their Black-Scholes value) Weshow that rms with large shareholders smaller boards and soon are better able to charge their CEOs and better able to removethis gift component by reducing the other components of pay Inother words principal agent models work best when there are infact individuals around to act as principals

Several unanswered questions remain First it is unclearwhat effects the reward for luck has on overall CEO utility Doescompetition in the market for CEOs force the mean level of pay atinitial hire to adjust so that there are no ex ante rents to be hadOr is the hiring process sufciently closed or captured by insidersthat such adjustments are small Second while formal models ofthe contracting view abound there is no careful analysis of theskimming model Without such an analysis our understanding ofskimming will necessarily remain vague What are the exactmechanisms by which skimming is constrained How specicallydoes better governance translate into better pay packages Theresults in this paper suggest that more energy should be devotedto clarifying the skimming alternative


APPENDIX 1 SUMMARY STATISTICSmdash50 LARGEST U S OIL COMPANIES CEOS

Mean S D

Age of CEO 58562 7892Tenure of CEO 10181 9781Total compensation 608269 597194ln (Total Compensation) 6125 722

a Data set is 50 of the 51 largest U S oil companies over the period 1977ndash1994b Total Compensation is dened as the sum of salary and bonus (cash and stock bonus) company

contributions to thrift plans other annual income and the value of the options granted to the CEO duringthat year in thousands of 1977 dollars

APPENDIX 2 PAY FOR LUCK FIRST-STAGE REGRESSIONS

(LUCK MEASURE IS EXCHANGE RATE SHOCKS)A

Dep Var Income Inc to Assets Ln (Sh Wealth)

(1) (2) (3)

2 lt Appr lt 4 2 56588 2 006 2 039(Current) (26408) (004) (047)2 lt Appr lt 4 2 15428 004 2 027(Lagged) (24271) (004) (048)Appr gt 4 2 68903 2 013 2 034(Current) (32039) (005) (058)Appr gt 4 2 12045 006 053(Lagged) (30646) (005) (055)2 lt Depr lt 4 76642 2 000 153(Current) (24647) (004) (045)2 lt Depr lt 4 85858 010 114(Lagged) (25942) (004) (047)Depr gt 4 45482 007 094(Current) (27761) (005) (050)Depr gt 4 76345 017 046(Lagged) (29791) (005) (054)Exch rate index growth 2 19273 2 000 2 077(Current) (167134) (030) (302)Exch rate index growth 216140 038 237(Lagged) (175302) (031) (316)Sample size 1737 1729 1713Adjusted R2 622 700 873F-stat 348 26 247(prob gt F 2 stat) (000) (004) (006)

a Dependent variable is the level of income in column (1) the ratio of operating income to total assetsin column (2) and the log value of shareholder wealth in column (3) Income and shareholder wealth areexpressed in millions of 1977 dollars 2 lt Appr lt 4 is dummy variable that equals 1 if the industry-specic exchange rate index appreciated by more than 2 and less than 4 since the previous year All theother appreciation and depreciation dummies are dened in a similar way

b Each regression includes rm xed effects and year xed effects All regressions also include aquadratic in CEO age and a quadratic in CEO tenure

c The 3 regressions are the rst-stage regressions associated with columns (2) (4) and (8) in Panel A ofTable 3



PRINCETON UNIVERSITY NATIONAL BUREAU OF ECONOMIC RESEARCH AND CENTER

FOR ECONOMIC POLICY RESEARCH

MASSACHUSETTS INSTITUTE OF TECHNOLOGY AND NATIONAL BUREAU OF ECONOMIC

RESEARCH

REFERENCES

Abowd John M and David S Kaplan ldquoExecutive CompensationSix Questions ThatNeed Answeringrdquo Journal of Economic Perspectives XIII (1999) 145ndash168

Aggarwal Rajesh K and Andrew A Samwick ldquoThe Other Side of the Trade-OffThe Impact of Risk on Executive Compensationrdquo Journal of Political Econ-omy CVII (1999a) 65ndash105

Aggarwal Rajesh K and Andrew A Samwick ldquoExecutive Compensation Stra-tegic Competition and Relative Performance Evaluation Theory and Evi-dencerdquo Journal of Finance LIV (1999b) 1999ndash2043

Baker George Robert Gibbons and Kevin J Murphy ldquoSubjective PerformanceMeasures in Optimal Incentive Contractsrdquo Quarterly Journal of EconomicsCIX (1994) 1125ndash1156

Baker George and Brian Hall ldquoUnderstanding Top Management IncentivesFirm Size Risk and CEO Effortrdquo Harvard Business School 1998

Baysinger Robert D and Henry N Butler ldquoCorporate Governance and the Boardof Directors Performance Effects of Changes in Board Compositionrdquo Journalof Law Economics and Organization I (1985) 101ndash124

Bertrand Marianne ldquoFrom the Invisible Handshake to the Invisible Hand HowImport Competition Changes the Employment Relationshiprdquo National Bu-reau of Economic Research Working Paper No 6900 1999

Bertrand Marianne and Sendhil Mullainathan ldquoCorporate Governance and Ex-ecutive Compensation Evidence from Takeover Legislationrdquo Princeton Uni-versity 1999

Bertrand Marianne and Sendhil Mullainathan ldquoAgents with and without Prin-cipalsrdquo American Economic Review XC (2000a) 203ndash208

Bertrand Marianne and Sendhil Mullainathan ldquoDo CEOs Set Their Own PayThe Ones without Principals Dordquo National Bureau of Economic ResearchWorking Paper No 7604 2000b

Blanchard Olivier J Florencio Lopez-de-Silanes and Andrei Shleifer ldquoWhat DoFirms Do with Cash Windfallsrdquo Journal of Financial Economics XXXVI(1994) 337ndash360

Brickley James A Jeffrey L Coles and Rory L Terry ldquoOutside Directors and theAdoption of Poison Pillsrdquo Journal of Financial Economics XXXV (1994) 371ndash390

Byrd John W and Kent A Hickman ldquoDo Outside Directors Monitor ManagersrdquoJournal of Financial Economics XXXII (1992) 195ndash221

Crystal Graef In Search of Excess The Overcompensation of American Executives(New York NY WW Norton Co 1991)

Denis David and Jan Serrano ldquoActive Investors and Management TurnoverFollowing Unsuccessful Control Contestsrdquo Journal of Financial EconomicsXL (1996) 239ndash266

Diamond Peter ldquoManagerial Incentives On the Near Linearity of Optimal Com-pensationrdquo Journal of Political Economy CVI (1998) 931ndash957

Durell Alan ldquoAttribution in Performance Evaluationrdquo Dartmouth College 1999Garen John E ldquoExecutive Compensation and Principal-Agent Theoryrdquo Journal

of Political Economy CII (1994) 1175ndash1199Gibbons Robert and Kevin J Murphy ldquoRelative Performance Evaluation for

Chief Executive Ofcersrdquo Industrial and Labor Relations Review XLIII(1990) S30ndashS51

Gibbons Robert and Kevin J Murphy ldquoOptimal Incentives in the Presence ofCareer Concerns Theory and Evidencerdquo Journal of Political Economy C(1992) 468ndash505

Haid Michael Incentive Compensation and the Market for Corporate ControlSubstitutive Forces to Discipline Management of Publicly Held Organizationsin the U S Empirical Evidence From the Oil Industry 1977ndash1994 (BernSwitzerland Haupt 1997)


Hall Brian J ldquoThe Design of Multi-Year Stock Option Plansrdquo Journal of AppliedCorporate Finance XII (1999) 97ndash106

Hall Brian J and Jeffrey B Liebman ldquoAre CEOs Really Paid Like BureaucratsrdquoQuarterly Journal of Economics CXIII (1998) 653ndash691

Hermalin Benjamin and Michael Weisbach ldquoThe Effects of Board Compositionand Direct Incentives on Firm Performancerdquo Financial Management XX(1991) 101ndash112

Himmelberg Charles and R Glenn Hubbard ldquoIncentive Pay and the Market forCEOs An Analysis of Pay-For-Performance Sensitivityrdquo Columbia Univer-sity 2000

Holmstrom Bengt ldquoMoral Hazard and Observabilityrdquo Bell Journal of EconomicsX (1979) 74ndash91

Holmstrom Bengt and Paul Milgrom ldquoAggregation and Linearity in the Provi-sion of Intertemporal Incentivesrdquo Econometrica LV (1987) 303ndash328

Hubbard R Glenn and Darius Palia ldquoExecutive Pay and Performance Evidencefrom the U S Banking Industryrdquo Journal of Financial Economics XXXIX(1994) 105ndash130

Janakiraman Surya N Richard A Lambert and David F Larcker ldquoAn Em-pirical Investigation of the Relative Performance Evaluation HypothesisrdquoJournal of Accounting Research XXX (1992) 53ndash69

Jensen Michael and Kevin J Murphy ldquoPerformance Pay and Top-ManagementIncentivesrdquo Journal of Political Economy XCVIII (1990) 225ndash264

Joskow Paul L Nancy L Rose and Catherine D Wolfram ldquoPolitical Constraintson Executive Compensation Evidence from the Electric Utility IndustryrdquoRand Journal of Economics XXVII (1996) 165ndash182

mdashmdash ldquoCorporate Performance and Managerial Remuneration An Empirical In-vestigation of Managerial Labor Contractsrdquo Rand Journal of EconomicsXVII (1986) 59ndash76

mdashmdash ldquoExecutive Compensationrdquo Handbook of Labor Economics O Ashenfelterand D Card eds (Amsterdam The Netherlands North-Holland 1999)

Revenga Ana L ldquoExporting Jobs The Impact of Import Competition on Employ-ment and Wages in U S Manufacturingrdquo Quarterly Journal of EconomicsCVII (1992) 255ndash284

Rosenstein Stuart and Jeffrey G Wyatt ldquoOutside Directors Board Indepen-dence and Shareholder Wealthrdquo Journal of Financial Economics XXVI(1990) 175ndash191

Shea John ldquoNominal Illusion Evidence from Major League Baseballrdquo Universityof Maryland 1999

Shivdasani Anil ldquoBoard Composition Ownership Structure and Hostile Take-oversrdquo Journal of Accounting and Economics XVI (1993) 167ndash198

Shleifer Andrei and Robert Vishny ldquoLarge Shareholders and Corporate ControlrdquoJournal of Political Economy XCIV (1986) 461ndash488

Tufano Peter ldquoWho Manages Risk An Empirical Examination of Risk Manage-ment Practices in the Gold Mining Industryrdquo Journal of Finance LI (1996)1097ndash1137

Weisbach Michael ldquoOutside Directors and CEO Turnoverrdquo Journal of FinancialEconomics XX (1988) 431ndash460

Yermack David ldquoDo Corporations Award Stock Options Effectivelyrdquo Journal ofFinancial Economics XXXIX (1995) 237ndash269

mdashmdash ldquoHigher Market Valuation of Companies with a Small Board of DirectorsrdquoJournal of Financial Economics XL (1996) 185ndash211

Zeckhauser Richard J and John Pound ldquoAre Large Shareholders EffectiveMonitors An Investigation of Share Ownership and Corporate PerformancerdquoAsymmetric Information Corporate Finance and Investment R Glenn Hub-bard ed (Chicago IL University of Chicago Press 1990)


luck therefore cannot provide better incentives and will onlymake the contract riskier [Holmstrom 1979]2

This paper starts by examining whether or not CEOs are infact paid for luck using three measures of luck3 First we performa case study of the oil industry where large movements in oilprices tend to affect rm performance on a regular basis Secondwe use changes in industry-specic exchange rate for rms in thetraded goods sector Third we use year-to-year differences inmean industry performance to proxy for the overall economicfortune of a sector4 For all three measures we nd that CEO payresponds signicantly to luck In fact we nd that CEO pay is assensitive to a lucky dollar as to a general dollar Moreover theseresults hold as well for discretionary components of paymdashsalaryand bonusmdashas they do for options grants

These results are inconsistent with a simple contractingview Motivated by practitioners such as Crystal [1991] we pro-pose an alternative skimming which can explain these results[Bertrand and Mullainathan 2000a] The skimming view alsobegins with the separation of ownership and control but it arguesthat this separation allows CEOs to gain effective control of thepay-setting process itself Both because of entrenchment such aspacking the board with supporters and because of the complexityof the pay process many CEOs de facto set their own pay withlittle oversight from shareholders Their pay level then becomesconstrained by an unwillingness to draw shareholdersrsquo attentionPay for performance arises in the skimming view because goodperformance may ease these constraints in essence creating

2 Note our emphasis on observable luck In any model given the random-ness of the world CEOs (and almost everybody else) will end up being rewardedfor unobservable luck Note also our emphasis on the fact that this predictionholds in simple agency models As we will discuss shortly complications to theagency model can in principle alter this result

3 Blanchard Lopez-de-Silanes and Shleifer [1994] present suggestive evi-dence on pay for luck by showing that windfall gains from court rulings raise thepay of CEOs It is only suggestive since court rulings may not be luck but rathera result of the CEOrsquos work In other domains Shea [1999] independently performsan exercise similar to ours for baseball players

4 This last test very much resembles the approach followed in the relativeperformance evaluation (RPE) literature [Gibbons and Murphy 1990 Janakira-man Lambert and Larcker 1992 Aggarwal and Samwick 1999a] Problems canarise with RPE as a special case of luck Filtering this specic kind of luck may notbe optimal from an agency theoretical point of view As Gibbons and Murphy[1990] note relative performance evaluation can distort CEO incentives if theycan ldquotake actions that affect the average output of the reference grouprdquo Aggarwaland Samwick [1999b] develop a formal model along these lines By using othershocks to performance that are even more objectively beyond managerial inu-ence we circumvent these problems












(1) s 5 a 1 b ( p 2 d o) 5 a 1 b (a 1 u)







































82 215 mdash mdash(16) (104)


mdash mdash 38 35(03) (17)

Age 05 07 05 05(02) (03) (02) (02)

Age2p 100 2 04 2 05 2 04 2 04

(02) (02) (02) (02)Tenure 01 01 01 01

(01) (01) (01) (01)Tenure2

p 100 2 03 2 03 2 03 2 03(02) (01) (02) (02)























Mean S D










ln (totalcomp)

































































SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(007) (036) (016) (053)

Governance 2 009 018 2 017 411 2 006 084 100 480(011) (018) (049) (240) (021) (033) (108) (356)
























No Yes No Yes

SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)




























SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)







SpecicationGeneral

(9)Luck(10)

General(11)

Luck(12)




















SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(016) (053) (004) (012)

Governance 2 010 127 094 365 2 019 2 002 007 170(021) (038) (109) (357) (007) (010) (029) (085)

Firm Size p









IV CONCLUSION






Mean S D







(1) (2) (3)










RESEARCH

REFERENCES
























































(1) s 5 a 1 b ( p 2 d o) 5 a 1 b (a 1 u)







































82 215 mdash mdash(16) (104)


mdash mdash 38 35(03) (17)

Age 05 07 05 05(02) (03) (02) (02)

Age2p 100 2 04 2 05 2 04 2 04

(02) (02) (02) (02)Tenure 01 01 01 01

(01) (01) (01) (01)Tenure2

p 100 2 03 2 03 2 03 2 03(02) (01) (02) (02)























Mean S D










ln (totalcomp)

































































SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(007) (036) (016) (053)

Governance 2 009 018 2 017 411 2 006 084 100 480(011) (018) (049) (240) (021) (033) (108) (356)
























No Yes No Yes

SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)




























SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)







SpecicationGeneral

(9)Luck(10)

General(11)

Luck(12)




















SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(016) (053) (004) (012)

Governance 2 010 127 094 365 2 019 2 002 007 170(021) (038) (109) (357) (007) (010) (029) (085)

Firm Size p









IV CONCLUSION






Mean S D







(1) (2) (3)










RESEARCH

REFERENCES

















































































82 215 mdash mdash(16) (104)


mdash mdash 38 35(03) (17)

Age 05 07 05 05(02) (03) (02) (02)

Age2p 100 2 04 2 05 2 04 2 04

(02) (02) (02) (02)Tenure 01 01 01 01

(01) (01) (01) (01)Tenure2

p 100 2 03 2 03 2 03 2 03(02) (01) (02) (02)























Mean S D










ln (totalcomp)

































































SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(007) (036) (016) (053)

Governance 2 009 018 2 017 411 2 006 084 100 480(011) (018) (049) (240) (021) (033) (108) (356)
























No Yes No Yes

SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)




























SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)







SpecicationGeneral

(9)Luck(10)

General(11)

Luck(12)




















SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(016) (053) (004) (012)

Governance 2 010 127 094 365 2 019 2 002 007 170(021) (038) (109) (357) (007) (010) (029) (085)

Firm Size p









IV CONCLUSION






Mean S D







(1) (2) (3)










RESEARCH

REFERENCES































































Mean S D










ln (totalcomp)

































































SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(007) (036) (016) (053)

Governance 2 009 018 2 017 411 2 006 084 100 480(011) (018) (049) (240) (021) (033) (108) (356)
























No Yes No Yes

SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)




























SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)







SpecicationGeneral

(9)Luck(10)

General(11)

Luck(12)




















SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(016) (053) (004) (012)

Governance 2 010 127 094 365 2 019 2 002 007 170(021) (038) (109) (357) (007) (010) (029) (085)

Firm Size p









IV CONCLUSION






Mean S D







(1) (2) (3)










RESEARCH

REFERENCES


















































ln (totalcomp)

































































SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(007) (036) (016) (053)

Governance 2 009 018 2 017 411 2 006 084 100 480(011) (018) (049) (240) (021) (033) (108) (356)
























No Yes No Yes

SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)




























SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)







SpecicationGeneral

(9)Luck(10)

General(11)

Luck(12)




















SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(016) (053) (004) (012)

Governance 2 010 127 094 365 2 019 2 002 007 170(021) (038) (109) (357) (007) (010) (029) (085)

Firm Size p









IV CONCLUSION






Mean S D







(1) (2) (3)










RESEARCH

REFERENCES



























































































SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(007) (036) (016) (053)

Governance 2 009 018 2 017 411 2 006 084 100 480(011) (018) (049) (240) (021) (033) (108) (356)
























No Yes No Yes

SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)




























SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)







SpecicationGeneral

(9)Luck(10)

General(11)

Luck(12)




















SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(016) (053) (004) (012)

Governance 2 010 127 094 365 2 019 2 002 007 170(021) (038) (109) (357) (007) (010) (029) (085)

Firm Size p









IV CONCLUSION






Mean S D







(1) (2) (3)










RESEARCH

REFERENCES






























































No Yes No Yes

SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)




























SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)







SpecicationGeneral

(9)Luck(10)

General(11)

Luck(12)




















SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(016) (053) (004) (012)

Governance 2 010 127 094 365 2 019 2 002 007 170(021) (038) (109) (357) (007) (010) (029) (085)

Firm Size p









IV CONCLUSION






Mean S D







(1) (2) (3)










RESEARCH

REFERENCES






























































SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)







SpecicationGeneral

(9)Luck(10)

General(11)

Luck(12)




















SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(016) (053) (004) (012)

Governance 2 010 127 094 365 2 019 2 002 007 170(021) (038) (109) (357) (007) (010) (029) (085)

Firm Size p









IV CONCLUSION






Mean S D







(1) (2) (3)










RESEARCH

REFERENCES

























































SpecicationGeneral

(1)Luck(2)

General(3)

Luck(4)

General(5)

Luck(6)

General(7)

Luck(8)





wealth)(016) (053) (004) (012)

Governance 2 010 127 094 365 2 019 2 002 007 170(021) (038) (109) (357) (007) (010) (029) (085)

Firm Size p









IV CONCLUSION






Mean S D







(1) (2) (3)










RESEARCH

REFERENCES















































IV CONCLUSION






Mean S D







(1) (2) (3)










RESEARCH

REFERENCES















































Mean S D







(1) (2) (3)










RESEARCH

REFERENCES

















































RESEARCH

REFERENCES














































are ceos rewarded for luck? the ones without...

Documents