ESSAYS O N EXECUTIVE COMPENSATION AND MANAGERIAL INCENTIVES

Xianming Zhou

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Economics University of Toronto

o Copyright by Xianming Zhou 1997

1+1

.

National Library ,nada

Bibliothéque nationale du Canada

Acquisitions and Bibliographie Services

Acquisitions et services bibliographiques

395 Wellington Street Ottawa ON K I A ON4 Canada

395. nie Wellington Ottawa ON KtA ON4 Canada

The author has granted a nonexclusive licence allowing the National Library of Canada to reproduce, loan, distribute or sell copies of this thesis in microform, paper or electronic fomats.

L'auteur a accordé une licence non exclusive permettant à la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme de microfichelfiIm, de reproduction sur papier ou sur format électronique.

The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial extracts fiom it d othenvise may be p ~ t e or reproduced without the author's permission.

L'auteur conserve la propriété du droit d'auteur qui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent être imprimés ou autrement reproduits sans son autorisation.

Essays on Executive Compensation and Managerial Incentives Xianming Zhou Doctor of Philosophy 1997 Graduate Department of Economics University of Toronto

Abstract

The first chapter of this dissertation examines the regdatory impact of executive compensation disclosure on managerial pay schemes. Analysing a three-level agency mode1 arnong the shareholders, directors, and manager of a firm, this chapter shows that before disclosure, the directors may underinvest in monitoring the manager's activities because of asymmetric information on the penomance of the directors and a divergence of interests between them and other shareholders. Compensation disclosure rules mitigate this ineficiency and increase the pay-performance linkage. Empincaily, this chapter examines the impact of compensation disclosure by comparing the sensitivity of C E 0 pay to finn performance, before and after the new Ontario disclosure regulation (1993), between groups of firms that were differentidly afEected by the legislation. It is found that after introducing the Ontario regulation, the pay-performance sensitivity increased much more in the firms that were affected by the

regulation than in those that were unaffected by the regulation. In contrast to previous studies, these findings suggest that disclosure of executive compensation strengthens the pay-performance

..

11

relationship. The second chapter provides the first systernatic examination of the relationship between

CE0 compensation. fimi size, and corporate performance for Canadian f m . Consistent with previous studies, t h i s chapter fmds that C E 0 pay rises with fm size and that compensation is tied to Company performance. Some novel fmdings are also documented. First, options and stock ownership tend to play a relatively more important role for CE0 incentives in small firms. And second, while C E 0 turnover probability is generally negatively related to the firm's stock

performance, the threat of dismissal is less pronounced in small firms. The third chapter compares executive pay-performance sensitivities between Canadian

firms and US firms. Examining the data for 365 Canadian h

s and 678 US f m s over the years

1991- 1994, it is found that the pay-performance sensitivity associated with direct pay and stock

ownership is smailer in Canadian fimis than in US firms but that the difference diminishes as firm size increases. This finding, together with poorer corporate performance and lower C E 0 pay of Canadian firms, is consistent with the hypothesis that managerial incentives affect firm

Acknowledgements

1 would like to thank Dwayne Benjamin, Morley Gunderson, Michael Berkowitz, and

Michael Baker for their comments and encouragement. Specid thanks go to Aloysius Siow who paid close attention to my progress, spent countless hours discussing various issues with me and

provided many valuable suggestions. I am especially grateful to my supervisor. Arthur Hosios, for his guidance and inspiration, his constructive criticism and insighdul suggestions at every

stage of my research, and for his effort in the many revisions of this dissertation. Finally, 1 want to express my gratitude to my farnily to whom 1 dedicate this work.

Table of Contents

Chapter 1 Executive Compensation Disclosure and Management Incentives 1.t Introduction ..................................................................................................................... 1 1.2 The Issue and the Literature ............... ...................................................

.................. 5 1.3 The Mode1 ..................................................................................................................... 10 1.3.1 The Basic Mode1 ..................................... ,.,................................................... 10 1.3.2 Political Pressures ......................................................................................... 12 1.3.3 Pay Envy ....................................................................................................... 14 1.3.4 Shareholden' Scmtiny ................................................................................... 15 1.3.5 A Surnmary of Disclosure Effects .............................................................. 25 .. 26 1-4 Evidence for Chief Executive OFficers ................................................................ 1.4.1 The Ontario Regulation Act ........................................................................... 26 1.4.2 The Methodology and the Data .................................... ............................. 28 1.4.3 Empiricd Results .......................................................................................... 32 1.5 Conclusion .................................................................................................................. 38 Tables and Figures ..................................... ........................................................................ 40 Appendices ......................................................................................................................... 44 References .......................................................................................................................... 52

Chapter 2 C E 0 Pay. Firm Size. and Corporate Performance: Evidence From Canada 2.1 Introduction ................................................................................................................. 56

...................................................................................................................... 58 ....................................................................... 62 2.3 C E 0 Pay and Firm Size .................... . 2.4 C E 0 Pay and Corporate Performance ........................................................................ 65 2.4.1 Pay-Performance Sensitivity ....................................................................... 67 2.4.2 C E 0 Stock Ownership .............................................................................. 70 2.2 The Data

2.4.3 C E 0 Turnover and Corporate Performance .............................................

71

.................................................................................................................... 75 Tables and Figures .............................................................................................................. 76 Appendices ......................................................................................................................... 87 References ........................................................................................................................... 93 2.5 Conclusion

Chapter 3 Executive Compensation and Managerial Incentives: A Cornparison

Between Canada and the United States 3.1 Introduction ................................................................................................................

96

.......................... .................................................... 98 . 3.2.1 Previous Studies ............................. . . . ................................................. 98 3.2.2 Mode1 Specification ................................................................................... 100 3.2.3 The Data ................................................................................................... 102 3-3 Incentive Strength ..................................................................................................... 105 3.3.1 Incentives fiom Direct Payments ............................................................... 105 3.3 -2 Incentives fiom Stock Ownership ..............................................................110 3.3.3 nie Effect of Stock Options ....................................................................... 113 3.4 Incentives and Performance ................................................................................. 114 3.5 Conclusion ................................................................................................................ 1 17 Tables and Figures .......................................................................................................... 118 References ....................................................................................................................... 129 3.2 The Methodology and the Data

List of Tables

Chapter 1

............................................................................................... 40 Table 1.2 Changes in Pay-Performance Sensitivity (Canadian fimis) .................................. 41 Table 1.1 Selected Statistics

Table 1.3 Changes in Pay-Performance Sensitivity: Compustat Fims vs. non-Compustat fims .............................................................................. 42 Table 1.4 Changes in Pay-Performance Sensitivity: US firms vs . Compustat and non-Compustat Canadian Firms .................................................................. 43

Chapter 2

.................................................................. 76 Nurnber of Firms with Different Incentive Plans ................................................ 76 Surnmary of Selected Statistics ........................................................................... 77 Firm Size Elasticity of CE0 Total Pay ................................................................ 78 Firm Size Elasticity of CE0 Relative Pay ............................................................ 79 Pay-Performance Sensitivity (OLS estimation) .................................................. 80 C E 0 Stock Ownership ........................................................................................ 81

Table 2.1 A Bief Comparison of the Evidence Table 2.2 Table 2.3 Table 7.4 Table 2.5 Table 2.6 Table 2.7

Table 2.8 C E 0 Turnover and Firm Performance (Probit estimation) ................................. 81 Table 2.9 Probability of C E 0 Turnover .............................................................................. 82

Chapter 3

............................................................................................... 118 Table 3.2 Estimation of Pay-Performance Sensitivity (total sample) ................................. 119 Table 3.3 Selected Statistics about Sample Heterogeneity ................................................. 120 Table 3.4 Estimation of Pay-Performance Sensitivity (subsample 1 and 2) ....................... 121

Table 3.1 Selected Statistics

vii

Table 3-5 Estimation of Pay-Performance Sensitivity (subsarnple 1) ............................. .. 122 Table 3.6 Estimation of Pay-Performance Sensitivity (subsarnple 2) ..............................

123

Table 3.7 The US-Canada Differential and Firm Size (total sample) ................................. 124 Table 3.8 CE0 Stock Ownership as the Percentage of Shares Outstanding ...................... 125 Table 3.9 Regressions of C E 0 Stock Ownenhip Against F i m Size ................................. 125 Table 3.10 The Ratio of Stock Gains from Exercising Previously Granted Options Over Total C E 0 Pay ........................................................................... 126

............... 126 Table 3.1 1 Cornparison of Fims' Systematic Performance ......................... . . Table 3.12 Cornparison of the C E 0 Pay Level .................................................................

viii

127

List of Figures

Chapter 2

Figure 2.1 Distribution of Firm Size ................................................................................. 83

....................................................................................... 84 Figure 2.3 Components of C E 0 Pay ....................................................................................... 85 Figure 2.4 Structure of CE0 Pay .......................................................................................... 86 Figure 2.2 Distribution of CE0 Pay

Chapter 3 Figure 3.1 Distribution of Firrns .....................................................................................

128

List of Appendices

Chapter 1 A . Proof of Proposition 1 .......................................................................................................

44

B . Proof of Proposition 2 ...............................................................................-................... 45

C. Proof of Proposition 3 ................... ....

............................................................-............... 46 D. Proof of Proposition 4 ...................................................................................................... 47 E. Proof of Proposition 5 ..................................................................................................... 48 F . The Integrated Post-Disclosure Model ............................................................................. 49

Chapter 2 List of Finns .........................................................................................................................

87

Chapter 1 Execut ive Compensation Disclosure and Management Incent ives 1.1 Introduction Since the Seciuity Exchange Commission of the United States enacted its first regdation regarding execiitive compensation disclosure more than half a centiiry ago, execiitive compensation has becorne a controversial corporate govemance topic. In recent years, the level and structure of executive compensation have attracted considerable academic

and public interest. C E 0 compensation, in particidar, has fascinated many scholars. and in one form or another it has been studied by economists. sociologists and experts in management and industrial relations. The debate on executive pay seems to have been part of the national agenda in the United States and becarne a widely disciissed popiilist

topic diiring the 1992 election year. There are two major issues in the debate. One concerns the level of execiitive pay

which inqtures whether or not executives earn "excessive" compensation. The other concerns the pay structure, fociising on the relationship between execiitive pay and corporate performance. On one side of the debate are critics who argue that execiitive pay is tuirelateci to firm performance and is excessive.' On the other side of the debate are

those who believe some execiitives may, in fact, be underpaid becaiw of the inherent iinfairness of arbitrarily Limiting an individual's wort h under public pressiires.2 m i l e 'A representative view of excessive executive pay is presented by Graef Crystal. In his book of In Seamh of Ezcess (1991), Crystal has siibstantiated the Ievel of overcompensation and has describeci the skewed procedure for setting compensation packages. 'See, e-g., the comment of Whitworth (1989).

1.1 INTRODUCTION

2

the viewpoints in the debate &en contrast sharply with one another. there seems to be a consensus that the Linkage betaieen executive pay and corporate performance is Explanations of a weak pay-performance linkage differ dramatically. Some critics

link it with excessive compensation and attribiite it to corporate dysfimction. They argue that executives siiccessfully lise an increase in company performance to jitsti@ large pay raises while poor performance still resirlts in siibstantial pay. According to t h i s populist view: there exists a fiindamental problem in corporate govemance

-

the board

of directors is not making compensation decisions in the best interests of the company and its shareholders. This view contends that regdatory measlires regarding executive compensation are both necessary and desirable to ensiire order- corporate operation and protec t the interests of the individual investors. Some economists. however. argue that

a lack of pay-performance iinkage may actiially result from govemment disclosine niles. It is the visibility of high executive pay that has caiised public concem. fiieled criticism.

and tiuned the debate into an emotional issue. Piiblic disapproval of high execiitive pa? fosters political pressires which constrain the types of contracts that are m i t ten berneen

managers and firms. With the market failing to freely set compensation. the pay system becomes less efficient than it shotdd be-

These two viem depict quite Werent perspectives concerning the impact of goyeniment disclosiire ndes on evecutive compensation. Thoiigh it is both interesting and important to i d e n t e the effects of disclosiire. few stiidies have directly examined this issue. The reason is the la& of information. Cjntil the early 1990s. the regdations re-

garding this issue were the US disclosure d e s implemented in the 1930s and the 1967 Companieç Act in Bntain.'

Most stiidies have fociised on 5s execiitive compensation

3Jensen and S f i q h y (1990a)give a bench mark estirnate of the intensity of the pay-performance relationship. Examining the Forber sample of CE0 compensation during the years 1974-1986. they find that CE0 compensation increases o d y S3 for each S10ûû increase in shareholder value. 'Japanese regdations do not require firrns to disclose execiitive compensation. The stiidies for

1.1 INTROD UCTION

3

perhaps becaise it contains the rnost detailed information concernuig the remmeration

of executived Since the US regulation has existed for more than half a century, it is difficult to document its i d u e n c e because of both insufEicient data and siibstantial changes in the institutional and market environment diiring siich a long period of t h e .

In October 1993, the province of Ontario enacted the first Canadian regdation regarding compensation disclosure of top exec~tives.~ All publicly-traded b

s in Ontario

were reqtùred, beginning in fiscal year 1993, to disclose the compensation of their five highest paid execiitives including the CE0 for the ciment fiscal year and the two immediate preceding years. Because a majority of large Canadian firw are traded on the Toronto Stock Exchange,' the regdation affects most large publicly-held Canadian corporations. The implementation of this new legislation provides a natiud experiment in which to evaluate the regdatory impact of disclosiue. Using the residting C E 0 compensation data, this paper condiicts a stiidy of the effects of disclosiire on the firms' managerial pay schemes nith an emphasis on the impact on the pay-performance relationship. In the public debate and law and econornics literatiue there have been m e r e n t informal arguments made regarding the impact of disclosiue. In this chapter 1 first develop a mode1 to analyze the issue. It extends a conventional owner-manager relationship to

a shareholder-direct or-manager relationship. 1 show t hat , before disclosiire, the board of directors may underinvest in monitoring the manager's activities because of asyrn-

metric information on the performance of the directors and a divergence of interests between thern and other shareholders. In addition to generating inefficient managerial Japanese h s either use the total salary and bonus earned by al1 directors as a proxy of execirtive compensation (Kaplan, 1994) or use tax returm to estimate the compensation of Japanese execiitives (Kato and Rockel, 1992). 'The British Act required firrns to show only the aggregate ernoltunents of the Chairman and t h e highest-paid director when other than the Chairman (see Cosh, 1975). 'Ontario Regdation 638,filed on October 14, 1993. 7For example, by either total assets or sales in fiscal year 1993, more than 90 percent of the 300 largest Canadian firms are traded on the Toronto Stock Exchange.

1.1 INTRODUCTION

incentive schemes, the model shows that the manager may also "bribe" the board of directors and obtain rents. Pay disclosine ndes mitigate this inefficiency and increase

the pay-performance linkage. For cornparison, 1 also consider a simple model of the effect of polit ical pressures on managerial pay schemes (the Jensen-h.Iiirphy hypothesis).

This model shows that public disapproval of high payments, when it imposes a cost on the firm, weakens the pay-performance linkage. Additionally, 1 show that pay envy or jealousy among managers enhances incentives and strengthem the pay- p erforxnance relationship. The total regdatory impact of disclosure on manageriai incentives is thiis theoreticdy arnbigiioiis. 1 then proceed to examine the evidence concerning disclosiue. Exarnining C E 0 com-

pensation of 384 large publicly-held Canadian firms, 1 find that the sensitivity of C E 0 pay to firm performance (the ratio of changes in C E 0 pay over changes in shareholder

wealth) increased significantly after the new regdation. To control for the effects of other potential factors, 1 also examine the evidence by a "difference in differences" analysis.

This is done by comparing pay-performance sensitivity, before and after the regdation. between Canadian firms whose shares were piiblicly traded in the United States before

1991 (and were siibject to US disclosiire regtdations) and those whose shares were not, and between Canadian firm and US b s . 1 End that after the Ontario regulation, (i) the pay-performance sensitivity increased siibstantidy more for Canadian firms whose shares traded excliisively in Canada than for Canadian firms that also traded in the

United States, and (ii), whde the sensitivity increased less in US firms than in Canadian h

s whose shares traded excliisively in Canada, there is Little difference between US

firms and other Canadian

that also traded in the United States. In contrast to pre-

vious stuclies, these findings siiggest that executive compensation disclosure regtdations strengthen the pay-performance retationship.

1.2

THE ISSUE AND THE LITERATURE

5

This chapter proceeds as follows. Section 1.2 introduces the issue and briefly reviews some related st~idies. Section 1.3 develops a model to characterize regdatory effets of g o v e r n e n t disclosure d e s on managerial pay contracts. Section 1.4 disciisses the evidence. The conclusion is provideci in Section 1.5.

1.2 The Issue and the Literature An incent ive problem arises when the individiial wit h ownership rights, the principal. delegates decision-making or prodiiction tasks to anot her individiial, the agent, and where the principal cannot observe the agent's effort. According to standard agency theory? the reward scheme must be a fimction of the agent's performance t o provide incentives for the agent to work in the interest of the principal. Agency costs imavoidably occiu in stich an environment due to the CO-existenceof privately observed effort and ris k-averse agents, and so the pay scheme m w t be constructed to strike a balance between incentives and insurance, The relationship between a h ' s shareholders and its CE0 has been modeled as a principal-agent relationship and extensively analyzed in the economics literatiue. In this context, the compensation scheme is designeci so that the pay-performance Linkage reflects the incentive-insimance trade off and, with a cornpetitive managerial labo1ir market the

base s a l a q adjiists so that total expected compensation eqtials the CEO's expected marginal prodiict. This model suggests that piiblic concern with C E 0 pay is a piizzle. The problem, however, is that the relationship between shweholders and managers

in the real world does not present a simple agency problem. It is often argued that the compensation system is not working in the best interests of the shareholders becaiise the pay scheme is not controlled directly by the shareholders but by the board of directors. Commenting on the disciplining role of corporate democracy, Car1 (1993, p8) notes,

1.2 THE ISSUE -4ND T H E LITERATURE

'In theory, directors set an executive's compensation and, in accordance with their fiduciary duties to the shareholders, compensate execiitives in the manner most advantageous to the Company and its stockholders. if the directors fail to do so, the shareholders shoidd, in theory, elect new directors.' This raises the key issue: do directors have the incentive to perform their fidiiciary diities in the best interests of the shareholders? Doiibts exist. Becaiise directors and execiitives are often b o w d by reciprocal self-interest , it may not be in the best interests of directors

for them to be honest with shareholders. It is widely believed that directors are captive.

"The independent director has tunied out to exist only in theory; in the real world al1 direct ors are beholden to management" (Cari, 1993, p34) .8 Withoiit information aboiit how managers are rewarded, it is difficult for shareholders to know how the directors perform their diities in monitoring and motivating the management. Becaiise the market may, in consequence, fail to implement an effective compensation system, governrnent intervention that makes execiitive compensation public may be needed. Governrnent disclosure d e s me a direct regdatory response to shareholder concern aboiit potential 'looting' by managers in colliision with directors. As execiitive pay is exposed to the public, both insiders (shareholders and managers) and outsiders respond accordingly. Varioits informa1 arguments have been put forth to describe the impact of disclosure. The following are some often mentioned influential factors. (1) Shareholders' scmtiny of pay schemes With disclosiire, the informational asymmetry concerning managerial pay between shareholders and the board of directors is removed; shareholders becorne aware of how ' ~ o k(1993) condiided: "Most top execritives are in an ~musiiailystrong position to strike a favorable bargain, because t hey exert such idtience over the process (fixing exectit ive compensation). CEOs almost always serve as chairman of the board. They typically have a good deal to say aboiit the choice of new board members. They are the key people who decide on the fees paid to the directors, fees that average over $40,000 for only a few days of work each year."

1.2 THE ISSUE AND THE LITERATURE

7

much and, to some extent, how the management is paid. This provides a monitoring scheme that disciplines the behavior of the firm's board compensation committee."~

the

directors of the committee becorne more focilsed and more concerneci about the relation-

ship between executive compensation and corporate performance, the pay-performance relation is intensifieci. Cornmenthg on the new Ontario disclosure niles, Dimma (1994) predicts that they d reçult in more emphasis on both short-term cash boniises and longterm stock options or stock appreciation rights, with less emphasis on salaries. Dimma (p45) explains this impact as the fouowing:

'With increased disclosure, shareholders activism - primarily but not entirely that of institutional investors - wïll increase. As aiways, knowledge is power. Fiirthermore, compensation committee mernbers will be incresingly sensitive to charges of being captive to management or of looking sok or ganting increases out of line with what their peers, sitting on compensation cornmittees of companies in the same indiistry cr in broadly similar industries, are approving. ' (2) Political pressures on high payments

The visibility of manageriai compensation is the source of the piiblic debate concerning excessive executive pay. In response to a gronring piiblic concern that execiitive pay is imjiistifiably high in the United States, some large corporate shareholders. labor and

consimer representatives, and public officials have c d e d for imposing constraints on execiitive compensation. These constraints can be in the form of either regdatory measiires or mord s u a ~ i o n . 'While ~ the purpose of these constraints is to ciirb high payments, they are expected to result in a lower compensation package and a weaker pay-performance relation as well. Jensen and Miirphy (1990a, p262) provide an explanation for this effect: gSome commentators point out that a thorough disclosiire reqiiirernent paves the way for shareholders to take back their companies (see Johnson, 1995). loin the early 1990s, the media and the recession combined to focils public attention on the payments of America's CEOs, and excessive execiitive compensation was publicly targeted. As a response to the public otitcxy to control excessive execiitive pay, the Secirities and Eschange Commission, the Financiai Accounting Standards Board, and even Congress were moving to change the rides. See Walters, Hardin, and Schick (1995), and Brownsterin and Panner (1992).

1.2

THE ISSUE AND THE LITERATURE ' h c a t i n g the iipper tail of the payoff distribution reqiiires that the lower tail of the distribution also be truncated in order to maintain levels of cornpensation consistent with eqiiilibriiim in the managerial labor market.'

(3) Pay envy among managers It has been argueci that pay cornparisons can lead to higher levels. Some call it a bootstrap effect while some others call it pay envy or jealoiisy. In disciissing the changes in Canadian C E 0 compensation due to the new Ontario disclosiire law, PvlcHiitchion

(1996) notes that compensation experts believe that 'the disclosiire law has actually helped to drive iip execiitive pay packages. ... Elvecutives tend to develop a case of s d q envy and Say "me too," when they see what other corporate heads in their industry are making.'

While this behavior-related managerial effect is widely believed to raise pay levels, its relevance to the pay-performance linkage has not been noted. As yet, there has been no (theoretical or empirical) stiidy in the economics literattire that directly investigates the disclosiire issiie. Given the empirical natiire of the issiie and a lack of data, this is not siirprising. Nevertheless. there have been some empirical stiidies that either indirectly investigate the regdatory effect or deal with a related issiie. Among these stiidies notable are Jensen and hliuphy (1990a), Joskow, Rose, and Sheperd (1993)' and Hiibbard and Palia (1995).

To explain why the linkage between CE0 pay and corporate performance is very weak in large US firms, Jensen and Mv~iirphyexamine the difference in pay-performance sensitivity and pay levels between the ûrms in the 1980s and those in the 1930s (when regdatory pressures were less evident). They fhd that both the sensitivity and pay level (with inflation adjiisted) have declined since the 1930s. They attribiite this observation

to the idhience of public and private political forces that residt frorn the visibility of

1.2 THE ISSUE AND THE LITEMTURE

9

executive compensation and which impose constraints on managerial pay contracts ,and conseqiiently suppress the pay-performance sensitivity. Joskow, Rose, and Shepard address a sirnilar issue using a different approach. To explore whether political pressures impose a constraint on management compensation, they compare CE0 compensation

between regiilated f k n s (which are presumably siibject to the direct infiilences of political forces) and linregdateci firms. They h d that CEOs of regidated firms are paid less than those of iuiregulated h s , and that the compensation eanied by CEOs of regulated finns is l e s responsive to Company profitability than is the compensation for

CEOs of iinregdated firrns. They concliide the observation is broadly consistent with the presence of binding political constraints on CE0 compensation, a s mediated throiigh regdatory rneasiires and an increased visibility of execiitive pay. Comparing C E 0 pay in the banking industry before and after deregdation of interstate banking, Hiibbard and Palia reach a similar concliision.

While these studies are consistent with a negative impact on the pay-performance relation of increased pay visibility, the evidence so far is inconcliisive. Becaiise they examine the issue only indirectly, these stiidies have difficiilty in disentangling the impact of disclosiue due to political pressures from other potential factors. l'

Besides, t hese

stiidies are miite on the role of siibstantially increased shareholderç' scriitiny of the pay system which is the most debated issue concerning governrnent disclosiire rules. Direct stiidies are needed before a firm conclilsion can be reached.

"For example, in the comment on Joskow, Rose and Shepard, Meyer points out that it is difEctdt to disentangle political constraints from prodtictivity difierences between regidated firms and iinregidated firms.

1.3 The Model 1.3.1 The Basic Model 1start with a standard simple agency mode1 withoiit disclosiue. Consider a fim and

a single worker, the manager, in a one-period contract. To obtain an anaiyticai soliition. 1 adopt the following commonly used assumptions: a risk-neiitrd h, a risk-averse

manager, a linear production technology for the manager, and a linear pay contract. The production function is

where e is the input of the manager's effort in production which is imobservable to the

firm, and

E

is a n o r m d y distributed noise with zero mean and variance

0 ' .

The linear

pay contract is

where a is fked pay, and

0 is pay-performance sensitivity or incentive dope.

The manager has an exponential iitility U(W,C) = - exp [-p(GV

- C ) ] where C =

'2 k e 2 is cost of effort,p is the coeficient of absoliite risk aversion, and k is a constant. For

a normally distribiited random variable X with mean E ( X ) and variance V(X) there is

E [ e ~ p ( - ~ x=) exp[-pE(X) ]

+ + p 2 v ( ~Hence ) ] . the manager's expected iitility is

E(U(W,C)) = - eup - p ( a

1 + @e - -ke2) + 2

The manager chooses effort e to maximize the expected utility, which gives the incentive compatibility constraint:

p - ke = 0.

The k m maximizes expected profits E(II) = E ( Y

opt imization pro blem:

- W ) by

solving the following

1.3 THE MODEL

siibject to

where Uris the manager's reservation utility. It is straightforward to obtain the following solution for the incentive slope:

This solution has been disciissed in many previoiis studies.12 Given ,B and a binding participation constraint, the soliition for fixed pay can be expressed as

The h s t term describes the linkage between the pay level and the worker's ability (reflected in reservation iitility). The second term reflects the incentive-insixance trade off

of the pay scheme. The expected total pay is

which increases with

P. Eqiiation (1.4) and (1.5) characterize the pre-disclostue optimal

incentive contract. Now 1 turn to the issue of compensation disclosiire. To keep the analysis simple, 1

deal with the effect of different factors separately while leaving an integrated mode1 to 12See, e.g., Holmstrom and Milgrom (l98?), Rosen (l992), and Garen(l994).

1.3 THE MODEL

12

an appendix. 1 e s t analyze the regulatory effect associated with political pressures and psy envy imder the conventional owner-manager Framework, assiuning that shareholders

of the hbehave as a single owner. 1 then open the black box of firm ownership by distinguishing shareholders who are directors from non-direct ors and describe a threelevel (shareholder-director-manager) model to analyze the effect of shareholders' scnitiny of the pay system.

1.3.2 Political Pressures Political pressures influence the firm's compensation practice mainly in two ways.

One way is negative piiblicity, a direct effect of piiblic disapproval of hiige managerial pay fueled by the criticism from piiblic activists. The other channe1 is regdatory mea-

siires imder government intervention, an indirect effect of the piiblic opposition t o high compensation. A featiue of politizal pressure effects is that they impose unfavorable constraints on the h m ' s operation. To describe the effects, 1 assume that when the piiblic observes managerial pay beyond a certain level. political pressures emerge t hat explicitly or implicitly impose a cost on the firm. Because political pressures do not impose a clear-ciit iipper limit on managerial pay.13 1 model the effects by defining a probability Above W the that the paÿ to the manager is higher than a piiblic norm, Prob(W > W). pay is considered imacceptable by the public.

W

reflects piiblic opinion on the value of

managers, and is exogenoiis to this model. ) the probability density fimction of Let d ( ~denote

E.

The probability is

where 9 is a normal distribution function. The more often W exceeds

W ,the larger

I3Commenting on the position of the U.S.Congress in the debate on execritive compensation in 1992, Johnson (1995) noted, "while dissatisfied, Congres appeared hesitant to act so drasticallÿ as to set explicit lirnits on execirtive compensation."

1.3 THE MODEL the probability Prob(CV >

13

W )and

the higher political pressures. The cost associateci

with political constraints thiis increases with the probability. Introdticing cost parameter

0 3 0, the objective fimction of the firm becomes

In particidar, 6 = O when there are no political pressures or when there is no piiblicly available information on W. With other conditions unchangeci fiom the basic model, solving the problem leads to the following concliisions:

fin.compensation disreduces the incentive paramefer, 8, reduces the fked composent 01pay, a. when

Proposition 1 W h e n political pressures impose a cost o n the closure

kpa2 > 1 , and reàuces the ezpected wage, E(M7).

Proof.

Appendix A.

Proposition 1 predicts a negative impact of political pressiires on the incentive slope.

It conf~rmsthe hypothesis of Jensen and hlurphy (199Oa) that public and private political forces suppress the pay-performance sensitivity. However. the intuition here is somewhat different. Because a higher pay-performance sensitivity causes an iipward shift of the pay clistribiition and thus more Likely poiitical presstues, the marginal benefit hom an

increased sensitivity is rediiced compared to the situation wit hoiit political pressures. The effect on h e d pay is arnbigiiotq and depends on model parameters. This ambiguity reflects the trade off between incentives and insurance. When the manager is sufficiently risk-averse or when production is siifficiently risky, the trade off will dictat e that k e d pay changes in the sarne direction as does the incentive slope. With reservation utility imchanged, the expected total pay is bound to change with pay riskiness, and

1.3 THE MODEL

conseqtiently, it is lower with political pressures.

1.3.3 Pay Envy Supposing that there exists pay envy or jealoiisy among managers, a manager's iitility

is increasing in the difference between her pay and those of her peers. To determine the effects of pay envy, I assume a manager's u t i l i ~declines as the expected pay differential

E (W, - W ) rises, where W, is the wage of peers. With this assiunption the manager's utility fimction is rewritten as

U(cK CoW,) = - exp [-p(W - C - hE(Wp- W ) ) ] where h 2 O. Siich a treatment is similar to that of peer pressure in Kandel and Lazear

(1992) where peer pressure is a coût to utility. Parameter h characterizes the intensity of the manager's concern aboiit a lower pay.

Though the nature of the manager's tastes do not change nith the openness of managerial compensation, the intensity of envy and its effect on iitility certainly depend on the available information and piiblic attention. So. h = O is iised to mode1 the pre-disclosine effect of pay enky. With the sarne information on other managers' pax the firm and its manager wil1

form the same belief on the expected pay of other managers. U$.

Y = e + E and pay contract W = cr

Under production

+ PY, the expected iitility of the manager becomes

Talung W,bas given, the manager maximizes the expected iitility. The incentive compatibility constraint becomes (1+ h)B - ke = O. Given 0 , the manager woidd now like to work more. Because pay envy stimulates the manager to earn more, it enhances working incentives.

1.3 THE MODEL

15

When the belief on other managers' expected pay is correct and when managers are hornogenoiis, t hen

With siich a setiip, the following proposition follows immediately:

Proposition 2 When managers exhibit pay en%

compensation disclosure increases the

incentive parameter, ,O,increases the jixed component of pay. a

. when kpa2 + h2 > 1.

and increases the expected wage. E ( W ) .

Proof.

AppendixB.

Piiblic critics have attacked executives for seeking high compensation. They are rnissing an important point: the desire to earn more than others encourages managers to

work more and, thus, to accept a more risky contract. which is beneficial to mitigating agency costs. The expected total pay is increased because pay is tied more closely to

performance. The public is right in assuming that the race for earnings among managers resiilts in higher pa-yments. However, Proposition 2 gives an efficiency e.xplanation of this phenomenon. Cornpetition for earnings stirnidated by pay envy acts iike a toiunament

incentive scherne.

1.3.4 Shareholders' Scrutiny In the above analyses, the firm is synonymous with shareholders who contract directly with the manager. When shareholders are divided into two groiips , directors and non-directors, the ''£km" no longer refers to a i d e d groiip of investors, and a conflict of interest may mise between the board of directors and other shareholders. As the black box of h m ownership is opened, another agency problem emerges: how are directors

1.3 THE MODEL

16

induced to perform their fiduciary diities in the best interests of ail shareholders? The incentive issues in such an environment call for a model describing the relationship between shareholders, directors, and managers. In this section 1 fbst describe a three-level pre-disclosure mode1 in which, by assumption, the directors and manager are inclined to collude (at a cost to shareholders) and then introdiice disclosiire to determine its disciplining effects. The issues of political pressures and pay envy are ignored in this model.

The Pre-Disclosure Mode1 Consider a piiblicly-held h m wit h many risk-neutral shareholders. The shareholders delegate decision making aiithority to their representative, the board of directors, to

design and implement the managerial pay contract. For simplicity, there is assiunecl to be only one director on the board. The non-director shareholders will simply be called shareholders. There are two activities in the model: the manager exerts effort to prodiice firrn oiitpiit; and the director monitors the manager, and designs and implements the managerial reward scheme. The folIowing assiunptions define the informational asymmetries between the shareholders, director, and manager. (i) PNhile the shareholders h o w the director's ability, they observe neither the director's monitoring effort nor her choice of managerial contract."

(ii) The director knows the ability of the manager, observes the

b ' s performance, but does not observe the manager's effort. (iii) The manager observes both the ability and monitoring effort of the director. There are two agency relationships, one between the shareholders and the director n I ' reality, shareholders certainly have information, other than executive compensation, on the directors' performance. We assume here that this information is irrelevant to manageriai pay schemes, and does not reveal what directors do in motivating the manager. Later, disclosilre will reveal the managerial contract to shareholders.

and the other betnreen the director and the manager. The foilowing time line depicts the seqiience of events in the model:

Shareholders choose director's contract F

Director chooses manager's contract W

Production and monitoring effort invested

Firm output obsewed

Paj-ments made

The shareholders offer to pay F to the director for her fiduciary dtities? and then the director contracts with the manager with pay scheme W for the task of prodiiction. There is only one period of production. Both F and W are specified before prodiiction

and monitoring activities begin. At the end of the period, the otitcome of h m prodiiction is observed and payments are made accordingly.

( 1) The Shareholders

The firm's net profit is Ii = Y

- CV - bT

where Y is total revenue net of prodiiction

costs. bT is a benefit (side payrnent) obtained by the director that is privately paid by the manager to the director at a cost to the firm. T denotes the managerial rente and is described in more detail below. The shareholders observe components because they are rmable to distinguish W

II but

cannot identib its

+ bT from other prodiiction costs.

The shareholders choose the director's pay scheme, F, to mavimize the expected net payoff:

(2) The Director

1.3 THE MODEL

18

Given F , (which is detailed later below), the director faces a three-fold decision: choosing the managerial pay scheme W , deciding the monitoring effort m, and determining the rent T received by the manager. The objective fimction of the director is

where O 5 S < 1 is the portion of the director's interest in total shrtreholder wealth.

kgm2is the cost of monitoring effort, where g

2 O and m 2 0.

In conventional incentive models the performance measure for the manager is welld e k e d and observable to both the principal and the agent, and so it is costless to the principal. In reality, however, performance measiirement is far more complicateù where smbiguities abotmd. It is the complexity of performance measilrement that calls for monitoring effort from the director. Taking Y = e

+E as a costless performance measiue.

an adjiisted measiue of performance imder monitoring becomes15

Q = e +~ ( 1 m).

(1.10)

The key feattue of (1.10) is that the director's effort rediices the noise in Y. When

m = O, Q = Y which is the case withoiit the director. &%en m = 1' perfect monitoring is achieved and managerial effort becomes observable to the director. Based on Q, the

pay contract for the manger is

W=a+OQ (3) The Rilanager 151t is implicitly assimed that rn

5 1.

1.3 THE MODEL

19

The manager is risk-averse and with exponential iitility

U

= - exp [-p(CV

- C)]

where C = $ke2 is the cost of managerial effort. The participation constraint is now revised to d o w a rent for the manager:

Ur is the component of reservation utility corresponding to the manager's prodiictivity. deterrnined imder a cornpetitive market. T is non-negative. This formulation allows the director to collude n i t h the manager by setting T > 0: the director pays the manager excessively (T > 0); and the manager provides a benefit, bT, to the director a t a cost to

shareholder weal t h.

{T:b} is assimed to be an implicit contract between the director and the manager that is enforced through bilateral reputational concems. Becaiise b does not impose a cost on the manager and the director dways prefers a higher 6, it woidd be imboimd in an one-period model (siibject to a constraint on reolized valiies of ri - F). So b is taken here to be exogenoiis. The model, therefore, does not endogenoiisly determine the existence of a managerial rent. It will identiS the lower boiind of b that @es the necessary and

sificient condition for T > 0. (4) The Soliition

The model is solved by backward indiiction. The manager's problem is straightforward. Taking the director's monitoring effort m as given, the manager chooses e to rnrru-

+

1 imize the eiupected iitility, E ( U ) = - exp { - p [(a B e ) - $e2]

+ ip2a2(1- n ) 2 ~. 2 )

which gives the incentive compatibility constraint: ,L? - ke = 0. To solve the director's problem, the pay scheme for the director needs to be specified. Without information on the director's actions and decisions, the h m ' s net profit is the

1.3 THE MODEL

20

only observable and verifiable measiire on which the director's pay c m depend. Consider a linear scheme,

where O 5 sl < 1. In a piiblicly-held firm, the director w o d d not be a residiid claimant.l6 Therefore' s t is smaller than imity.

The director chooses {a,0, rn,2') by solving the following problem:

subject to

Proposition 3 siunmarizes the main results of the solution to (1.14): given {so, SI} .

Proposition 3

(2)

The director rnonitors the manager ( m > 0 ) and specifies excessive

incentives for the manager

(0> &)

ij she is also a shareholder (6 > O ) ; (ii)

> 0; and (iii) the manager captures rents (T > 0 ) if only if

and

2 >O

1- 6+( 1-613,) p(-Ur)) (

6+(L-6)?i

,

1.

Proof.

Appendk C.

Parts (i) and (ii) are intuitive. As long as the director shares some interests with

the shareholders, as is the mual case in reality, the director will invest effort to monitor management, i.e., rn > O. This resiilts holds even if the director is paid a fixed fee 16Risk-sharing and public financing are two important reasons for the existence of a pu blicly-held Company, which excliide the possibility for the director to be a residiid claimant.

1.3 THE MODEL (sl = O).

With monitoring effort, the managerial pay scheme becomes more efficient

because of the rediiced noise in the performance measure. Both monitoring effort and incentive dope increase with the common interest parameter, 6. As shown in the proof of the proposition,

fl=

,++.when m = O, which is the zero-monitoring solution of the 2

basic mode1 without the director. When 6 = 1, the director's problem rediices to the one where the shareholders directly rnonitor the manager, and t hm, the ideal investment of effort is achieved and ,O is optimized in the interests of the shareholders. Part (iii) airns at a controversial issue: are managers paid excessively? The answer is inconcliiçive yet interesting. Mihen T > O, both the manager and director receive excessive pay and the payoff to the shareholders is accordingly rediiced. The proposition gives the condition under which director-manager coilitsion occius. Several factors affect this condition. Because of the conflict between the collusive benefit received by the director and the cost of collusion associated with the rediiced firm profit, the director's willingness to collude depends cnicially on 6. T becornes zero when 6 is sificiently close to iinity. Another factor is b which reflects the marginal benefit to the director from a rent captiired by the manager. The value of b is related to the ways in which a benefit from colliision can be delivered and the intensity of public scriitiny of the colliisive condiict.

Finally, the shareholders' problem is to choose {so. s 1) to rnaxirnize the e'ipected payoff, E(I1 - F ) . Consider two sitiiations. (i) The shareholders do not know the manager's ability (Le., k and

L& are iinknown parameters to the shareholders) or they

are unaware of the constraints on the fiinctional form of the managerial pay contract. In this situation, niithoiit the information needed to solve a maximization problem. the shareholders simply pay the director a k e d fee at reservation wage Fr. Or equivalently, so = Fr and si = O. (ii) Alternatively, suppose the shareholders know k and

Ur, and have

the knowledge of the stochastic process of Q and functional form of W even thoiigh they

do not directly observe Q and W. Then the shareholders solve the following problem to determine {sot sl}:

Max E(II - F ) = ( 1 - s l ) [(l -,O)e - a ] - b ( l - s l ) T - so

(1.15)

{so*s1)

The first and the forth constraint are the incentive compatibility constraint for the manager and the director respectively. The second and fifth are participation constraints. Both the third and forth constraint corne from the first-order conditions of the direct or's pro blem.

In either situation, the following concliiçions hold. Proposition 4 Before disclosure. ( i ) directors underinuest in monitoring managers and

the incentive parameter,

,O,zs consepently smaller than is the parameter when share-

holders directly monitor managers; and (zi), incentive parameter'

Proof.

0,are

both the directors' monitoring effort and

independent ofmanagena1 rents. T .

AppendixD.

Clearly, the director woidd not be a perfect representative of t,he shareholders. The

imder-invat ment of monitoring effort occurs becarise of the divergence of interests between the director and the shareholders. This agency cost, unlike that in standard agency rnodels (which exists because of privately observable effort and risk-averse agents), restdts

Erom privately observable performance and the ownershipsharing nature of a piibliclyheld fum.Even a firm performance-based pay scheme for the director would not eliminate

1.3 THE MODEL

23

the agency cost, becaiise siich a scheme only increases the direct or's interest share in the

£km and does not change the nature the problem. The second part of the proposition disentangles t h e incentive issue from excessive managerial pay.

Public critics attribute both high execiitive pay and a weak pay-

performance relation to director-manager collusion. The proposition dismisses siich a linkage. Thoiigh the director may have an incentive to pay the manager excessively. it is not in the interest of the director to weaken the incentive strength.

The Post-Disclosure Mode1 Under disclosiue niles, the board compensation committee has to detail both pay and performance related information. and thiis, imder ideal sitiiations all a. 3. Q. and

Y become piiblicly observable. Two things happen accordingly. First. a managerial rent no longer exists. Though t h e shareholden do not observe T. piiblic information on CV

and Urforces the participation c0nstra.int to be set as E ( U ) 2 Ur. Second, with information on Y, I.V and Q, the shareholders can evaliiate the performance of the director.li For simplicity. assiune a linear performance measiire for the director's monitoring activity, ill = rn + [, where [ is normally distributed with zero mean. hl is public information. The shareholders sign a Linear pay contract with the direct or,

Disclosiue d e s do not change the information flow between the director and the manager, and so the manager's problern remains imchanged. But the problem for the -

-

....--

--

---

17For example, Y - Q = rn can serve as an performance measure.

1.3 THE MODEL

director becomes the foilowing:

subject to p

- ke = O and E(U) 2 Ur.

With the director-manager problem being determined by ( 1.17). the shareholders solve the following optimization problem to determine { fo, fi):

M a x E ( n - F ) = (1 - 0 ) e - a - fo - f i m

IjoJi )

The constraints in (1.18) correspond to those in (1.15).

Proposition 5 ,4fter disclosure. (i) m = m* and

9 = 0' where m* and 8'

are the jirst-

best moniton'ng solution (when the shareholders directly contract with the manager): ( i i ) the f i e d component of pay, a? is increased if kpa2 > 1 and T = O before disclosure: and (iii) T = O before disclosure 2s a suficient condition for the expected wage. E ( W ) . to increase, and T > O before disclosure i s a necessary condition for the expected wage to decrease.

Proof.

Appendix E.

As the shareholders can directly evaliiate the direct or's performance, the reward scheme for the director becomes efficient. The solution to (1.18) is the same as that to the pre-disclosure mode1 with 6 = 1. In other words, after disclosiire, the director is indiiced to perform as a perfect representative of the shareholders. This residt is not

1.3 THE MOD EL

25

surprising. Since the director is risk neutral, the unobservability of the director's effort is not sificient to pose a moral hazard problem. The public scrutiny of the managerid pay scheme thus solves the agency problem between the shareholders and the director.

The incentive slope of the managerial pay scheme is increased becaise of the increased monitoring effort from the director. The positive disclosiire effect on the pay-performance

linkage is anticipated by many public commentators. The intuition is simple: shareholder scriitiny of the compensation system forces the board of directors to invest more effort

in monitoring management and to better perform their Bdiiciary duties. Changes in fixed pay and e-xpected total pay are arnbiguoiis, depending on mode1 parameters. It is worth noting that part (iü) of the proposition points otit a possibility to identi& the existence of pre-disclosiire managerial rents.

1.3.5 A Summary of Disclosure Effects It is straightforward to incliide political pressires and pay envy into the shareholderàirector-manager hamework. However, since none of above residts will change qiialitatively, 1leave the description of the integrated mode1 to an appendk. The following table s i m a r i z e s the partial effects disciissed above. Influencial Factors Shareholders' scriitiny Pay Envy Polit ical Pressures Pre-disclosiue rent (T > 0) -

+ +

+ or + or -

-

-

-

no

-

-

-

-

or

+

+ + -

. .

-

-

-

A@,Acr, A E ( W ) denote changes due to compensation disclosiire in incentive slope, Lxed pay, and expected total pay r e s p e c t i ~ e l ~ . ~ ~ -

-

-

-

p p

-

181t needs pointing out that there is a difference between the table and Proposition 5 with respect to the effect of shareholders' scnitiny on E ( W ) . In the table, the effect of pre-disclostire rent is separately

iisted.

1.4 EVTDENCE FOR

CHIEF E X E C U T m OFFICERS

26

In the public debate and economics literatiire, arguments iisiiaily focus on one effect while ignoring the others. When all these effects are taken into accoimt, the total regdatory impact on both the pay-performance sensitivity and expected wage becornes ambigioiis. It is essentially an empirical matter. Now 1 proceed to examine the evidence. Because the post-disclosure period is short in the current data, it is difficidt to estimate changes in the expected wage. Given the general interest in the efficiency of the pay system, 1 wili focus on the impact on the pay-performance sensitivity.

1.4 Evidence for Chief Executive Officers 1.4.1 The Ontario Regulation Act Until October 1993, proxy disclosiire reqiurements imder the Ontario Security Act called for disclosure of aggregate total compensation for ail execiitives. In contrast, US

niles governing disclosiire of execiitive compensation reqilired individual named disclos u e , in addition to aggregate information.

As a response to piiblic criticism of the

disparity between Canadian and US reqiurements, in the fall of 1990 the Ontario Seciirity Commission began to review executive compensation disclosiire reqiiirements. In early 1992, the commission settled iipon a proposa1 for aggregate disciosiire of the top five executives with individual disclosiire only when an execiitive e m e d more than forty percent of the total compensation received by the five persons receiving the highest aggregate compensation. The new regulation was announced on October 14' 1993, which went much hirther than the proposal. requiring thorough, named disclosiire of the five highest paid executives' compensations. The new regdation makes the disclosiire reqilirements

very sirnilar to those in the United States, piitting "Ontario at the forefront of openness and accoimtability to shareholders" (Laiighren, p5106). The regulation took effect on October 31, 1993. The disclosure rides include three

1.4 EVZDENCE

FOR C H E F EXECUTWE OFFICERS

main components. First , corporations must detail total executive compensation in several tables with a standardized format. Completing the tables reqiiires compensation information for the ciment fiscal year and the two immediately preceding years. The ndes reqtiire compensation disclosure regarding the C E 0 and the f o u highest paid executives with total compensation in excess of $100,000 in the most recent fiscal year.

Such a reporting format provides shareholders with a concise, comprehensive o v e ~ e w of compensation awarded, earned or paid in the reporting period, and makes it easy for shareholders to compare compensation practices between reporting corporations. Second, the compensation cornmittee of a corporation mtist disclose to the shareholders its compensation policy - the b a i s for the compensation granted to senior execiitives. While a general statement of compensation policy is reqiured for all reported senior mecutives, a specific statement is required for the C E 0 to specify the relationship of corporate performance to compensation. The compensation committee report reqiiirement permits shareholders to assess how well directors are representing their interests. In order to ensure that shareholders know which directors are to be credited or blamed for rompensation polices. the report is made dong with the names of the directors serving on the compensation comrnittee. Third, the disclosiire rtdes reqiure a corporation to graphically present the corporation's performance by cumulative total shareholder rettun. This figue miist be compared to both a market index and an indiistry or peer group index. The graph m u t show data for a minimiun of five years incliiding the base year. This reqtlirement complements the compensation committee report and siunmarizes information on the relationship of executive compensation to corporate performance in a given fiscal year. Given the close relationship between the Canadian and the US market, it is important to understand the US disclosure rides to properly document the impact of the

1.4 EWDENCE FOR

CNlEF EXECUTWE OFFICERS

28

Ontario regdation. Ever since corporations began ta disclose top execiitives' compensa-

by corporate directors to motivate senior tion in the 1SOS, compensation practices i ~ e d managers have corne under public scrutiny in the United States. In the early 1990s execritive compensation becarne an issue for public debate. blany believed that sornething was wrong: "First, executive compensation was disproportionate when compared wit h

the salaries paid to the senior exeçiitives of foreign corporations. Second, the level of execritive compensation had increased much more rapidly than salaries of the remainder of the American work force. Third, executive compensation was not related to corporate performance." (Johnson, 1995, p191) Institutional investors were bringing increasing pressures to bear on the regulatory agencies to support mechanisms which wodd increase institutional investor influence on corporations. Faced with the threat of impending Congressional action, the Seciuities Exchange Commission responded by adopting regdatory changes in late 1992. The regdation was r e h e d by a second set of changes near the end

of 1993. There are three changes in the US law. First, the new regidation reqtùres corporations to disclose senior execiitives' compensation iising a standard format. As for the visibility of executive pay, this reqiiirernent is not miich different barn the early niles. Second, the new regulation reqiiires the compensation cornmittee to disclose its compensation policy to the shareholders. Third, cornpanies are reqillred to provide a performance graph. as described above for the new Ontario regdation. The new Ontario regidation was clearly influenceci by, and siibstantially similar to, the new US disclosiire reqiiirements adopted in late 1992.

1.4.2 The Methodology and the Data Regdatory impact is evamined in two ways. First. 1 examine the pay-performance relation before and after the regdation was introduced for all piiblicly- traded Canadian

1.4 EVIDENCE FOR CHIEF EXECUTIVE OFFICERS

29

firms. Second, I compare post-disclosure changes in the pay-performance Linkage between firms affected by the regulation and others that are less d e c t e d or imaffected by the regulation. The latter is a "ditference in difFerences" approach. which has an advantage in controbng for the effects of factors other thao the regulation.

The "difference in differences" analysis includes two comparisons. One comparison is made between Canadian fLms whose shares were exclusively traded in Canada before

1991 and other Canadian firms whose shares also traded in the United States. Before 1991 Canadian compaoies publicly-traded in the US.were bound by the U.S. regulations and so the information on top executives' compensation for these companies were accessible to shareholders through the were already rdFected by the

US. market. I thus identify these firms as ones that

US.disclosure rules and would be less a£Fected by the new

Ontario regulation. The other comparison is made between Canadian firms and US firms (which were not affected by the Ontario regulation). Under the assumption that the compared samples were subject to different regulatory 'shocks' and in an otherwise similar environment,'"he

difference in post-disclosu~rechanges of managerial pay schemes

is expected to reflect the regulatory change in Ontario.

We may assume that the disclosure rules were not anticipated in early 1993. and so take the fiscal years before and including 1993 to be the predisclosure period. This assumption is based on the following discussion. The proposal of new disclosure rules was put forth to the public in March 1992.~' The proposed rides require disclosure of the

aggregate compensation of the five highest paid executives, which is similar to the existing lg More specifically, this assumption includes: (i) the introduction of the disclosure law provides an exogenous variation in exec~~tive compensation schemes, (2) all other factors including secular trends, but other than the discloslue law, are common to the compared samples. While this assumption defines the ideal conditions for the "difference in differences" approach, it describes qttite closely the situation between Cornpustat and non-Cornpustat Canadian firms. The second condition, however, may not be held desirably for the relationship between Canadian firms and US firms due to the 1992 and 1993 revisions of US disclosure laws. This issue is addressed in the next section in the discussion of the results for the comparison between Canadian firms and US firms. 20 See Wright (1992).

1.4 EVIDENCE FOR CHrEF EXECUTIVE OFFICERS

30

disclosure requirements in regard of the opemess of execiitive compensation. Althoiigh disciissions on possible regulatory changes in Ontario can be traced to as early as April 1991,2i the public did not seem to anticipate the new disclost~rendes before mid-1993. For example, the editorial article in the Financial Post on May 12, 1993 complained that "As the U.S. moves forward, Canada treads water. The Ontario govenunent cannot even bring itself to adopt the OSC proposal for limited disclostue." Given that companies rnake decisions on pay schemes iisudy at the beginning of a fiscal year and that the new regulation was annomceci in late 1993, it is then reasonable to take fiscal year 1994 to be the k s t year when the new regdation infltienced the compensation system.

Canadian C E 0 compensation data are compiled hom the corporate proxy statements a e d with the Ontario Seciirity Commission and mailed to shareholders. The data cover a l l proxy statements 6led diuing October 1993 to July 1995, which report execritive

compensation for fiscal year 1991 throtigh 1994. The data contain detailed information of diffesent cornponents of exectitive remuneration. Ciment cash payments sala^, anniial boniiç, and varioiis benefits) are reported, dong with stock options (or stock appreciation rights), grants of restricted stock shares, and long-term incentive plan payoiit. Financial data are obtained fiom the Financial Post data files. Excliiding partial compensation diie to turnover and firm reorganization, and eluninating observations mith missing data after matching C E 0 compensation with fim financial data, the Canadian sarnple contains 381

firms for a total of 1209 observations. Becaise some firms do not have complete f o u years of data, the sample constitiites an unbalancd panel. 1 identify the Canadian fimis that were publicly traded in the United States before

1991 as those that appeared in the Compustat data file for fiscal year 1991. I cal1 them Compustat £hm.Arnong the Canadian companies piiblicly-traded in the United States * ' ~ h efirst article in the Financial Post that mentioned the Ontario Seciwities Commission's plan to introduce regulatory proposais was piiblished on April 15, 1991 (see Whyte).

2.4 EVTDENCE FOR CHlEF EXECUTWE OFFICERS

31

in 1991, more than 500 were listed in the file. Because the CE0 compensation data essentidy cover all large publicly-traded Canadian firms,those included in the sample and also traded in the US market are expected to be listed in the Compustat data file.

Of the total h

s in the sample about 70 percent are Compiistat finns.

A Forbes sample over the period 1991-1994 is used for US 6rms. The Forbes report of CE0 compensation includes total salary and bonis, stock gains. and other payments. Total salary and bonus is the same as the siun of sdary and annual bonus in the Canadian

C E 0 compensation data. Stock gains are the net value realized fiom the exercise of options or stock appreciation rights previoiisly awarded. Other payments inclitde long-term incentive payments and varioiis benefits. The financial data for US firms are obtained fkom the Compustat data files. The US sample contains 666 firrns Mth a total of 2133 observations. Selected statistics of C E 0 pay as well as some f k n variables are siimmarized in Table 1.1. There are two things worth mentioning. First, there is a notable difference in firm size between Compustat and non-Compiistat Canadian firms, and a siibstantial

difference in h m size between Canadian firms and US finns. Second, cornpared with the pre-disclosiue period, fhns generally performed poorer in 1994. Because of limitations of the data,22 the investigation of this paper will focus on C E 0 direct pay excliiding

stock options. Specifically, the disciiçsions will be based on two pay variables: salary plus bonus and total direct pay which apply to both the Canadian sample and the US sample. The effects of stock ownership and options will be briefly disctissed later.

22~tock option data are incomplete: the Forbes stirvey does not report ciirrently awarded stock options the Canadian firms' proxy statements did not disclose the relevant stock market price, exercise price, and expiration date for stock options granted in previoiis fiscd years.

1.4 EVIDENCE FOR

CHIEF EXECUTlVE OFFICERS

1.4.3 Empirical Results As in Hubbard and Palia (1995), the foilowing specification is iised to estimate the pay-performance sensitivity:

(CE0 Pay), = a + Bi (Shareholder Wealth), + & (Shareholder Wealth),-, .

C E 0 pay is either salary pliis bonus or total direct pay. Total direct pay incliides sala% boniis, long-term incentive payrnents and fringe benefits, excliiding stock options. Shareholder wealth is defined as the stock retiirns earned diring the year, times the price at the beginning of the yearo times the niimber of shares otitstanding. As past performance may also have an influence on ciment compensation,23 a lag term of shareholder wealth is included as a regressor. The total sensitivity is ,BI

+ ,O2. A diunmy variable for fiscal

year 1994 is introdiiced to captiire post-disclosiire changes in C E 0 pay schemes. 1 first examine how the pay-performance sensitivity changed in Canadian firrns after

the new Ontario regdation. Table 1.2 presents the regession restdts with all Canadian firms. The pre-disclosiire sensitivity for C E 0 salary+bonits and total pay is abolit 8 cents and 11 cents, respectively, in response to every $1000 change in shareholder wealth. Mter disclosiire the sensitivity increased to 34 cents for salary+boniis and 30 cents for total pay respectively. The large magnitude and high significance level of the sensitivity increase siiggest a substantially strengthened pay-performance linkage. To the extent that sensitivity declines with h m size and increases Mth a firrn's market p e r f ~ r r n a n c e , ~ ~ the changes in sensitivity may be imderestimated because Brm size was larger and performance poorer in fiscal year 1994 (see Table 1.1).

An obvious criticism on these estimates is that there may exist other factors that -

-

- -

pp

23See, e.g., Jensen and hIiirphy (l990a)and Joskow and Rose (1994). 2 4 ~ist well documented that the pay-performance sensitivity changes inversely

with firm size (sw Jensen and Miirphy, 1990a;and Garen, 1994). Some early studies note that managerial pay is relâtiveIy less reçponsive to poor performance than to good performance.

1.4 EVlDENCE FOR C H E F EXECUTWE OFFICERS

33

caused the observed changes in the sensitivity. To address this issiie. 1examine the differ-

ence in the post-disclosure changes of the pay-performance sensitivity between Cornpiistat £hmand non-Compustat firms. Recd Compustat firms are Canadian fbms siibject to

US disclosure ndes, which were, presumably, less aifected by the new Ontario remdation than were non-Compiistat Canadian firms. Assiiming that the eifects of non-regdatory

economic factors are common to both groups of firms, the difference in their response to the regdation (and therefore, different changes in sensitivity &ter disclosiue) will reveal

the regdatory impact. The residts with a diimmy variable for non-Compiistat firms are presented in Table

1 . 3 . ~The ~ coefficients on the difference in post-disclostue changes (i.e. the terms with d i m y variable YEAR94x NONCOLIPUSTAT) are ail positive, and becorne simgpifmint at the one percent level for the regression with salary and bonus. Incliiding the coefficients

on both the ciment and the previous year's performance, the sensitivity increased by 0.000155 with salary plus boniis, nnd by 0.000353 with total pay. for Cornpiistat fums. The increase is as high as 0.000814 with salary pli= boniis and 0.000779 with total pay, respectively, for non-Compiistat f i r m . After disclosiue. while the sensitivity increased in both Compiistat and non-Compiistat firms, i t increased much more in non-Compristat

k m s . These results strongly siipport the view that disclosiue riiles strengthen the payperformance linkage. Fiirther, if Compustat firms and non-Compustat firm were in a different regdatory environment before the new Ontario regdation, we shoidd be able to observe a difference in their pre-disdosure sensitivity consistent with the regdatory infliience. This is also verified in Table 1.3. The pre-disclosure sensitivity is estimateci by the coefficients on 2 5 ~ h estirnates e are adjusted for hetemscedasticity between Cornpiistat and non-Compiistat firms irsing the two-step, GLS estimator. This adjistment is applied also to the regressions in Table 4 which compare US firms with Compiistat and non-Compuçtat Canadian finns.

1.4 EWDENCE FOR

CHIEF EXECCTTm OFFICERS

34

performance variables withoiit diunmy variable YEAR94. It is 0.000093 with salary and

bonus and 0.000123 with total pay for Compiistat firms, and 0.000052 and 0.000094, respectively, for non-Cornpustat

Consistent with a positive regdatory impact on

the pay-performance relation, the sensitivity is Iarger for the firms that were afFected by the US disclosiue regidations. Taking into accoimt both the pre-cùsclosiue sensitivity and the post-disclosirre changes. the above estimates indicate that the percentage increase in sensitivity for non-Compiistat firms is 3 to 9 times as large as for Compiistat firms. Because the non-Compiistat sample is made tip of s m d capitalization firms, one may argile that the effect of disclosure disciissed above rnerely represents a year (1994) specific effect on C E 0 compensation in s m d firms. Fiirther, becaise small Enns are iisually relatively young, the residts may reflect the effects of aging. To clai@ this problem, 1 divide the total sarnple into srnail h

s and large firms while ignoring the

"Compiiçtat" dimension. The mode1 is reestimated by introdiicing a d i m y variable for small firms. While the increase in the post-disclosiue sensitivity is rnodestly larger in

s m d firms than in large firrns, the difference is statistically insignificant. -4s the predisclosiire sensitivity is taken into accoimt the difference between small firms and large firms becomes even more obsciired.

In siunmary, the residts in Table 1.3 show that (i) before the Ontario regdation was introdiiced, the firms that were not affected by US disclosiue rides had a smaller

pay-performance sensitivity, and (ii) after the Ontario regdation was introdiiced, the sensitivity of these firms increased siibstantially relative to those firms that were already covered by the US regulations. This observation is consistent with many public 2 6 ~needs î pointing out that the significance level is quite low for the difference in the pre-disclosire pay-performance sensitivity between Compiistat and non-Compustat fkms. This is expected to reflect a firm-size effect. As mentioned earlier, non-Compustat h s are small and the sensitiviw parameter is larger in smaü firms. The observeci difference in the pre-diçclosiire sensitivity is downward biased when fims that have a smaller pre-disclosiue sensitivity are with a s m d e r size. This discussion aiso applies to the residts for the cornparison between Canadian firms and US firrns.

1.4 EVIDENCE FOR

CHIEF EXECUTWE OFFICERS

35

commentators' prediction of the regulatory impact: Execiitive compensation disclosiire increases the public scnitiny of the pay system, and thiis forces the board of directors to tie managerial pay more closely to corporate performance. Given the observed difference between Compiistat and non-Compiistat firmso i t is natiital to ask how the two groups of Canadian firms are different from US firms in response to the new Ontario regdation. Presumably: Compustat Canadian 6rms were in a similas regulatory environment as were US £hmregarding to execiitive compensation disclosine. So, if the difference between Compustat and non-Cornpiistat firms reflects the effects of disclosure, then we should be able to observe similar post-disclosiue changes in sensitivity between Compustat Canadian h m and US firms, and to observe a larger increase in sensit ivity for non-Compiistat h

s t han for US firms. To verify this conject lire,

1 nin regressions for the pooled sample of ail Canadian and US firms, and use a diunmy

variable for Compustat and non-Compustat fkms respectively. The resiilts are presented in Table 1.4, and are consistent with the conjecture. The terms with diunmy variable

COMPUSTAT capture the difference between Compustat Canadian firms and US firrns. and the terms with diunmy variable NONCOhIPUSTAT capture the difference between

non-Compiistat Canadian fhms and US firms. On one hand, after the Ontario regdation was introdiiced, the sensitivity increased rniich more for non-Computat Canadian firms

than for US firms. With salary and bonlis. the post-disclosine (absolute or percentage) change for non-Compiistat firrns is about 4 tirnes as large as that for US firms, and the difference is statistically si,dficant. While the estimate for total pay is weaker, the qiialitative resiilt remains the sams. On the other hand, there is no sigdicant difference between Cornpustat Canadian b

s and

US firms. even though the post-disclosiue

changes are slightly smaller for Cornpiistat firms. This observation tends to siiggest that the Canadian firms that were already subject to the US disclosiire regdations were not

1.4 EVIDENCE FOR C H 1 . F EXECUTTVE: OFFICERS

affecteci by the new Ontaxio regulation. While an increase in the sensitiviiy of both Canadian fims and US firms may partly

result from similar stock market fliictuationç, it is possible that there existeci similar regulatory effects on US fkms in 1994. As mentioned earlier, the Seciinties and Exchanges Commissions in the United States revised the long-standing US disclosiire ndes in late 1992 and 1993. While the new disclosiire rules do not change the visibility of execiitive compensation (and thiis the effect of public pressure and managerial behavior), they siibstantially increase the public scnitiny of the b ' s compensation poiicy. The revision of disclosiire ndes is widely believed to have a profoimd influence on the fu?n's compensation practices in tying executive pay more closely with corporate performance. There was another event occ~medin 1993 in the United States that was also expected to increase sensitivity. In response both to the public oiitcry concerning excessive execiitive compensation and the administration's need to raise federal reveniie, the US congress passeci new restrictions on the arnoimt of execiitive compensation rvhich corporations can dediict from their income (the Revenue Reconciiiation Act of 1993). Under the new t a codeoa publicly-held corporation generally cannot dediict any compensation over $1 million paid to eseciitives from its g o s s income. To encourage corporations to adopt performance based compensation practices, however, a corporation can avoid this lirnit if the compensation is characterized as performance related. The new bill. dong with the revised disclosure niles, woidd aùnost certainly force ptiblicly held corporations to strength the linkage of execiitive pay to firm performance. So, becaiise of these possi-

ble regdatory effects on US firms, the observed difference in the post-disclosive changes in the pay-performance sensitivity between US firm and non-Compiistat Canadian firms may underestimate the regdatory impact of the introdiiction of the new Ontario regdation.

1.4 EVIDENCE FOR

CHIEF EXECUTWE OFFICERS

37

The discussions above do not consider incentives from CE0 stock ownership and options which, as previous studies note, provide important components of incentives for executives. Thus, there may be concem for whether or not the changes in direct compensation accurately reflect the total change in managerial incentive strength. Siich a concern, thoiigh reasonable, does not seem to seriously adfect the above discussions. First, in a short post-disclosilre period, executive stock holdings are iinlikely to be affectecl

by her firm's irnmediate change in compensation strategies. Also, execiitive stock gains from exercising previoiiçly ganted options (which reflect long-term incentives of stock options) are largely independent of policy changes that affect the crirrent or very recent option grants. Second, because direct compensation (incliiding salary? boniis, gants of stock shares, long-term incentive plan payout. and vmiciis benefits) axe the most visible components of CE0 pay and are iuider a close public scnitiny, they are expected to be most sensitive to govenunent disclosiue n~Jes.*~ This siiggests that changes in direct pay are more indicative of the total incentive change in response to the new regdation.

Thoiigh a shift of pay striictiue may occiir in response to disclosixe regiilations. changes in the most visible components of pay are expected t o coincide with the total chcange in incent ive strength. Finally, 1 briefly look a t the estimates of the 6xed pay component. As shown in Table 1.2 throiigh 1.4, the fixeci component increased in 1994 for ali samples. The absoltite changes can be directly trnnsformed into percentage changes so as to minirnize the effect of firm size on the estirnates. The resiilts are m L d : there is no particidar pattern

for the difference either between Compiistat firms and non-Çompustat h m s or between 27Given a comparable si& of wealthTbenefits h m options and stock ownership are les visible than direct payments. In fact, many Canadian h s note in proxy statements that the information on directors' (incliiding the CE0 in most cases) stock shares is beyond the knowledge of the finri and is provided by directors thernselves. &O, many firms claim that they do not know how to value execirtive options.

Canadian tirms and US fkm. The general increase in the h e d pay cornponent for a.ll samples seems to siiggest a cornmon trend of changes in the level of C E 0 compensation. Since the model described in the last section does not give an imambiguous prediction to

fixed pay, no conclusion can be drawn with these estimates abolit the effect of disclosiire on the fixed component of CE0 pay.

1.5 Conclusion Discussions of the regulatory impact of execiitive compensation disclosiire on incentive contracts have recently focined on political forces that, explicitly or implicitly, impose conçtraints on the firm's compensation practice and adversely affect the effectiveness of managerial pay schemes. Little has been done, however. to examine the regulatory impact on the pay-performance relation of the piiblic scriitiny of the pay system and managerid behavior. Analyzing an agency model with a shareholder-director-manager stnict lire, this chapter has shown that governrnent disclosiire rides have a two-fold effect on the optimal incentive contract. WXle political forces due to piiblic disapproval of high payments suppress the pay-performance sensitivity, shareholders' scriitiny enhances the effectiveness of the pay system and strengthens the pay-performance Unkage. -4s well. pay envy among managers encoirages them to accept a more risky pay contract and also helps the firm to tie compensation more closely with performance.

Examining C E 0 compensation for 384 large Canadian firms and 666 large US firms before and after the introduction of the new Ontaxio disclosiire regdation, this paper obtains two empirical residts. First, following the introduction of disclosiue regdation in Ontario, the pay-performance sensitivity increased for al1 Canadian finns, but increased siibstantially more in those firms which were not already siibject to the US disclosilre reg~dations.Second, while the pay-performance sensitivity increased less at US firrns than

at Canadian firms that were unaffectecl by the US regtdations, there was Little difference between US frm.s and other Canadian firms that were already boimd by the US regulations. These findings are consistent with a positive impact of regdatory disclosiue on the pay-performance linkage. They confirm the widely-held belief of legislators and piiblic cornmentators: with disclosure, increased piiblic scrutiny of compensation decisions made by corporate directors further aLigns managerial pay with corporate performance. Sorce caveats: Post-disclosiue sensitivity increased by more than 200 percent for all sarnples. This change in sensitivity seems to be large. As weU, the significance level is generdy low for the estimates with total pay. A possible explanation is that the post-disclosire period is too short, which affected both the magnitude and efficiency of the estimates. In particidax, because of the short post-disclosiire period, the evidence obtained with the ciment data may not reflect the long-term effects of goverment disclosiire rules. It is possible that the downward effect of political constraints is not evident at the early stage of disclosiue, becaiise piiblic pressures take time to biuld and to exert an influence on firrns' compensation practices. Therefore, the residts of this stiidy do not

necessarily conflict wit h t hose of earlier stiidies t hat sriggest a negative disclosrue effect on managerial incentive strength (e.g., Jensen and Miirphy, 1990a; and Joskow. Rose. and Shepard, 1993). The difference between the Ontario and US disclosure reqtiirements may also play a role. Compared with the US disclosiire regidation enacted in the 1930s, the new Ontario regdation (and dso the revised US regdation) added reqiurements that reved aspects of a h ' s compensation policy in addition t o the amoimts actiially paid. This change would certainly strengthen the influence of piiblic scnitiny of h m ' s execiitive compensation syst ems.

Table 1 . 1 Selected Statistics -

-

-

-

-

Canndian Fims US Firms

The Tolal Sumplc

Compustat Firms

Non-Compusiat Finns

Mcdian Mcan Obs.

Mcdian Merin Obs.

Median Merin Obs.

The Pre-Disclosure Period: 1991-3993

ÇEO Pav (Sthousand)

Salory+Bonus Total Pay

Finn Vnriables (SrniIlion) Assets Sales Market Value Shareholder Rctiim The Post-Disclosure Period: 1994

ÇEO Pnv (Sthousrmd) Salaryt Bonus Total Pay

Assels

Sales Market Value Shareholder Reiurn Note: AI1 variablçs rire annual and in 1991 US dolltirs. Coriipustut liniis arc thosc ihai appcnr in th;: 1991 C'OMPUSTJTda!a file.

Median Mean

Obs.

Table 1.2 Changes in Pay Performance Sensitivity (Canadian firms) -.

Dependent Variables

Independent Variables Salary+Bonus

Total Pay

Shareholder Wealth (Shareholder Wea1th)-, YEAR94x(SharehoIder Wealth) YEAR94x(Shareholder Wealth).,

O. 106~ IO-' (3.20)

Observations

1,309 - -

-

-

-

-

-

. .- -

--

-

Note: Individual (fixed) effects are controlled. t-statistics are in parentheses. YEAR94 is a dummy variable for fiscal year 1994.

Table 1.3 Changes in Pay Performance Sensitivity : Compustat Firms vs. non-Compustat Firms Dependent Variables Independent Variables SaIary+Bonus

Total Pay

Y EAR94

95.48 (3.63)

YEAR94x NONCOMPUSTAT

-14.80 (-0.23)

Shareholder Wealth

0.079 x 1O-J (2-35)

(Shareholder Wea1th)-,

0.043 x 1O-' ( 1.20)

NONCOMPUSTATx(Sharehoider Wealth)

-0.046 x 1 O" (-0.45)

NONCOMPUSTATx(Shareho1der Wealth)_,

0.0 18x 1W3 (O. 13)

YEAR94x(Shareholder Wealth)

0.243 x 1O-3 (2.14)

Y EAR94x(Shareholder Wea1î.h)-,

0.1 10x IO-) ( 1 .94)

Y EAR94xNONCOMPUSTATx(Shareholder Wealth)

0.295~IO-^ (0.82)

Y EAR94x NONCOMPUSTATx(Shareho1der Wea1th)-,

0.131~10-~ (0.66)

RZ Observations

0.0635 1,209

Note: Individual (fixed) effects are controlled. t-statistics are in parentheses. YEAR94 is a dummy variable for fiscal year 1994. NONCOMPUSTAT is a durnmy variable that equals one for Canadian firms that did not appear in the 1991 COMPUSTATdata file.

Table 1.4 Changes in Pay Performance Sensitivity : US Firms vs. Compustat and non-Compustat Canadian firms Dependent Variabtes Independent Variables

YEAR94xNONCOMPUSTAT

Salary+Bonus

TotaI Pay

205.14 (8.26)

222.78 (4.03)

-1 11.62 (-3.0 1 )

- 127.30

-(-4.1 169.60 0)

(- 1-87)

-142.10 (- 1.35)

Shareholder Wealth

0.03 1 x 1 0 ' ~ (3.40)

0.047 x 1 (2.29)

(Shareholder Wealth).,

0.022~1O" (2.9 1 )

1O" 0.053~ (3.10)

COMPUSTATx(Shareho1derWealth)

0.024~1O-' (0.65)

0.032~1 0m3 (0.59)

0.016x 10-3 (0.42)

-0.0 1 o x 104 (-0.17)

NONCOMPUSTATx(Shareho1der Wealth)

O-' -0.0 1 5 1~ (-0.27)

-0.0 14%1om3 (-0.1O)

NONCOMPUSTATx(Shareho1der Wealtfi).,

0.014~lC3 (O. 18)

0.008%10" (0.04)

YEAR94x(ShareholderWealth)

0.084~1O" (3.78)

0 . 2 0 0 ~1O" (4.03)

Y EAR94 x(Shareho1der Wealth).,

0.125~IO4 (5.83)

0.200 x 1 (4.20)

Y EAR94xCOMPUSTATx(Shareholder Wealth)

-0.0 IO^ 104 (-0.10)

0.043 x IOJ (0.24)

YEAR94xCOMPUSTATx(Shareholder Wealth).,

-0.044~1O-' (-0.70)

-0.090~1O'3 (-0.92)

YEAR94xNONCOMPUSTATx(ShareholderWealth)

0.466~1O-' (2.45)

0.337~10" (0.66)

YEAR94xNONCOMPUSTATx(Shareholder Wealth)-,

0.139~10" ( 1-29)

0.042 x 1O" (O. 14)

R2 Observations

0.09 15

0.0447

3.342

3.342

Note: Individual (fixed) effects are controlled. t-statistics are in parentheses. YEAR94 is a dummy variable for fiscai year 1994. COMPUSTAT is a dummy variable that equals one for Canadian fims that appear in the 199 1COMPUSTATdata file.

Appendix A. Proof of Proposition 1 The Lagrange of the optimization problem for the firm is

where E ( U ) = - exp [-&

+ Be - $ke2)+ $~?a'fl~], and XI

and A? are rniiltipliers for

the constraints. The first order conditions are -1-

-B- @ ( 0

O - ke

-

w-a!

= 0.

-.)

and

-&E(U)p=O,

E(U) - Ur = 0.

Jointiy solving these conditions yields

(Al) The first term is the pre-disclosiue pay-performance sensitivity,,& = *.

The second

term characterizes the effect of political pressures. Becaiise t9 > O, to prove fi < Bo we need t o show accomplished by contradiction. Suppose there is

means

W

-a

+ p'P2

W - a + ,oo2/3* 5 O.

& or P(l + k p o 2 ) 2 1. RecaUing ,6' = ke, we have

> O. This is From ( A l ) this

-

W-cr+pa3$

2 W+(i-fi)e-a = W + E(Y - W ) .

For the firm to accept the contract, there miist be E ( Y - UT) 2 O. CI/ - 4

+

Then there is

2 W > O. This contradicts the presiunption. Therefore. 4
O sttch that the soliition of m and ,3 deterrnined by

(AM) and (A15) satisfy (A17) and (A18). Solve (A15) for m and let it eqiial m*. Then.

Equation (A14), (-415): (A19) and E ( U ) = Urjointly determine the soliition of a ,

8,

f, and m. They are the first-best monitoring solution as long as ( A U ) and (A18) are satisfied. With the same argument as that in the proof of Proposition 5, they are also the solution to (A16). In (A17) and ( A B ) , while all partial effects of each influentid factor on p and m are qualitatively the same as those of Proposition 1 through 5: the total disclosrire impact is ambiguous, depending on the relative strength of different partial effects.

Further, letting E ( U ) = Urgives

Cornparing (A20) with t h e pre-disclositre solution, (A12), and the change of P. the resiilt is consistent with Proposition 1 throiigh 5 in regard of the partial e f k t of different factors on a. The total impact is again ambiguous.

Findy, I check the change in the total expected managerial pay. From the participation constraint for the pre-disclosiue and post-disclosiire model, there is

AE(1V)

=

1

-- log P

1 (Uru+ T ) + [ ( 1 + /-#p' 2k

-

&]

where mo and Po are the pre-disclositre solution defined by (A5). CVhen T = 0, obvioiisly,

sign[AE(W)]= sign[P - Po]. When T > 0, however, this relation may not hold. which is consistent with the residts siunmarized in Section 1.3.5.

Q.E.D.

References Bok, Derek, The Cost of Talent; Hou Executives and Professionals are Paid and How it Aflects Amen'ca, 1993. Bogus, Car1 T., "Excessive Execiitive Compensation and the Failiire of Corporate Democracy," Buffalo Law Review, vol. 41, Winter 1993: 1-83. Brownsterin, Andrew R. and Morris J Panner, "Who Shoidd Set C E 0 Pay? The Press*? Congress? Shareholders?" Haniard Business Review, May- Jime 1992, 28-35. Crystal, Craef S., In Search of Ezcess: The Overcompensation of American Executives, New York: W.W. Norton, 1991.

REFERENCES

53

Dirnma, William, "A Peek Into the Boss's Pocket Book," Business Quaderly. Aiitiunn 1994, 41-45. Dnicker, Peter F., "Reform Executive Pay or Congres WLII," The Wall Street Journal. .4pril 24, 1984. Editorial, "Time for fider s a i q disclosiue," The Financial Post, May 12_ 1993. Garen, John E., "Execiitive Compensation and Principal-Agent Theory," Jovnlal of Political Economy, 102, 1994, 1175-1199. Gibbons, Robert and Kevin J . Miirphy, "Optimal Incentive Contracts in the Presence of

Career Concerns: Theory and Evidence," Journal of Political Economy, 100, 1992. 468-505.

Holmstrom, Bengt , "Moral Hazard and Observability," Bell Journal of Economics. 10. 1979, 74-91. Holmstrorn, Bengt and P a d R. Milgrom, "Aggregation and Linearity in the Provision

of Intertemporal Incentives?" Econornetrica, 55, March 1987, 303-328. Hiibbard, R. G. and Darius Palia, "Execiitive Pay and Performance Evidence from the

US Banking 1ndiistr-y." Journal of Financial Econornics, 39' 1995. 105-130. Jensen. hlichael C. and Kevin J - hlilrphy, -Performance Pay and Toph~ianagement Incentives," Journal of Political Economy, 98, I990rr, 225-264. Jensen, hlichael C. and Kevin J. Mruphy, "CE0 Incentives - It's Not How hliich Yoii

Pay, But How," Harvard Business Review, May-Jiine 1990b, 138-149. Johnson, Tracy S., "Pay for Performance: Corporate Execiitive Compensation in the 1990s," Delaware Journal of Corporate Law, 20, 1995, 183-224. Joskow, Paul, Nancy Rose, and Andrea Shepard, "Regdatory Constraints on C E 0 Compensation," with discussions, Brookings Papers: M~c~oeconomics 1993. 1-69. Kandel, Eiigene and Edward P. Lazear, "Peer Pressiire and Partnerships," Journal of Political Economy, 100, 1992, 801-517.

Kaplan, Steven N., "Top Execiitive Rewards and Firm Performance: A Cornparison of Japan and the United States," Journal of Political Economy, 102- 1994. 51û-546. Kanter, Rosabeth Moss, "The At tack on Pay," Haruard Business Reuzew, 65(2): 1957.

60-67. Kato, Takao and Mark Rockel, "Experiences, Credentials, and Compensation in the Japanese and US Managerial Labor Markets: Evidence hom New Micro Data." Journal of the Japanese and International Economics76, 1992, 30-5 1. Kesner, Michael S.: "Execiitive Compensation in the US: A More Objective Evaliiation and Examination," a speech to the National Association of Corporate Directors. December 1991.

Laughren, Floyd, "Statement to the Legislatiire on Disclosiire of Execiitive Compensation," Ontario Securities Commission Bulletin, October 15. 1993, 51045106. Loornis, Carol J., "The Madness of Execiitive Compensation," Fortune Magazine, Jidy 12, 1982, 42-53. hlirrlees, J. A., "The Optimal Striictiire of Incentives and Aiithority within i\n Organization," Bell Journal of Econornics, 7, 1976, 105-31. bfurphy Kevin J., "Top Executives are Worth Every Nickel They Get." Harvard Business Reuzew, hkrch- April 1986, 125-131.

Pak, Charles C., "Toward Reasonable Execiitive Compensation: Oiitcry for Reform and Regdatory Response," Annual Suniey of Amen'can Law, 1994 Voliune, Issiie 4, October 1995, 633-667.

Rogers, David, "GrassRoots Issiie: Congress's Pay Raise, Slated for Next Weeks. Stirs a Spitited Debate," Wall Street Journal, Febriiary 1, 1989.

Rosen, Sherwin, "Contracts and the Market for Execiitives," in Contract Economics, edited by Lars Werin and Hans Wijkander, bfassachiisetts: Basil Blackweil Ltd., 1992.

Smith, Clifford W. and Ross L. Watts, 'The Investment Opportiinity Set of Corporate Financing, Dividend and Compensation Policies," Journal of Financial Econ,omics.

32, 1992, 263-292. Walters, Bruce, Tim Hardin, and James Schick, "Top Executive Compensation: Eqiiity or Excess? Implication for Regaining Arnerican Compet itiveness," .Journal of

Business Ethics, 14, 1995, 227-234. Whitworth, Ralph, "How Lhch Pay 1s Too Miich? Michael Milken's Hiige Compensation Stirs a Heated Debate Aboiit What Execiitives Are Really Worth To Their Companies," L. A. Times,April 9, 1989, 3B.

Whyte, Heather D., "Books to open on execiitives' pay," The Financial Post, April 15. 1991.

Wi tzer, Edward J ., "Remarks at Execiit ive Compensation Disclosiire Insight Conference," Ontario Securities Commission Bulletin. Febniary 4, 1994. 490-491. Wright, Robert J., "Remarks at Seciuities Law Superconference," Ontario Secvrities

Commission Bulletin, May 1, 1993, 1845-1847.

Chapter 2

C E 0 Pay, Firm Size, and Corporate Performance: Evidence frorn Canada Int rodudion In recent years a nimber of empirical stiidies have examined the determinants of managerial pay and its relation to corporate performance. These stiidies fociis on large companies in the United States (e-g., hliirphy, 1985; Coughian and Schmidt. 1985; Jensen and kliirphy, 1990), Japan (Kato and Rockel, 1992; and Kaplan, 1994). and Britain

(Cosh, 1975). It only became possible to condiict such stiidies with Canadian firms after 1993, when ail piiblicly-traded companies in the province of Ontario were reqiured to disclose top execiitives' compensation imder the new Ontario Seciirities Regdation. It is interesting to duciment the Canadian evidence concerning execiitive compensation both becaiise it has not been described in the literatiire and becaiise it yields resiilts that complement those of earlier stiidies. h4ost of the previoiis stiidies examine large US

and Japanese h m , and so little work has been done on smaller firms. Bas& on total assets or sales, the Canadian firms in oiu sarnple are, on average, abolit ten percent as large as the US firms in the widely tised Forbes sarnple. So, examining the Canadian data provides an opportunity to determine whether or not the widely docimented empirical regiilarities regarding managerial pay schemes hold for relatively small B r m as well. The execiitive compensation data used in this stiidy are the first piiblicly available data that cover al1 Çrns whose shares were piiblicly traded on the Toronto Stock Exchange (TSE) in fiscal years 1993 or 1994. Discussion in this paper fociises on the

2.1 INTRODUCTION

57

compensation of chief executive officers. Section 2.3 examines the relat ionship between

C E 0 pay and h m size. It is found that C E 0 eamings from direct compensation increase by about 0.25 percent for every one percent increase of the firrn's total assets or sales. This elasticity is surprisingly close to that identified in previoiis stiidies with large US, Japanese, and UK h s . Table 2.1 summarizes the major findings of this chapter in cornparison with the evidence dociunented in some representative earlier stiidies. I t has been a piizzling phenornenon that the elasticity changes little across time, industries, and countries. Given the siibstantial difference in firm size between the Canadian sample and those of previoiiç stiidies, this finding makes the iinifonnity of the fîrm-size elasticity of

C E 0 compensation even more puzzling: it also holds for miich s m d e r firms. Section 2.4 disciisses the pay-performance relation. With the widely iised Linear sensitivity measiire that linlis changes of C E 0 pay with changes of shareholder wealth. 1 estimate a sensitivity of 0.00018 with C E 0 direct compensation. When C E 0 stock holdings and options are incliided, the sensitivity increases to 0.00355. Siich a sensitivity means that for every $1000 change in shareholder wealth, CEOs7 earnings change by $3.55. This residt is qiialitatively similar to those of earlier stiidies: C E 0 compensation is positively related to firm performance, but the pay-performance lidcage appears to

be weak. As shown in Table 2.1, the sensitivity (based on ~al~aryiboniis or total cornpensation including options and stock holdings) is notably larger than that for US and Japanese firms. Because the pay-performance sensitivity is inversely related to firm size (see Garen, 1994; and Schaefer, l995), this difference reflects the siibstantial difference in fim size bet~veenthe Canadian sample and the US and Japanese samples. Some novel findings are also docilmenteci. First, options and stock onmership tend to play a relatively more important role for CE0 incentives in small &m.And second' while CE0 turnover probability is, in general, negatively related to the firm's stock

3.2 THE DATA

performance, the threat of dismissal is Iess pronounced in small firms.

2.2 The Data The C E 0 compensation data are compiled £kom the firms' corporate pr0.q statements fileci with the Ontario Seciirity Commission and mailed to shareholders. Since October 1993: ail companies publicly traded in Ontario have been reqiiired to disclose the payments to their CE0 and their other four most highly paid execiitive officers.'

The data cover dl companies that siibrnitted corporate proxy statements during October 1993 to Jiùy 1995, with most firms' reporting executive compensation for two conswiitive years. Becaiise firms are reqiured to disclose execiitive compensation for both the ciment

fiscal year and the last two previois years, the sarnple contains execiitive pay data for fiscal year 1991 through 1994. A total of 939 firms are incliided in the sample. among which 778 have at least two years' data (see the appendiw for a list of firms). Thoiigh the sarnple is based on the firms siibject to the Ontario regiùations, most large Canadian companies are piiblicly traded on the Toronto Stock Exchange. Based on the Financial

Post fhn ordering, more than 90 percent of the 300 largest Canadian firms (by either total assets or sales) in fiscal year 1993 were traded on TSE. The compensation data incliide 98 percent of TSElOO h

s and 91 percent of TSE300 &m.So the companies

in our sample have a reasonable representativeness of large Canadian h m s . The disclosure reqiiirements imder the Ontario regdation are ver). simila to the US disclosiire d e s that reqiUre corporations to detail total execiitive compensation in several tables with a standardized f ~ r r n a t .There ~ is a variety of components of executive remuneration. Ciment cash payments are reported, dong with stock options (and stock lExecutives apart From CEOs who earn less than $100,000 (Canadian) a year in salary and bonus are exempt from disclosiire. * S e Laughren (1993)for a brief summary of the disclosure reqiiirements.

2.2 THE DAT4 appreciation rights) , grants of restricted stock imits, long-term incentive plan payoiits, and fiinge benefits? Indirect-pay related earnings reported in pro-

statements include

stock gains from exercising previoiisly rewarded stock options and stock holdings of directors (in most cases a C E 0 is also a director). Becatise the disclosure rides reqillre fmm to provide only the current year's information on execiitive stock gains and directors' stock ownership, the data of these hÿo items are only available for fiscal years ending 1993 and

1994. There is also no information on the relevant stock market price, exercise price and expiration date for options granted in previous fiscal years, which is needed to estimate the Vitliie of option grants. Sol disciissions concerning stock ownership and options will focus on fiscal year 1993 and 1994.

In a broad sense, the components of CE0 compensation c m be classified into Eoir categories: salary,bonus, long-term incentive rewards, and benefits. Salary is the major component of execiitive pay in Canadian h. Bonis is the anniid variable cornponent of remuneration or short-term incentive pay. Long-term incentive rewards contain three i terns: options, restricted shares, and Long-term incentive plan payoiits. Table 2.2 de-

scribes the niunbers of firms that adopted different incentive plans by the end of fiscal year 1994. Anniial bonuses and option gants axe the two most important schemes that

are cornrnonly used, particularly, in large firrns. Among TSE300 firms. about 94% had an annital boniis plan and had iised stock options in compensating their execiitives. Benefits incliide all other payments that can not be adeqiiately reported imder any above pay components, siich as payments for life insurance, contributory pension plan. impiited interest benefits for debt, tax subsidy, car and hoiising allowance, etc. Excliiding partial compensation due to turnover, firm reorganization, fiscal year change, and part-time --

-

-

-

-

3Stock appreciation rights d o w an employee to realize the appreciation in value of a specified nimber of common shares withoiit req~iiringthe employee to make a cash investment in the stock or caiising dilution of the employer's shareholder eqiuty. A long-tenn incentive plan provides compensation intended to serve as incentive for performance to occiu over a period longer than one financiai year.

2.2 THE DATA

service,' the compensation data contain 830 firms for a total of 2472 observations. Financial data are obtained from the Disclosure tape of Financial Post data files. Table 2.3 summarizes selected statistics of CE0 pay and h variables. As s h o m in the table and by Figure 2.1, the distribution of firm size is highly skewed to the right.' For example, the s m p l e mean of fimi sales is more than six times as large as the sample median. The distribution is even more skewed when based on total assets. Because of this featiire of the data, and given the observation fkom earlier stiidies that the responsiveness of execiitive pay to f k n performance changes markedly nith fùm size, the disciission of the pay-performance relation in Section 2.4 will highlight the difference between large 6rms and s m d firm.

In the firms' proxy statements all components of execritive compensation except stock options are reported in values. The value of options shown in Table 2.3 is the s i m of the ex ante value of ciirrently granted options estimated by the Black-Scholes (1973) f o m d a

and stock gains from exercising previoitsly awarded options. Figure 2.2 presents the distribution of C E 0 total pay incliiding options for fiscal years 1993 and 1994. The pay distribution is also skewed, thoiigh not as strong as that of firm size. In these turo years

CEOs earned a total of $708,000 on average while the median CE0 earned $319.000. The statistics of the pay variables show that salary is the most important pay cornponent of direct C E 0 compensation (which excliides options): on average, s a l q constitutes 76% of t o t d direct pay. The next important component is a m i d boniis which contributes about 19% of total direct pay. This is close to what firms often report: anniid bonus is set at roughly one third of base salary6 Stock options have become increasingly "here are some small firms wbich do not have a fidl time chef execiitive office and the service of siich a position is provided by an executive from another Company on a part-time basis imder certain inter-company employment contract. 'The true distribution of firm size is more skewed than that is showm by Figue 1 because, for the convenience of presentation, the horizontal axis is not evenly scded. 6~siially,there is a performance target t hat has to be met for execiitives to receive a bonus. Among

2.2 THE DATA

61

important in executive compensation in recent years, whose value, based on the sample mean, is about 15% of total direct compensation. Figures 2.3 and 2.4 illustrate the relationship between the composition of C E 0 pay

and firm size for fiscal years 1993 and 1994. As shown by Figure 2.3, ail pay components are higher in larger £hm,and there is obviously a strong positive correlation between

firm size and CE0 total compensation. It is interesting to notice that the role of different pay components tend to be different among h

s of different size. This is illiistrated in

Figure 2.4. As firm size increases, salary consistently becomes less important. For the bottom 25% of firms, the percentage of salary and incentive pay (bonus, stock options,

and long-term incentive pay) in total compensation is aboiit 76 and 18 respectively. The percentage becomes 52 and 44 for salary and incentive pay, respectively, for the top 25%

of £inm.Because the portion of long-tem incentive payments is negligible, Figure 2.4 also siiggests that stock options tend to be relatively more important than other components

of incentive pay in smaller firms. Two things need pointing out here about the sample of this chapter relating to that tised in Chapter 1. First, the sarnples are both from the period of 1991-1994. Becaiise the fociis of this chapter is the Canadian evidence of the general relationships between C E 0

pay, firm size and corporate performance, differences in pay schemes between different years are ignored in this ~ h a ~ t e Second; r.~ the sample of this chapter also incliides the

firms that do not have complete information on stock performance which is needed in Chapter 1 where the fociis is the linkage between C E 0 pay and shareholder wealth. the firms with a bonus plan, aboiit 30 percent, each year, did not grant a bonus to their CEO. ' ~ tcan be easily verified that the estimate of the pay-performance sensitivity changes from year to year. Many factors incltiding stock market fluctuations rnay play a role. While the strategy that assumes a constant pay-performance sensitivity over certain period of time may have an iintriviai effect on the test, it is widely adopted in the literatiire of execiitive compensation for simplicity and discussion convenience.

2.3 C E 0 PAY AND FIRM SIZE

2.3 C E 0 Pay and Firm Size Executive Compensation bas attracted considerable piiblic attention and academic interest because of both the magnitude of pay and its relation t o corporate performance.

Early stiidies of managerial pay schemes focus on the determinants of pay level, particdarly, the role of firm size on C E 0 earnings. According to the allocation theory of control, "in a market eqiiilibrium, the most talented execiitives occupy top positions in the largest firms,where the marginal prodiictivity of their actions is geatly magnifieci over the many people below them to whorn they are linked." (Rosen, 1992, p182) This provides the theoretical groimd for a positive relationship between execittive pay and firm size. Evidence has b e n reported that iuianimously supports a strong positive pay-size relation (see, e-g., Roberts, 1956; Cosh, 1975; hIiirphy, 1985; and Kostiuk, 1989).

1use the widely employed elasticity specification to examine this relationship, which takes the following form:

where siibscript t denotes fiscal year. To be consistent with previoiis stiidies, C E 0 total direct compensation excliiding stock options is iised as the dependent variable. Firm size is iiçually proxied by either total assets and sales, and so, both are examineci in regressions. Firm-size elasticity of C E 0 pay is estimated by b. Retiun denotes annual rates of retiirn. The effect of performance on C E 0 pay is captiued by semi-elasticity c. To take into account the debate on the choice of firm performance measiues in siich a test, three often discussed rates of return axe iised; the accounting retiini on total assets, retiini on equity, and the market return on common stock.$ 8~ccoiuitingretirn on assets (ROA), retiini on eqiiity

(RTS)are defined as foilows:

(ROE),and market

retirn to common stock

2.3 CE0 PAY AND FTRn/I SIZE Table 2.4 presents the estirnates of fim size elasticity of C E 0 total pay. Coliimnç 1 to 3 of Panel A report the OLS estimates of the total assets elasticity and Columns 4 to

6 report the estirnates of sales elasticity. The estimate is very robtist to the size variable

and to the choice of r e t i m Mnables, indicating ~manimouslyan elasticity of about 0.25. For every ten percent increase in either total assets or sales, the total compensation paid

to the average C E 0 increases by 2.5 percent. This resiilt is very close to the finding of many previoiis stiidies that identiS a firm size elasticity between 0.2 to 0.3 for large USI

UK, and Japanese f i r m ~ .To ~ accoimt for a possible difference in the elasticity between large fhx and srnall firms, the mode1 was reestirnateù by dividing the total sample into large firms and s m d firm according to total assets or sales. Biit little difference is foimd in the elasticity between small firms and large firms. After reviewing the evidence doc~imentedby previoiis stiidies, Rosen (l992. p 186) concliides that 'The relative iiniformity of the elasticity of execiitive pay with respect to scale across h m s , indiistries, coiintries, and periods of time is notable and ptuzling

becaiise the technology that siistains control and scale shoidd vary across these disparate

ROA =

ROE =

RTS =

after-tau income to total assets (toal assets + total assests- l)/2 '

after-tax income to common equity (comrnon eqiuty common eqiiity, 1 )/2 '

+

stock price - stock price- 1 + dividends stock price-1

Siibscript '-1' denotes previoiis-year valiles. Following Leonard (1990), after-tax income to total assets is c d c d a t e d as "net income i interest expenses x (1-tax rate)", and after-tau income to comrnon equity is calcidated as "net income - minority interest income - preferred dividends paid." 'For example, with CE0 salary and bonus data for 73 large US corporations over the 1969-1981 period, Kostiuk (1989) obtains an estimate of sales elasticity of abolit 0.24. Exarnining a Forbes sample over the 1970 through 1990 period, Joskow and Rose (1994) estimate a sales elasticity of C E 0 total compensation of about 0.28. Baker, Jensen, and Murphy (1988) give a simmary of sales elasticity changes over time and cross firm for large US cornpanies. Similar residts are aIso reported for UK and Japanese firms (see Cosh, 1975; and Kaplan, 1994).

2.3 C E 0 PAY AND FIRn/I SUE units of cornparison." Given that the average fmn size of the Canadian sarnple is subs t a n t i d y s m d e r than those of other studies, the above resiilts make the iuiiformity of the elasticity even more plizzling: it also holds for miich smaller firms. Panel B presents the regressions with fkn hed-effects being controlled. The estimates are close to the OLS estimates though sales elasticity is notably s m d e r . This difference seems to suggest that the sales elasticity is overestimated with OLS estimation. Kowever, becaiise the time series of the data are short and the fit of the fiued-effects regressions is poor, it is difficdt to draw a conclusion with these residts on the difference.

In both Panel A and B, ail coefficients on rates of retiun are positive. Althoiigh the magnitude and significance level of the estimate vary among the regressions with different retirn variables, the sign of the semi-elasticity of CE0 pay with respect to 6rm retum is robiist, consistently indicating a positive relationship bet-ween C E 0 pay and k m performance. I'U leave detailed discussions of the pay-performance relation to the next section where I fociis on the responsiveness of C E 0 eamings to corporate performance.

A firm-size elasticity larger than zero and smaiIer than imity siiggests that C E 0 pay increases with firm size at a decreasing rate. In other words, there exists a negative relationship between h size and C E 0 relative pay. WXle this phenornenon has long

been realized, it has caiight little attention from the empirical stiidies of eseciitive compensation. To obtain a direct estimate of this relation, I also estirnate the mode1 with log(CE0 relative p l ) as the dependent variable. C E 0 relative pay is defineci as the ratio of CE0 total pay over the firm's market value, which gives a reasonable pro.xy of the CEO's relative importance in a hm. Table 2.5 presents the regression residts. As expected, the size elasticity of C E 0 relative pay is negative in all regressions. The estimates indicate that for every 10 percent increase in total assets, C E 0 pay relative to the h m ' s market value decreases by about

2.4 C E 0 PAY AND CORFORATE P E R F O M A N C E

65

5.3 percent. The decreasing rate is surprisingly large. For example, for a fkm with total assets of $10 million and an asset retiun of 5 percent, the relative pay is 0.012;

when the firm's assets rise to $10 billion, the relative pay drops to 0.0003. The public debate and academic stuàies have extensively explored the linkage between execiitive pay

and 6n-nsize, focusing exclusively on the magnitude of top executives' compensation in large companies. Biit the other side story has essentidy been ignoreci: the portion of executive earnings declines dramatically as firms grow. Behind this observation there may be fimdarnental issues related to the structure and efficiency of the firm7sorganization

and the match between managerial cspability and the firm's complexi5. While this paper does not airn to provide an explmation for this observation, it is nevertheles noteworthy

that the relative importance of top exectitives changes inversely with firm size. As argiied in the next section, this fact may offer some insight to the seemingly small magnitiide of the pay-performance sensitivity of managerial pay schemes.

2.4 CE0 Pay and Corporate Performance The relationship between execiitive pay and corporate performance is the foociis of the recent literatiire of execiitive compensation. The t heoretical literatiire concerning agency problems develops optimal contracts that iink an agent's pay to variations in her performance as a means of aligning the interest of the agent with that of the principal

( s e , e.g., Minlees, 1976; and Holrnstrorn, 1979). Taking the shareholders of corporations to be the principal and the top execiitives to be the agent, a nimber of empirical stiidies have been undertaken which investigate the relationship between executive compensation

and firm performance. There are two main issues concerning the pay-performance relation. One is whether or not executive pay is positively related to corporate performance. If it is, then a firther

2.4 C E 0 PAY AND CORPORATE PERFORMAlVCE

66

issue concerns the extent to which executive pay is tied to corporate performance. Early studies focus on the first issiie but obtain rnixed results ( s e , e-g., Lewellen and Himtsman. 1970; Masson, 1971; and Ciscel and Carroll, 1980). As Miuphy (1985) points out, the reason why early stiidies fail to identdjr an unambigiioiis pay-performance relation may be because the data used is cross-sectional in nature. In addition, the sales-profits debate in early stiidies complicated the issue because of the multicollinearity between size variables and performance variables (see discussions in Rosen, 1992). Since the late 1980s, most stiidies have iised longitudinal data and unanimoiisly identified a positive relationship between corporate performance and managerial pay. h o n g the representative stiidies are hliuphy (l985), Coiighlan and Schmidt (1985), Jensen and bhrphy (1990). Kaplan (1994), and Joskow and Rose (1994).

As it has become a well-established empirical regtdaxity that managerial pay increases with fum performance, more recent stiiùies fociiç on the second issiie and investigate the intensity of the pay-performance relation or incentive strength. Using the Forbes sample of US CEOs' compensation diiring the years 1974-1986 and taking into accoimt cash payrnents, option, stock ownership and dismissal-related factors. Jensen and RLixphy ( 1990) estimate a pay-performance sensitivity of 0.003. This means that C E 0 compen-

sation in the largest US firm increases $3 for each $1000 increase in shareholder vdiie.

The magnitude of this pay-performance sensitivity has been extensively disciissed and debated in the execritive compensation literatiue. While different arguments have been put forth to explain the seemingly small sensitivity (see, e-g., Haiibrich, 1994; Garen,

1994; and Mclaiighlin, 1994), there has b e n no agreement that it necessarily confirms an insensitive pay system for the top exectitives of the largest companies. Wliile disc~issionsin this section nriU also be based on the estimate of the payperformance sensitivity, 1 will fociis on the qualitative relationship between CE0 pay

2.4

CE0 PAY AND CORPORATE PEWORILfANCE

67

and corporate performance. I k s t examine the pay-performance Linkage in C E 0 pay

and then disciiçs managerial incentives associateci wit h execiitive stock ownership and the threat of dismissal.

2.4.1 Pay-Performance Sensitivity There has been little agreement concerning the choices of model specification and firm performance measures for testing the pay-performance relation. Some favor the elasticity or semi-elasticity specification (e.g., Joskow and Rose, 1994; and Brian and

Liebman, 1997), while others prefer the anthmetic specification that links, in a linear fahion, changes of CE0 wealth with changes in shareholder valiie (e-g., Jensen and kliuphy, 1990; and Hubbard and Palia, 1995). Also, some argue that the h m ' s performance shoidd be measured by stock valiie becaiise i t reflects shareholder wealth, but others contend accoiuiting performance measures are usefid because they are less affected by an iuicontrollable market environment and hence are more informative concerning manage-

rial contribution (see Rosen ( 1992) for a review). Thoiigh different model specifications and performance measiires often lead to qiiite different estimates of the performancerelated coefficierit, they ail reveal a statisticaily significant: positive pay-performance relation. Follonring some recent stiidies, 1 iise the arithmetic specification proposed by Jensen and Mtirphy (1990) becaiise of its straightfomard interpretation,

A(CE0 Pay), = a + 0 A (Shareholder WeaIth)t

+ y A (Shareholder Wealth),-,

. (2.2)

The independent variables are changes in shareholder wealth which are defined as the product of the rate of r e t i m on comrnon stock in a year and the h ' s market value a t the

beginning of the year. A lagged term is included in the model to dlow past performance

2.4 CE0 PAY AND CORPORATE PERFORI1fANCE

to also have an influence on current c~mpensation.'~ The dependent vanable is changes

in C E 0 pay. Three measures of C E 0 pay are examineci. They are salary plis bonus. total direct pay, and stock options. Table 2.6 reports parameter estimates for the specification above. Column 1 presents the estimates when the dependent variable is the change in C E 0 salary and boniis. The coefficients on the current year's and the previoiis year's change of shareholder wealth are positive and statistically significant at the one percent level. The total sensitivity is 0.00011 which is the siun of these two coefficients on the performance measiires. That is, for every $1000 change in shareholder wealth, the total salcuy plis boniis paid to the average C E 0 increases by 11 cents. Column 3 reports the resiilt with total direct pay. When stock iinits, long-term incentive plan payoiits, and varioiis benefits are incliided. the sensitivity increases to 18 cents for every $1000 change in shareholder wealth. These nimbers are consistent with many previoiis reports that CEOs' eaniings are tied to corporate performance, but that the linkage appears to be very weak. These estimates, 11 cents and 18 cents per $1000 for salary plus boniis and total pay respectively, are miich larger than the corresponding ones. 2.3 cents and 3.3 cents-

estimated by Jensen and Miirphy (1990) for US firms with the Forbes sample diuing the period of 19741986. There are two factors contributing to this differerice. First, as indicated by Joskow and Rose (1994) and Hall and Liebman (1997), C E 0 compensation in large US firms has become significantly more sensitive to firm performance since the

1970s. With the Forbes sarnple for the later period 1986-1990, Schaefer (1995) estimates a sensitivity about three times as large as that of Jensen and Miu-phy. Second, and more

importantly, the pay-performance sensitivity is inverseiy related to firm size (see Garen, ' ' ~ 0 t h Jensen and Murpliy (1990)and Joskow and Rose (1994)confinn a significant influence of t,he firm's previoiis year's performance on CEO's ciment compensaiton.

1994; and Schaefer, 1995). Because the average hxn size for the Canadian sample is much smaller than that for the US sample siuveyed by Forbes magazine, part of the difference in the pay-performance sensitivity shodd be explained by h - s i z e ciifferences. To capture the effect of firm size on pay-performance sensitivity. the total sample is divided into large firms and s m d firmç at the sample mean of sales and a dummy variable

for large firms is included in the model. The median and mean sales for large firnis are

$773 million and $1.87 billion respectively. For s m d firms: the median and mean are $84 million and $98 million respectiveiy. Coliirnns 2 and 4 report resiilts when a large-firm

diimmy variable is included. As expected, the sensitivity is significantly s m d e r for large

fkms than for s m d firms. The sensitivity from total direct pay is 83 cents for s r n d firrns but only 14 cents for large firms. This residt verifies the negative relationship between

f3-m size and pay-performance sensitivity. It is interesting to notice that the sensitivity of large fimm is miich closer to the total sample sensitivity. This siiggests that large h

s

tend to play a dominating role in the total sarnple estimation. The test with stock options is presented in Coliunn 5 and 6. The sensitivity for the total sample is $1.67 per $1000 change in shareholder wealth which is about 9 tirnes as large as that from direct p . This difference is much larger than that obsenred for

US firms; Jensen and hfiirphy (1990) and Hubbard and Palia (1995) show that the payperformance sensitivity of stock options is three to four times as large as that of total direct p . This observation seems to siiggest that stock options play a larger role in

incentive pay in smaller firms than in larger firms. This conjectiire is siipported by the results in Column 6. While the sensitivity £rom stock options is 0.00151 for large £hm, which is about 11 times as large as that of direct pay, it jiimps to 0.01254 for small firms which becornes 15 times as large as that of total direct pay. This residt is also consistent

with the evidence in Figure 2.4 that CE0 earnings from stock options plus long-term

2.4 C E 0 PAY AND CORPORATE PERFOIZhIANCE

70

incentive pay is relatively more important than bonus payments in srnaller k m s .

2.4.2 C E 0 Stock Ownership The incentive problem exists because of the separation of ownership and management. A natural way to mitigate this problem is to increase CEOs' holdings of their firm's stock shares. Though executive stock ownership is not iisiidy dictated by the board of directors, the h ' s compensation strategy of rewarding stock iinits and options certauily

has a direct impact on execiitive stock holdings. According to Jensen and Murphy (1990)' the incentive intemity of CE0 stock ownership in terms of pay-performance sensitivity is abolit eight times as high as that of direct payments. Examinhg the elasticity of C E 0 pay with respect to stock performance, Hall and Liebman (1997) even concliide that managerial incentives are almost entirely from stock ownership. It has been a consensus in the execiitive compensation literatiire that stock ommership plays the most important role in mo tivating top executives.

The data on CE0 stock onmership are from the firm' reports of stock holdings of members of the board of directors for fiscal years ending 1993 and 1994. Because some CEOs are not directors (thoiigh in most cases they are), these data may slightly overestimate CEOs' stock holdings; i.e. non-director CEOs who are excliided from the sarnple are expected to hold Iess stock shares. Changes of C E 0 wealth frorn stock ownership are perfectly correlated with changes in shareholder wealth. Hence, the percentage of a firm's total shares outstanding that are

owned by CE0 gives a direct estimate of the pay-performance sensitivity associated with stock ownership. Table 2.7 siunmarizes the related statistics of CEOs' stock holdings.

The fUst row shows the percentage for the total sample. The CEOs o a n a n average of 2.89 percent and median of 0.17 percent of their h m ' s stock. In other words, for every

2.4 CE0 PAY AND C O R P O U T E PER.FORn/IANCE

71

$1000 hcrease in the h m ' s stock valueo CEOs' wealth £rom stock ownership increases by

$1.7 when she holds a median percentage of her b ' s stock.

The distribution of the stock holding percentage is also skewed. This reflects the skewed firm size distribution and the fact that CEOs of larger fim hold a smaller portion of their firm's stock. The second and third row of Table 2.7 verifj- the negative relationship between fùni size and C E 0 stock ownership. While CEOs of 6rms larger than the sample median hold an average of 1.49 percent of their fim's stock. those of h m smaller than the sample median hold an average of 4.3 percent. This difference is

miich larger when compared based on the median percentage. Including total cash pa-nents, options. and stock ownership (median). the above residts give a total sensitivity of $3.55 for every $1000 change in shareholder wealth for the total sample. The sensitivity is $2.15 for large firms and $25.57 for srnall firms. A typical e>cplanation for the negative relationship between the sensitivity and fimi size is that it reflects a shift of the pay scheme from incentives to insiirance as a response to a higher output iincertainty of larger fi-

(Garen, 1994). The observation that CEOs'

relative contribution decreases with hm size (Table 2.5) may also partly esplain the negative relationship between the pay-performance sensitivity and firm size. Becaise

CEOs contribute relatively less in larger Enns and because firm performance miist be balanced with total ernployee contribution, a sensi tivity that Links a CEO's indiviciiial pay with the h m ' s aggregate performance is boimd to be smaller in iarger fhns.

2.4.3 CE0 Turnover and Corporate Performance The £inalcomponent of the pay-performance linkage 1 examine is the threat of dismissal. The hypothesis behind the dismissal incentiw scheme is that, a change in a h ' s chef execiitive officer is more likely to occiir after bad years than after good years.

2.4 C E 0 PAY ,4ND CORPORATE PERFORlbIANCE

/2

m

1 use probit regression to test the negative relationship between the probability of CE0 turnover and the firm's stock performance:

"

where the dependent variable is dichotomoiis, having a value of one when a C E 0 turnover occtirs and a value of zero otherwise. The model estimates the latent variable of C E 0 turnover probability

(a

+

Retuml

+ y - R e t ~ u n ~ -which ~ ) is

the standard normal

distribution h c t i o n . The independent vaiables are the firm's market retiirn net of the retiini to the TSE300 index."

The information on C E 0 timover is directly available Erom the firm's p r o - statements. But for the firmç that changed CE0 at least twice (in different years) and hired a nea 930 from the market instead of interna1 promotion, the information for 1991 and 1992 is generally incomplete, because these firms usiially do not reveal previoiis turnover information in their disclosiue dociunent for the current fiscal year. However. becaise the nimber of siich fimis is small and this incomplete information does not came a Sample selection problem, al1 four years' (1991-1994) data are incliided in the test. l 3 C E 0 turnover can occiir a t any time during a year. It is reasonable to link the performance impact on turnover with the CE07sprevious Ml years' service. So a turnover is recorded whenever the C E 0 senres her last full fiscal year. Column 1 in Table 2.8 presents the resdts with the base specification. Consistent "Some studies (e.g., Coughland and Schmidt, 1985; and Jensen and Miuphy, 1990) iise logit regression to test the relationship between management turnover probability and firm performance. Since estirnates in siich a test are quite close between logit and probit rnodels and the choice of the model is risiiaily discretional, 1use probit regression for presentation convenience (it gives t-statistics and R2, consistent with the reports of other estimates). 121 compared the resiilts between tising the firm's stock rettirn and u i n g the net-of-market retiirn. Little difFerence was foiind between the two choices, though the estirnates are slightly irnproved when net-of-market retum is used. 13The estirnates are only slightly different when the data of 1991 and 1992 are excliided.

2.4 C E 0 PAY AND CORPORATE PERFOMANCE

73

with previous studies, the coefficient on the net-of-market return to cornmon stock is negative, confirming that a £km's stock performance and the probability of a change in C E 0 are inversely related. The estimate is marginally significant. The explanatory power of the regression, however, is low. The residt is slightly improved in Coliunn 2 where the previous year's retuni is included. To illustrate the performance effect on C E 0 tiunover probability, consider a firm that has a return eqiiai to TSEJOO index r e t ~ u nfor two conseciitive yews. The turnover probability, calculateci as

(-1.19 - 0.063 x O - 0.013 x O), is

0.117. It rises to 0.1247 when the firm eams a -50 percent net-of-market retint for two consecutive years. Given the siibstantid change in fhrn performance. the corresponding

change in turnover probability is very small. On the other hand, since the definition of C E 0 tiirnover does not distingillsh a disciplinary leave from a non-disciplinary leave (due to normal retirement or job switching), these incentive intensity estirnates are best viewed to be lower boirnds.

It is possible that the threat of dismissal for managers is different among h

s of

different size. With the wide distribution of firm size, the Canadian sample provides a good opportiinity to address this issue. Colimn 3 and 4 present resiilts when a d i m y variable for firms that have a size larger than the sample median is incltidecl in the model.14 The resiilts are iuiexpected but interesting: the impact of firm performance on

C E 0 turnover is more evident in large firms. For small firms, the coefficients on retiun variables are s m d and insignificant, though still negative. The large magnitude and

high significance levei of the coefficients on the dummy variable indicate a much stronger dismissal-performance relation in large firms. Table 2.9 presents some point estimates of the probability of C E 0 tiirnover based on the results of Coliunn 2 and 4 in Table 2.8. 14The number of observations in Column 3 and 4 is reduced because of missing financiai data from some firms that are needed to identify finn size.

2.4 C E 0 PAY AND CORPORATE PERFORMANCE The estimates indicate that the probability of C E 0 turnover in small fums is almost inva,riant to b performance. But for large 6mis,the performance effect is qiiite strong. For example, after a firm e m s a net-of-market r e t i m of -50 percent for two conseciitive years the CE0 is about three times as Iikely to leave the h m as when the firm earns 50 percent above the market retim.

This observation seems inconsistent with the finding of Jensen and Murphy (1990). With the Forbes sample of US k m s they obtain an iipper-boimd estirnate of the dismissal related sensitivity of 52.25 for smail firms but rnerely 5 cents for large h n s per

$1000 change in shaxeholder wealth.15 These estimates imply a miich stronger negative dismissal-performance relation in s m d firms. While it is possible that the effect of fkn performance on C E 0 turnover probability does not change monotonically with

firm size, the large difference between large h s and small Ei.rms in opposite directions is siiiprising.

Because l u g e Canadian firms are more iikely to be piiblicly traded in the United States and then are more iduenced by the US managerid labour market. the residts for Canadian fkm may reflect a generally greater job sectirity in Canadian firms than in US h s .

This conjecture is natural, given that in the Canadian sample abolit 74.5 percent

of large 6rms (larger than the sample median) and 27.1 percent of small firrns were traded in the United States before 1991, and that the estimates for large Canadian firms are

qiiite similar to the US resiilts ( s e , e.g., Table 4 in Jensen and Miirphy). To clarify this issue, the regression of Coliunn 4 in Table 2.8 is estimated, separately, for fimis that 15The iipper-boiind estimate of the dismissal related sensitivity is estimated as the following. The point estirnates of CE0 turnover probability are obtained by assuming the h earns the market rate of retirn for two years versirs when the firm realizes r e t i m 50 percent below the market in two conseciitive years. Assume that the CE0 wiU never work again if dismissed but will work for his fkm until age 66 if not dismissed. Then the expected wealth l o s of the CE0 due to dismissai is the present value of her potential earnings from the year of dismissal to retiring age 66. The àismissal performance sensitiviw is obtained by dividing the CEO's potentid wealth loss by the shareholder loss associated with earning 50% percent below-market r e t i n s for two years.

2.5 CONCLUSION

75

were traded in the United States before 1991 and for those that were not. Thoiigh the difference between large h

s and small h

s does become smaller, the qiialitative result

remains unchanged.

2.5 Conclusion This chapter has provided the first systematic evidence on the relationship arnong

C E 0 compensation, firm size. and corporate performance for Canadian companies. The evidence is consistent with, and large- similar to, the h d i n g s of previous stiidies for

other countries, partic~ilarlythe United States: CE0 pay rises with hm size and compensation is tied to Company performance. The similarity in C E 0 compensation between Canada and the United States is not surprising, because "given the e'rtensive economic (e-g.,trade) and institutional (e.g., corporate, labor iuiion) linkages that have developed between Canada and the United States at both the macro and microeconomic levels, we rnight reasonably expect significant cross-national influences on compensation practices." (Chaykowski and Lewis, 1996, p2).

As the first empirical analysis of execritive compensation for Canadian firms. this chapter has fociised on the evidence parallel to those doclunented by previoiis s tiidies, althoiigh references have Freqiiently been made to eariier findings. There are important issues concerning the Canadian execiitive compensation system imaddressed in this chapter. For example, how are managerial pay schemes in Canada different from those in other countries, particidarly, in the United States? Given a nimber of institutional and market differences between Canada and the United States, the pay systems and their effectiveness are evpected to be different . Becaiise sample heterogeneity iisiially plays an important role in an inter-sample cornparison, s ~ i c han issue can be properly addresseci with more comparable samples.

Table 2.1 A Brief Cornparison of the Evidence -

US. Sales elasticity of CE0 pay

0.282*

-

--

-

--

Japan

Britain

Canada

0.247***

0.26 I ****

0.247

Pay-performance sensitivity with CE0 salary+bonus ( x 1O-') Pay-performance sensitivity with CE0 pay, options, and stock holdings (x 1

-

2.950**

-

3-551

*

Sales elasticiiy of C E 0 total compensation for the Frobes sarnple over the years 1970- 1990 (Joskow and Rose, 1994). ** Pay-performance sensitivity for the Forbes sample over the years 1974-1986 (Jensen and Murphy, 1990). *** Sales elasticity of per director (as a proxy for the president) salary+bonusand pay-performance sensitivity of per director salary+bonus, respectively, for the Japanese companies From the Fortune sample of foreign industrials in 1980. The pay data are from the 1982 and 1984 Yuka Shoken Hokokurho (Kaplan, 1994). **** Sales elasticity of CE0 cash compensation for British firms during the years 1969- 1971. The data are fiom Department of Trade and Industry's standard Company accounts (Cosh. 1975).

Table 2.2 Nurnber of Firms with Different Incentive Plans -

-

-

- -

-

- -

-

--

- --

-

Incentive Plans Total Bonus Al1 firms

TSE300 firms

664 246 -

Options 8 17 247 -

LTIP 56 35

945 262

-

Note: The statistics are fkom the 1991- 1994 sample. LTIP denotes longterm incentive pay including grants of restricted shares and long-tenn incentive plan payouts.

Table 2.3 Summary of Selected Statistics Minimum

Median

Mean

Maximum

Observ.

107 104

1,92 1 659

23 7 32 0.5 10.5

797

168,248 2 1,065 14,708 1,053 10,305 19,400

2,245 2,068 1,240 2,199 2,100 1,888

C E 0 pay variables ($1000)

Salary Bonus Long-term incentive pay (LTIP) Other payments Total direct pay Stock Options Salary/(Total direct pay) (%) Bonus/(Total direct pay) (%) LTIPf(Tota1direct pay) (%) Other/(Total direct pay) (%) Options/(Total direct pay) (%) Firm variables ($million)

Total assets Sales Market vaIue Retum on total assets (%) Return on equity (%) Return to cornmon stock (%)

O. 1 0.1 0.7 - 1,583

-2,404 -97.5

-1.4 5.1

55.7 -

.

--

-

-

- -

-

Note: Al1 variables except stock options are ftom the 1991- 1994 sarnple. Stock options are for 1993 and 1994, which includes values of currently granted options estimated by the Black-scholes (1973) formula and stock gains Frorn exercising previously granted options. LTIP includes gants of restricted shares and long-term incentive plan payouts. Total direct pay includes ail pay components except options.

Table 2.4 Firm Size Elasticity of CE0 Total Pay - --

--

-.

Dependent Variable: log(Tota1 Direct Pay) (Total assets elasticity)

(Sales elasticity)

Independent Variables (1)

(2)

(3

A. OLS Estimation

Intercept log(Totat Assets)

Return on Assets Retum on Equity Return to Common Stock R2

Observations Fixed-effects Estimation

log(Tota1 Assets)

Retum on Assets Return on Equity

-

Retum to Common Stock

-

-

-

Note: The data are from the 199 1 - 1994 ~ample.t-statistics are in parentheses.

(4)

(5)

(6)

Table 2.5

Firm Size Elasticity of C E 0 Relative Pay Dependent Variable: Iog(Tota1 Direct Pay/Firm Value) --

(Total assets elasticity)

(Sales elasticity)

Independent Variables (1

(2)

(3 1

A. OLS Estimation

Intercept

log(Tota1 Assets)

Return on Assets Return on Equity

Return to Common Stock R2

Observations

B. Fixed-effects Estimation log(Tota1 Assets) Iog(Sa1es) Return on Assets Return on Equity

-

-0.195

-

(-4.2)

Return to Common Stock

Note: The data are from the 1991- 1994 sample. t-statistics are in parentheses.

(4)

(5)

(6)

Table 2.6 Pay-Performance Sensitivity (OLSestimation) Dependent Variables

Independent Variables

a(Salary+Bonus)

total Direct Pay)

~(Options)

Intercept LARGE

LARGE x ~(ShareholderWealth) x 1 O-3 a(Shareho1der Wealth) -, x 1 O-'

LARGE x ~(SharehoiderWealth)

-, x 1O-3

R' Observations -

-

-

..

Note: The data are from the 199 1- 1994 sample for regressions with saIary+bonus and total direct pay, and are fiom the 1993-1994 sample for regressions with options. LARGE is a dummy variable that equals one for fims with average sales larger than the sample median. t-statistics are in parentheses.

Table 2.7 C E 0 Stock Ownership (CE0shares)/(TotaI shares outstanding) Minimum Median Mean

Maximum

Num ber of Firms

Observations

Al1 F h s Large Firms Srnall Firms Note: The data are fiom the 1993-1994 sample. Large firms are those that have averages sales larger than the sample median.

Table 2.8 C E 0 Turnover and Firm Performance (Probit estimation) Dependent variable: CE0 turnover Independent Variables

Intercept LARGE Net-of-Market Retum LARGEx(Net-of-Market Return) met-Of-Market Retum)-,

R2 Observations

-

Note: The data are fiom the 1991 1994 sample. The dependent variable equals one when a CE0 tumover occurs and equals zero otherwise. LARGE is a dummy variable that equals one for firms with average sales larger than the sample rnedian. Net-of-Market Retum is the firm's stock return minus the return to TSE300 index. t-statistics are in parentheses.

Table 2.9 Probability of C E 0 Turnover Probability of CE0 Turnover -

Net-of-Market Retum

AI1 Firms

Large Firms

-

-

-

SmalI Firms

.-

Note: The estirnates of the probability are obtained fiom regression (2) and (4) in Table 2.8. Large firms are those that have average sates larger than the sample median.

Figure 2.1 Distribution of Firm Size

5

80

40

20

60

150 250 350 450 560 710 900 1100 1300 1500 1700 1900 21002300 110 200 300 400 500 630 800 1000 1200 1400 1600 t8OO 2000 2200 2400

Average Sales ($million)

Figure 2.2 Distribution of C E 0 Pay

Average CE0 Total Pay ($thousand)

Figure 2.3 Components of CE0 Pay

G

Salary

iBonus

a LTIP+Options a Benefits

Figure 2.4 Structure of C E 0 Pay

a Salary

a LTIP+Options

Bonus

.

1

25

.

m Benefits

t

i

I 50

........

IF..

1

I

I

75

Percentile by Average Sales (%)

100

Appendix: List of Firms 20I20 Financial Corp A.G.F. Management Ltd ABL Canada Inc ACC Telenterprises Ltd AGT Ltd AJ Perron Gold Corp AQM Automotive C o q ARC International Corp AT Plastics Inc ATCO Ltd ATS Automation Tooling Aber Resources Ltd Abitibi-Price Inc Accord Financial Corp Acklands Ltd A c m a International Inc Adex Mining Corp Adrian Resowces Ltd Advanced Gravis Computer Tech Advanced Material Res Ltd Adventure Electronics Inc Agnico-Eagle Mines Ltd Agra Industries Ltd Ainsworth Lumber Co Ltd Air Canada Airboss of Amenca Corp Akita DriIling Ltd Alarmforce Industries h c Alberta Energy Co Ltd AIberta Natural Gas Co Ltd Alcan Aluminium Ltd Alert Care Corp Algo Group Inc Algoma Central Corp Algoma SteeI Inc Algonquin Mercantile Corp Allelix Biopharmaceuticals Alliance ~ommunicationsCorp Allied Northern Res Ltd Alpine Oil Services Corp Alta Genetics Inc American Barrick Resources Corp American Econornic Corp American Leduc Petr Ltd Amoco Canada Petr Co Ltd Anchor Lamina Inc Anderson Exploration Ltd Andres Wines Ltd Angio-Canadian Tel Co Angoss Software Corp Applied Inventions Management

Arbor Capital Inc Arbor Mernorial Services Archer Resources Ltd Argus Corp Ltd Armbro Enterprises Inc Arrowlink Corp Ashgrove Energy Ltd Ashton Mining of Canada Inc Asia Minerals Corp Astra Finance Ltd Astral Communications Inc Atcor Resources Ltd Atlanta Gold Corp Atlantis Communications Inc Atlantis Resources Ltd Atlantis Submarines (Internation Aur Resources Inc Aurizon Mines Ltd Avco Fin Serv Canada Ltd Avenor Inc BCBancorp B.C.Pacific Capital Corp BC Gas Inc BC Sugar BC Telecom Inc BCE Inc BCE Mobile Communication Inc BPI Financial Corp Ballard Power Systems Inc Ballistic Energy Corp Band-Ore Resources Ltd Banister Inc Bank of Montreal Bank of Nova Scotia Barrick Gold Corp Barrington Petr Ltd Baton Broadcating Inc Baîtery Technologies Inc Battle Creek Developments Ltd Baytex Energy Ltd Beamscope Canada Inc Bearing Power (Canada) Ltd Beau Canada Exploration Ltd Beaufield Consld Res Inc Bell Canada Berna GoId Corp Bethlehem Resources Corp Big V Pharmacies Co Ltd Biochern Pharrna Inc Biomira Inc Bionaire Inc

Bioniche Inc Biovail Corp International Biron Bay Resources Ltd Black Swan Gold Mines Ltd B Iake River Exploration Ltd B lue Range Res Corp Bombardier Inc Bonar Inc Bone Health Inc Boomerang Resources Inc Bovar Inc Bow Flex Inc Bow Valley Energy Inc Bracknell Corp Brarnalea Ltd Brampton Brick Ltd Brandevor Enterprises Ltd Brandselite International Corp Brascan Ltd Brenda Mines Ltd Brick Brewing Co Bruncor Inc Brunswick Mining & Smelt Corp C-MAC Industries Inc CAE Inc CCL Industries Inc CFS Group Inc CGC Inc CG1 Group Inc CHC Helicopter C o q CML Industries Ltd COGECO Inc CS Resources Ltd CSA Management Ltd CT Finanical Services Inc Cabre Exploration Ltd Cadillac Fairview Corp Ltd Caledonia Mining Corp Calian Technology Ltd Call-Net Enterprise Inc CarnVec Corp Carnbior Inc Cambridge Shopping Centres Ltd Carnco Inc Camdev Corp Cameco Corp Campbell Resources Inc Canada Malting Co Ltd Canada Southem Petr Ltd Canada Tnistco Mortgage Co Canada Tungsten Inc

Canadian 88 Energy Corp Cinar Films Inc Czar Resources Ltd Canadian Angus Resources Ltd Cindy Mae Res Inc D.A.Stuart Ltd Canadian Bank Note Co Ltd Cineplex Odeon Corp DMR Group Inc Canadian Conquest Exploration C inmm L td Dalmys (Canada) Ltd Cintech Tele-Management System Davis Distributing Ltd Canadian Foremost Ltd Circo Craft Co Inc Dayton Mining Corp Canadian Fracmaster Ltd Dedicated Technologies Corp Canadian Frobisher Resources Ltd Clairvest Group Inc Denbridge Capital Corp Canadian Imp Bank of Commerce Claude Resources Inc Clearly Canadian Beverage Corp Denison Mines Ltd Canadian Manoir Ind Ltd Clearport Petroleums Ltd Canadian Marconi Co DeprenyI Animal Health Inc Canadian Natural Resources Ltd Cliff Resources Corp Deprenyl Research Ltd Derlan Industries Ltd Canadian Newscope Resources Ltd Co-Steel Inc Cobi Foods Inc Develcon Canadian ~cciden'talPet- Ltd Coca-cola Beverages Ltd Devrar~Petroleum Ltd Canadian Pacific Ltd Cogas Energy Ltd Devtek Corp Canadian Pioneer Energy Inc Cogeco Cable Inc Dia Met Minerais Ltd Canadian Satellite Comm Inc Cognos Inc Dickenson Mines Ltd Canadian Tire Corp Ltd Colortech Corp Canadian Utilities Ltd Difhcto Ltd Cominco Fertilizers Ltd Canadian Westem Bank Discovery West Corp Cominco Ltd Canadian Westem Natural Gas Dofasco Inc Cominco Resources International Doman Industries Ltd Canam Manac Group Inc CompAS Electronics Inc Canbra Foods Ltd Domco Industries Ltd Cornputalog Ltd Canfor Corp Dominion Textile Inc Cornputer Brokers of Canada Inc Dominion and Anglo Invt Corp Canhorn Mining Corp Comstate Resources Ltd Canlan Investment Corp Domtar Inc Confederation Life Insurance Co Donohue Inc Canrise Resources Ltd Conpak Seafoods Inc Canstar Sports Inc Dore1 Industries Inc Consld Canna Corp Canuc Resources Inc Dorset Exploration Ltd Consld Eurocan Ventures Ltd Canwest Global Comm Corp Draxis HeaIth Inc Consld Firstfund Capital Corp Capilano International Inc Dreco Energy Service Ltd Consld Mercantile Corp Cam Operations Ltd DuPont Canada Inc Consld NRD Resources Ltd Carena Developments Ltd Dundee Bancorp Inc Consld Ramrod Gold Corp Caribbean Utitities Co Ltd Dundee-PaIIiser Resources Inc Consoltex Group Inc Caribgold Resources Inc Dusa Pharmaceuticals Inc Consumers Gas Carmanah Resources Ltd Dylex Ltd Cascades Inc Consumers Packaging Inc Dynacare Inc Continental P h m a Cryosan Cassidy's Ltd E-L Financial Corp Ltd Continental Precious Minerals Castle Capital Inc Eagle Precision Tech Inc Cathedra1 Gold Corp Contrans Corp Economic Invt Trust Ltd Celanese Canada Inc Conwest Exploration Co Ltd Eden Roc Mineral Corp Celltech Media Inc Corby Distilleries Ltd Edper Enterprices Ltd Centra Gas Ontario Inc Corel Corp Eicon Technology Corp Cornucopia Resources Ltd Central Capital Corp Elan Energy Inc Corpome Foods Ltd Chai-Na-Ta Ginseng Prods Ltd Eldorado Corp Ltd Coscan Development Corp Champion Gold Resources Inc Electrohome Ltd Champion Road Machinery Con Corp Emco Ltd Chancellor Energy Res Inc Counsel Corp Encal Energy Ltd Channel Resl Ltd Crestar Energy Inc Enerfiex Systems Ltd Crestbrook Forest Ind Ltd Chase Resource Corp Enerplus Resources Corp Crown Butte Resources Ltd Chateau Stores of Canada Ltd Enserv Corp Chauvco Resources Ltd Crown Life Insurance Co Environmentai Techs Internationa Crown Resources Corp Cheni Gold Mines Inc Enviropro International Inc Chesbar Resources Inc Crownx Inc Epic Data International Inc Chieftain International Inc Cube Energy Corp Equisure Fin Network Inc Chum Ltd Current Technology Corp Equity Transfer Services Inc Cimarron Petroleum Ltd Cycomrn International Inc Euro-Nevada Mining Corp Ltd

Evans Health Group Ltd Exall Resources Ltd ExceI Energy Inc Exco Technologies Ltd Export Tyre Holding Co Extendicare Inc FCA International Ltd FPI Ltd FT Capital Ltd Fahnestock Viner Iioldgs Inc Fairfax Fin Holdgs Ltd Falcon Point Resources Ltd FaIvo Corp Federal Industries Ltd Fiducie Desjardins Fiming Ltd Firan Corp Fire River Gold Corp First Marathon Inc Firstservice Corp Flanagan McAdam Resources Inc FIeet Aerospace Corp Fonorola Inc Ford Motor Co of Canada Ltd Foremost Industries Inc Fortis Inc Forzani Group Ltd Four Seasons Hotels Inc Franco-Nevada Mining Corp Ltd Fundy Cable Ltd Future Shop Ltd G.E.F. Management Ltd G.T.C. Transcontinental Group GSW Inc Galtaco Inc Gandalf Technologies Inc Gardiner Oil and Gas Ltd Gaz Metropolitain Inc Geac Computer Corp Ltd Gendis Inc Genecan Financial Corp General Leaseholds Ltd General Motors Accept Corp Gemum Corp Gentra Inc Gesco Industries Inc Gibraltar Mines Ltd Glentel Inc Global Eiection Sys Inc Global Govt Plus Fund Ltd Glyko Biomedical Ltd Go Vacations Inc Gold Reserve Corp Goldcorp Inc Golden Eagle Capital Corp Golden Knight Resources Inc

Golden Rule Resources Goldfarb Corporation Goldhwiter Explorations Inc Goran Capital Inc GowIing, Strathy & Henderson Grad & Walker Energy Corp Grand Oakes Resources Corp Graph/Max Inc Great Lakes Minerals Great Lakes Power Inc Great-West Life Assurance Co Greenstone Resources Ltd Greyhound Lines of Can Ltd Greyvest Fin Services Inc Groupe Laperriere & Veneau. Guardian Capital Group Ltd Guilievin International Inc GuIf Canada Resources Ltd Gulf International Minerais Ltd Gu1fsîrearn Resowces Canada Ltd Gwil Industries Inc H Paulin & Co Ltd H. O. Financial Ltd HCO Energy Ltd HEC Investrnents Ltd Haley Industries Ltd Halozone Technologies Inc Hammerson Canada Inc Harnmond Manufachiring Co Ltd Harbour Petr Co Ltd Hard Suits Inc Harris Steel Group tnc Harrowston Inc Hartco Enterprise Inc Harte Resources Corp Hawker Siddeley Canada Inc Hayes-Dana Inc Hees International Bancorp Inc Helix Circuits Inc Hemlo Gold Mines Inc Hemosol Inc Highridge Exploration Ltd Hillcrest Resources Ltd Hillsborough Resources Ltd Hol-Lac Gold Mines Ltd Hollinger Inc Holmer Gold Mines Ltd Home Capital Group Inc Home Oil Co Ltd Home Products Inc Hongkong Bank of Canada Horsham Corp Household Fin Corp Ltd Hudson Bay Diecasting Ltd Hudson's Bay Co Hy & Zel's Inc

Hyal Pharmaceutical Corp Xycrofl Resources & Devei Corp 1.S.G. Technologies Inc [PLEnergy Inc :PSCO Inc S M Information Systems Imasco Ltd :mperial Life Imperia1 Oil Ltd :mperiai Parking Ltd .mutecCorp In-Flight Phone Canada Inc 1x0Ltd 'ndustra Service Corp 'ntensity Resources Ltd nter-City Products Corp ntera Info Techs Corp nteraction Resources Ltd nterlock Consld Enterp Inc Intermetco Ltd International Aqua Foods Ltd International CoIin Energy Corp International Contour Tech Inc International Curator Resources International Datacasting Corp International Dunraine Ltd International Forest Products International Gold Resources International Innopac Inc International Inter-Link Inc International Mahogany Corp International Musto Exploration International Northair Mines Ltd International Pastel Food Corp International Petroleum Corp International Potter Distilling International Pursuit Corp International Semi-Tech Micro international UNP Holdings Ltd International Venfact Inc Interprovincial Pipe Line Sys Interquest Technologies Inc Intertape Polymer Group h c Intrawest Corp Inverness Petroleum Ltd Investors Group Inc Invin Toy Ltd Island Tel Co Ltd Ivaco Inc Jannock Ltd Jean Coutu Group John Forsyth Co Inc Jordan Petroleum Ltd Jordex Resources Inc Joss Energy Ltd Joumey's End Corp

Joute1 Resources Ltd KaufeI Group Ltd Kelsey's International Inc Kerr Addison Mines Ltd Kinross GoId Corp Koivox Communications Inc LSI Logic Corp of Canada Inc Labatt, John Ltd Lafarge Canada Inc Laidlaw Inc Laird Group Inc Lakewood Energy Inc Lander Downs Larr Capital Corp Laurentian Bank of Canada Le Groupe Videotron Ltee Leadley, Gunning & Culp Internat Leon's Furniture Ltd Liberian Iron Ore Ltd Linarnar Corp Lincoln ~ a p k Corp I Liquidation WorId Inc ive Entert of Canada Inc Loblaw Companies Ltd Loewen Group Inc Logistec Corp London Insurance Group Inc Louvem Mines Inc Luscar Oil and Gas Ltd Lynx Energy Services Corp Lytton Minerals Ltd M-Corp Inc MAXX Inc MDC Corp MDS Health Group Ltd MICC Investments Ltd MPACT Immedia Corp MSV Resources Inc MVP Capital Corp MacDonald, Dettwiler & Ass t t d MacMillan Bloedel Ltd Mackenzie Financiai Corp Madison Ave Sports Ntwk Ltd Magna International lnc Malahide Petroleum Corp MaIette Inc Mannville Oil & Gas Ltd MapIe Leaf Foods Inc Maritime Elec Co Ltd Maritime Telgh & Tel Co Ltd Mark Resources h c Mark's Work Wearhouse Ltd Markborough Properties Inc MarIeau, Lemire Inc MarshaIl Minerals Corp Marshall Steel Ltd

Maude Lake Gold Mines Ltd M a u Petroleum Ltd McGraw-Hill Reyerson Ltd McNellen Resources Inc Melcor Developments Ltd Memotec communications Inc Merfin Hygienic Prods Ltd Meridian Technologies Inc Metail Mining Corp Methanex Corp Metro-Richelieu Micro Tempus Inc Microbix Biosystems Inc Midland Walwyn Inc Milltronics Ltd Minera Rayrock Inc Mining & Allied Supplies (Cm Miramar Mining Corp Mitel Corp Mobile Computing C o q Modatech Systems Inc Moffat Communications Ltd Molson Breweries Molson Companies Ltd Monarch Development Corp Montreal Trustco Inc Moore Corp Ltd Morden & Helwig Group Inc Morgan Financial Corp Morgan Hydrocarbons Inc Momson Petroleums Ltd Mortgage Insurance Co of Canada Mosaid TechnoIogies Inc Motion Works ~ 6 Ï - p Mountain Beaver Resources Ltd Mr Jax Fashions Inc Mullen Trucking Ltd Municipal Financial Corp Municipal Ticket Corp Muscocho Explorations Ltd Mytec Technologies Inc NI1 Norsat International Inc NQL Drilling Tools Inc Nabors Industries Inc Nalcap Holdings Inc Napier International Technologie National Bank of Canada National Sea Products Ltd National Tmst National Trustco Inc Needler Group Ltd NeoTel Inc New Brunswick Tel Co Ltd New Canamin Res Ltd NewTel Enterprises Ltd Newalta Corp

Newbridge Networks Corp Newcourt Credit Group Newfoundland Capital Corp Ltd Newfoundland Light & Pow Co Newhawk Gold Mines Ltd Newport Petroleum Corp Newscope Resources Ltd Noble China Inc Noble Peak Resources Ltd Noma Industries Ltd Noramco Mining Corp Noranda Forest Inc Noranda Inc Norcen Energy Resources Ltd North Arnerican Palladium Ltd North Canadian Oils Ltd North West C o Inc North West Trust Northern Reef Exploration Ltd N o h e m TeIecom Ltd Northern Telephone Ltd Northfield Minerals Inc Northgate Exploration Ltd Northridge Exploration Ltd Northrock Resources Ltd Northstar Energy Corp Northwest Sports Enterp Ltd Northwestem Uti tities Ltd Norwall Group Inc Nova Corp of Alberta Nova Gas Transmission Ltd Nova Scotia Power Inc Novicourt Inc Nowsco Well Service Ltd Nu-Gro Corp NuGas Ltd Nufort Resources Inc Numac Energy Inc OCS Technologies Corp OSF Inc Ocelot Energy Inc Offshore Systerms International Okanagan Skeena Group Ornega Hydrocarbons Ltd Ondaatje Corp Onex Corp Optima Petroleum Corp Orbit 0i1& Gas Ltd Orvana Minerais Corp Oshawa Group Ltd Osprey Energy Ltd Ottawa Structural Services Ltd PWA Corp Pace Corp Pacific Cassiar Ltd Pacific Energy Systems Ltd

Pacific Forest Products Ltd Pacific Northern Gas Ltd Pan Pacific Devel Corp PanCanadian Petr Ltd Pantorama Industries Inc Paragon Entertainment Corp Paragon Petroleum Corp Paramount Resources Ltd Park Lawn Cemetery Co Ltd Parkiand Industries Ltd Peerless Carpet Corp Pegasus GoId Inc Pembridge Inc Pengrowth Gas Corp Penn West Petroleum Ltd Pennington's Stores Ltd People's Telephone Co of Fore Pet Valu Inc Petersburg Long Distance Inc Petro-Canada Petromet Resources Ltd Petrorep Resources Ltd Petrostar Petroleums Inc Philip Environmental Inc Phillips Cables Ltd Pinetree Capital Inc Pinnacle Resources Ltd Pioneer Metals Corp Pipestone Petroleum Inc Place Dome Inc (Placer!) Place Resources Corp Planvest Capitai Corp Plasti-Fab Ltd Plastibec Ltd Platinova AIS Poco Petroleums Ltd Polyphalt Inc Potash Corp of Saskatchewan Power Corp of Canada Power Financial Corp Precision Drilling Corp Premdor Inc Prime Equities Interational Prime Resources Group h c Pnmex Forest Products Ltd Princeton Mining Corp Promatek Industries Ltd Promis Systems Corp Ltd Provigo Inc Prudentid Steel Ltd Qsound Labs Inc Quadra Logic Techs Inc Quadron Resources Ltd Q u m Mountain Gold Corp Quebec-Telephone Quebecor Inc

Quebecor Printing Inc Queenstake Resources Ltd Quest Capital Corp Quno Corp R-M Trust Company RAM Petroleums Ltd Ranchmen's Resources Ltd Rand A Tech Corp Ranger OiI Ltd Rayrock Yellowknife Resources Receptagen Ltd Reclarnation Management Ltd Red Oak Resource Inc Redfern Resources Ltd Redlaw Industries Inc Redstone Resources Inc Regal Greetings & G i a Reitrnans (Canada) Ltd Renaissance Energy Ltd Repadre Capitd Corp Repap Enterprises Inc Republic Goldfields Inc Resourcecan Ltd Resources Ste Revenue Properties Co Ltd Richards Buell Sutton Richelieu Hardware Ltd Richland Petroleum Corp Richmont Mines Inc Rigel Energy Corp Rimoil Corporation Rio Algom Ltd Rio Alto Exploration Ltd Rising Resources Ltd Riverside Forest Prod Ltd Rogers Cantel Mobile Comm Inc Rogers Communications Inc Rolland Inc Roman Corp Ltd Rothrnans Inc RoyNat inc Royal Aviation Inc Royal Bank of Canada Royal LePage Ltd Rusty Lake Resources Ltd SBN Systems Inc SHL Systemhouse Inc SNC Lavalin Group Inc SR Telecom Inc STN Inc Sagewood Resources Ltd Samuel Manu-Tech Inc Sceptre Inv't Counsel Ltd Sceptre Resources Ltd Schneider Corp Scintrex Ltd

Scott Paper Ltd Scott's Hospitality inc Seagram Co L td Sears Canada Inc Sedona Industries Ltd Seprotech Systems Inc Serenpet Inc Service Corporation Internationa Shaw Communications Inc Shaw Industries Ltd Shell Canada Ltd Shemt Inc Sidus Systems Inc Sifion Properties Ltd Signal Energy Ltd Signature Brands Ltd Silcorp Ltd Skyjack Inc Slater Industries Inc Slocan Forest Products Ltd Socanav Inc Sodisco-Howde Group Inc SofiKey Software Products Soiid State Geophysical Inc South China Inds (Cnd) Inc Southam Inc Southem Frontier Resources Inc Southernera Resources L td Spar Aerospece Ltd Spectra Group of Great Spectral Diagnostics Inc Spectnim Signnal Procesg Inc Speedware Corp Inc Speedy Muffler King Inc Spirit Corp Spruce Falls Acquisition St Clair Paint & Wallpp Corp St Genevieve Resources t t d St Lawrence Cement Inc Stackpole Ltd Stampeder Exploration Ltd Star Data Systems Inc Startech Energy Ltd Stelco Inc Strike Energy Inc Stroud Resources Ltd Structured Biofogicals Inc Summit Resources Ltd Sun Ice Ltd Suncor Inc Surrey Metro Savings Cdt Un Suzy Shier Ltd Synergistics Industries Ltd Synex International Inc TCG International Inc

TFH International Inc TIE/Comms Canada Inc TNT Financial Ltd TSB International Inc TSC Shannock Corp TTY Paramount Partnership TVI Copper Inc TVX Gold Inc Tai Energy Corp Taiga Forest Products Ltd Talisman Energy Inc Tandem Resources Ltd Tarragon Oil and Gas Ltd Tarxien Corp Teck Corp Tecsyn International Inc Teddy Bear Valley Mines Ltd Tee-Comm Electronics Inc Tele-Metropole Inc Telebec Life Teleglobe Inc Telemedia Inc Telepanel Systems Inc Tetular Canada Inc Telus C o q Tembec Inc Tesco Corp Texaco Canada Petroleum Inc Thomson Corp Tirnminco Ltd Tiverton Petroleums Ltd Tm Technologies Corp Tombill Mines Ltd Toromont Industries Ltd Toronto Sun Publishing Corp Toronto-Dominion Bank Toronto-Dominion Centre Ltd Torstar Corp Trader Resources Corp Traders Group Ltd Tram Mountain Pipe Line Co Trans-Dominion Energy Corp TransAlta Corp TransCanada PipeLines Ltd Transat A.T. fnc Transpacific Resources inc Transwest Energy Inc Trenton Industries Inc Tri Link Resources Ltd Tri Origin ExpIoration Ltd Tridel Enterprises Inc Trillion Resources Ltd Trilon Financial Corp Trimac Ltd Trimark Financial Corp Trimin Enterprises Inc

XL Foods Ltd Trizec Corp Ltd Xerox Canada Inc Trojan Technologies Inc Zenon Environmental Inc Tm-WaiI Group Ltd Truax Resources Corp Truscan ReaIty Ltd Trust General Du Canada UAP Inc Ulster Petroleums Ltd Uni-Select Inc Unican Security Systems Ltd Unicap Commercial Corp Union Gas Ltd Union Gold Inc United Canadian Shares Ltd United Corp Ltd United Grain Growers United Rayore Gas Ltd United Reef Ltd United Tire & Rubber Co Ltd United Tri-Star Resources Ltd United Westburne Inc Univa Inc Venezuelan Goldfields Ltd Ventra Group Inc Ventas Energy Services Inc Versacold Corp Viceroy Homes Ltd Viceroy Resources Corp Vitran Corp Inc WF4 Industries Ltd Wajax Ltd Wall Financial Corp Wascana Energy Inc Webex Resources Ltd Weldwood of Canada Ltd West Fraser Timber Co Ltd West Kootenay Power Westar Group Ltd Westcoast Energy Inc Western Corp Enterprises 1nc Western Gamet Co Ltd Western Quebec Mines Inc Western Star Trucks Hldgs Ltd Westfield Minerats Ltd Westmin Resources Ltd Weston, George Ltd Westrex Energy Corp Westward Energy Ltd Wharf Resources Ltd Wheaton River Minerals Ltd Win-Eldrich Mines Ltd Winchester Group Inc Winfield Energy Ltd Winpak Ltd Working Ventures Canadian Fwid X-Cal Resources Ltd

References Baker, George P., Evlichael C. Jensen, and Kevin J. bIixrphy, "Compensation and Incentives: Practice vs. Theory," Journal of Finance, vol. XLIII, No. 3. Jiily 1988,

593-616. Black, Fischer and Myron Scholeç, "The Pricing of Options and Corporate Liabilities," Joumal of Political Economy, May/Jime 1973, 637-54. Chaykowski, Richard P. and Brian Lewiso Compensation Practice and Outcomes in Canada and the United States, IRC press, Queen's UniversiQ, 1995. Ciscel, David H. and Thomas

M.Carroll, "The Deterrninants of Execiitive Salaries: An

Econometric Survey," Review of Economic and Statzstics, 1980, 7-13. Cosh, A., T h e Remimeration of Chef Execiitives in the United Kingdom," Economic Journal, 85? 1975, 75-94. Coiighlan, Anne T. and Ronald M. Schmidt, "Executive Compensation, Managerial Tirnover, and Firm Performancet" Journal of Accounting and Economics. T. 1985, 67-84. Garen, John E., "Execiitive Compensation and Principal- Agent Theory," Journal of Political Economy, 102, 1994, 1175-1199.

Gibbons, Robert and Kevin Rlurphy, "Relative Performance Evaliiation for Chief Execiitive Officers," Industrial and Labor Relations Review, 43, 1990. 30-51s.

Hall, Brian J. and Jeffrey B. Liebman, "Are CEOs R e d y Paid Like Biirea~icrats?" Discussion Paper Nimber 1789, Harvard Institiite of Economic Research, 1991. Haubrich, Joseph G., "Risk Aversion, Performance Pay. and the Principal-Agent Problem," Journal of Political Economy, 102, 1994, 258-276.

Holrnstrom, Bengt, "Moral Hazard and Observability," Bell Journal of Econornics, 10, 1979, 74-91. Hubbard, R. Glenn and Darius Palia, "Executive Pay and Performance Evidence from the U.S. Banking Indiiçtry," Journal of Financial Economics, 39, 1995, 105-130.

REFERENCES

94

Jensen, blichael C. and Kevin J. Murphy, "Performance Pay and TopEvIanagement Incentivest7' Journal of Political Econorny, 98, 1990, 225-264. Joskow, Paul and Nancy Rose, "CE0 Pay and Firm Performance: Dynamics, Asymrnetries, and Alternative Measureç of Performance," NBER Working Paper No. 4976. 1994. Kaplan, Steven N, "Top Execiitive Rewards and Firm Performance: A Cornparison of Japan and the United States," Journal of Political Economy, 102, 1994, 51û-546.

Kato, Takao and Mark Rockel, "Experiences, Credentials, and Compensation in the Japanese and U.S. hilanagerid Labor Markets: Evidence Erom New blicro Data," Journal of the Japanese and International Economics, 6, 1992. 30-5 1. Kostiiik, Peter F., "Firm Size and Execiitive Compensation," Journal of Human Resources, vol. XXV, no. 1, 1989, 91-105. Laughren, Floyd, "Statement to the Legislatiire on Disclosiire of Execiitive Compensation," Ontario Securities Commission Bulletin, October 15, 1993, 5 104-5 106. Lewellen, Wilber G. and Blaine Himtsman. "Managerial Pay and Corporate Performance," American Econnmic Review, 60, 19'70, 710-20. Masson, Robert, "Execiitive blotivations, Earnings. and Consequent Eqiuty Performance," Journal of Political Econorny, 79, 1971, 1278-92. McLaiighlin, Kenneth J., "Individual Compensation and F i m Performance: The Economics of Team Incentives," Workhg Paper No. 104, Center for the Stiidy of the Economy and the State, University of Chicago, 1994. Mirrlees, J. A., "The Optimal Stnictilre of Incentives and Aiithority witithin An Organization," Bell .Journal of Economics, 7, 1976, 105-31.

Murphy, Kevin J., "Corporate Performance and Managerial Remimerntion. An Empirical Analysis," Journal of Accounting and Economics, 7, 1985, 11-42.

Murphy, Kevin J., "Incentives, Learning, and Compensation: A Theoretical and Empiricd Investigation of Managerial Labor Contracts," Rand Journal of Economics, Vol. 17, No. 1, 1986, 69-77. Roberts, D. R., "A General Theory of Execiitive Compensation Based on Statistically Tested Propositions," Quarterly Journal of Economics, 70(3), 1956, 270-94. Rosen, Sherwin, "Contracts and the Market for Execiitives" with comments by Bengt Holmstrom, in Contract Economics, ed. Lars Werin and Ham Wijkander? Mas-

sachusetts: Basil Blackwell L td, 1992. Schaefer, Scott, "The Dependence of CE0 Pay-Performance Sensitivity on the Value of the Firm," Manuscript, Kellogg Graduate School of Management, Northwestern

University, 1995.

Chapter 3 Execut ive Compensation and Managerial Incentives: A Cornparison Between Canada and the United States 3.1 Introduction Studies of execiitive compensation and managerial incentives have had a long history in the United States where corporate execiitive compensation has been piiblicly disclosed for decades. In Canada, it only became possible to condiict stich stiidies after 1993 when

all piiblicly-traded cornpanies in the province of Ontario were reqtured to disclose top executives' compensation iinder new Ontario Securities Regdations. Using the newly available piiblic data on executive compensation of Canadian &ms. Chapter 2 identified a positive relationship between C E 0 pay and firm performance for Canadian fim. This chapter addresses a related qiiestion: Does the pay-performance relation differ across

Canadian and US firrns? On one hand, the extensive economic and institiitional relations between Canada and the United States siiggest that managerial compensation schemes will be similar. On the other hand, becairse of differences between Canada and the United States in areas siich as labor legislation, regdation intensity and taxation? the compensation schemes and thus pay-performance relation are expected to be different between the two coimtries. Understanding this difference may provide fùrther insights to these factors which influence a h ' s compensation system and internai managerial

incentives. This chapter examines a sample of 365 large Canadian firmç and 678 large US firms, diinng the period of 1991-N94, to compare the intensity of the pay-performance relationship, Le., incentive strength of C E 0 compensation, between the two coimtries. Althoiigh there is no consensus in the literatiue conceming the measurement of incentive strength, it is u s u d y taken to be the sensitivity of managerial pay to the firm's stock performance or to shareholder wealth. This chapter examines both the arithmetic sensitivity of C E 0

pay to firm performance (the ratio of changes in CE0 pay over changes in shareholder wealth), and the elasticity of C E 0 pay with respect tv shareholder weaith. The similarity in data between the two countries d o w s the cornparison to be made based on joint-sample regressions for Canadian and US firms. After firm-size heterogeneity between the Canadian sample and the US sample is adeqiiately controlled, 1find that the incentive strength associateci with direct pay and stock ownership in Cm-adian firrns is weaker than in US firms biit that the difference diminishes as h size increases. This resiilt is obtained both with the arithmetic sensitivity estimator and with the elasticity estimator. While it is difficiilt to provide a cornpiete explanation for this observation, it is consistent with a positive impact of disclosire regdations on managerial incentive strengths as disciissed in Chapter 1. The US execiitive compensation systern has been imder public scrritiny since the 1930s, but Canadian firms did not domestically disclose their top executives' compensation imtil the end of 1993.

This difference in disclosure regdatory environment between the two coimtries is smaller with large firms because most large Canadian h m were piiblicly traded in the United States and were siibject to the US disclosire r d e s before 1991. To evaluate the impact of rnanagerial incentive intensity on corporate performance, this chapter also imdertakes a Canada-US cornparison of the firms' stock performances.

3.2 THE METHODOLOGY AND THE DATA

98

1find that, diiring the period of 19841994, the average r e t i m to a h ' s common stock

and the ratio of the average retum to the return standard deviation are both lower for Canadian

than for US b. ' Compared with US firms, this poorer stock

performance and the lower compensation paid to Canadian CEOs is evidence of less productive management of Canadian firms. In t i m , this observation, together with the weaker pay-performance relationship at Canadian hm,is at least consistent with the

hypothesis that managerial incentives d e c t corporate performance. The rest of the chapter is organized as follows. Section 3.2 introduces the basic: issiies concerning the pay-performance relationship; i t briefly disciisses previocis empirical stiid-

ies and then describes the methodology and the data i w d in this paper. In Section 3.3. a detailed cornparison of incentive strength with C E 0 direct pay and stock ownership is

made between Canadian firrns and US fimis. Comparing the h m s ' market performance. Section 3.4 evaliiates the impact of the Canada-US difference in managerial incentive schemes on corporate performance. Section 3.5 concludes the chapter.

3.2 The Methodology and the Data 3.2.1 Previous Studies An incentive problem arises when the owner of a firm, the principal, delegat'es decision-making or production tasks to anot her individiial, the agent (manager). and where the principal does not observe the agent's effort in prodiiction. Standard agency theory predicts that, to indiice the agent to promote the interests of the principal, the optimal rewaxd scheme miist have the agent's pay V a r y with his o ~ i t ~ i iIn t . other ~ words, to align the interests of the agent with those of the principal, the agent's pay miist be lThe ratio is based on the Sharpe measiire for portfolio performance ( s e , e.g., Jobson and Korkie, 1981).

*since Ross (1973) k t posed the principal-agent problem, many stiidies have been done to examine the optimal compensation scheme of the agency model, among which notably ~ u eMirrlees (1976), HoIrnstrom (19791, Grossman and Hart (1983),and Hart and Holrnstrom (1987).

3.2 TNE METHODOLOGY AND THE DATA

99

performance-based. This is the starting point of varions empiricd stiidies which, from different perspectives, investigate the relationship between managerial pay and firm performance. Focusing on cross-sectional analyses, earlier stiidies tried to identify a positive relationship between managerial pay and firm performance, but the residts were mixed (see. e.g., Lewellen and Htmtsman, 1970; Masson, 1971; and Ciscel and Carroll, 1980). Since

the late 1980s, most stiidies iise longitiidinal data and overwhekningly co&m

a posi-

tive relationship between pay and performance, even thoiigh the estimated performancerelated coefficient depends criticdy on the mode1 specification and choice of performance measiire (see, e.g., Miirphy, 1985; Coiighlan and Schmidt, 1985: Jensen and hIiirphy, 1990; Kaplan, 1994; and Joskow and Rose, 1994). Jensen and Miirphy give a bench mark estimate of the intensity of the pay-performance relationship. Using the Forbes sample of US CEOs' compensation diiring the years 19741986 and taking into accoimt cash payments, options, stock ownership and dismissal-

related factors, they find a pay-performance sensitivity of 0.003. This means that C E 0 compensation in the largest US fimu increases $3 for each $1000 increase in shareholder value.

Kaplan (1994) reaches a similar conclusion for the compensation of Japanese

exe~iitives.~While different agtiments have been put fortli to e-xplain the seemingly

small magnitude of these pay-performance sensitivitie~,~ there is no consensus concern3 ~ a p l a nfinds that while the relationship of turnover and cash compensation to performance are sirnilar between Japanese execiitives and US. executives, the share ownership of Japanese exec~itivesis much lower than that of the US. coirnterparts, and conclitdes that top Japanese exectitives have smaller incent ives to increase stock rettirns than U.S. executives. 'Jensen and Mi~rphy(1990) conjecture that under disclosiue of execiitive compensation, public and private political forces impose constraints on the type of contracts that are written between shareholders and management, and thiis reduce sensitivity. Joskow, Rose, and Shepard's (1993) explanation for regdatory constraints on C E 0 compensation d s o belongs to the political pressure argument. Batlbrich (1994) and Garen (1994) tend to explain the small sensitivity in t e m of agency costs. Given the riskaverse behavior of managers, the compensation scheme must be structiued to trade off incentives with insimance. Under this argument, the srnall pay-performance sensitiviw is a response of the pay scherne to large otitput variation in the largest U.S.firms, reflecting the shift of the balance between incentives and insiuance.

3.2 TH%I1fETHODOLOGY AND

THE DATA

100

ing whether or not the compensation of CEOs in the largest US firms provides the appropriate incentives to m a e z e shareholder wealth. Since the general agency model fails to identi& a quantitative benchmark, it is diEcult to draw a concliiçion conceming the incentive implications of a s m d pay-performance sensitivity. Setting the absolute magnitude of the pay-performance sensitivity aside, one can still profitably piwiie a comparative analysis. In recent years, cornparisons of managerial incentives have been made between s r n d fhns and large h

s (Jensen and hliuphy,

1990; Garen, 1994); regulated fhm and imregulated firms (Joskow. Rose and Shepard. 1993; Smith and Watts, 1992); and Japanese firmç and US firms (Kato and Rockel. 1992: Kaplan, 1994). In this chapter 1imdertake a cornparison of Canadian firms and US firms.

3.2.2 Mode1 Specification There are a variety of different mode1 specifications in the empiricd Literatiire on managerial incentives. The dependent variable can be the log of compensation. changes in the log of compensation. or dollar changes in compensation. The measiires of firm per-

formance are even more diverse: stock price, varioiis rates of retiun, changes in these rates

of retiirn, dollar changes in shareholder wealth, etc. As a matter of interpretative convenience, however. the model proposed by Jensen and bhrphy has been increasingly iised in the literatiire. This model links changes in managerial pay to changes in shareholder

wealth. Denoting C E 0 pay as C1/, and hm oiitpiit as

x,where siibscript t denotes fiscal

year, a Linear compensation contract pays Wt = a + b x . Jensen and bvlurphy estimate the model in its differential form:

3.2 THE METHODOLOGY AND THE DATA

101

The regressor, A&, is d e h e d as the change in shareholder wealth cdciilated as rt&-17

where rt is the rate of retiun on cornmon stock realized in fiscal year t and

v-l is the firm

value at the end of the previous fiscal year. As a iinear approximation to marginal reward for fum performance, the arithmetic sensitivity of CE0 pay to shareholder wealth,

LI:

measures the strength of the pay-performance relation.

As observed in several previoiis studies, the sensitivity estimated imder (3.1) changes significantly with firm size.' This featiue of the mode1 imposes a potential problem for oiir cornparison becaiise of a siibstantial clifference in firm size between Ccmadicmh

s

and US firms. For this reason 1 also employ the specification iised by Gibbons and

Murphy (1992) that converts the regression variables to logarithmic changes as follows:

where A log(%) is approximated by log(1 + r t ) . Under (3.2) the effect of firm size on the estimate of the elasticity of C E 0 pay with respect to shareholder wealth, q, is siibstant i d y rediiced. In the next section, the two specifications, (3.1) and (3.2), are both iised in the cornparison of incentive strength between Canadian firm and US firms. Recognizing ~ term that p s t performance may also have an infiiience on ciment ~ o r n ~ e n s a t i oan ,lag is added in each specification. 'With US. CE0 compensation data, Jensen and Miuphy (1990), Garen (1994) and Schaefer (1995) al1 find an inverse relationship between the sensitivity and firm size. 'Both Jensen and Miirphy (2990) and Joskow and Rose (1994) confirm a significant influence of the previoiiç performance on c ~ u ~ ecompensation. nt Stich a phenornenon partly results fiom the ambiguity of the timing of performance payments in both the decision and the firm's report ( s e Jensen and Miirhpy). In addition, various long-tenn incentive payrnents that usually depend on performance over severai years directly link ciment compensation with previous performance.

3.2 THE METHODOLOGY AND THE DATA

3.2.3 The Data

The Canadian Data The compensation data of Canadian CEOs are compiled £rom the corporate pr0.x-y staternents that are Bled with the Ontario Seciirity Commission and which are also maileci to shareholders. Under the new Ontario Regdation (1993): aJl piibiicly-traded companies in Ontario are reqiiired to disclose all payments to their CEOs and their foiu other rnost

highly paid execiitive officers. Thoiigh the sarnple is based on the h subject to the Ontario regdation, most large Canadian firms are piiblicly traded on the Toronto Stock

.~ Exchange; the resulting sample is certainly representative of large Canadian f k m . ~ The compensation data are dram Erom dl corporate proxy statements filed between Janiiary

1994 and July 1995 which separately report executive compensation for fiscal year 1991 through 1994. Execiitive remuneration has several components. Current cash payments (salas; annual boniis. and various benefits) are reported, dong with stock options (and stock appreciation rights), g a n t s of restricted stock imites, and iong-term incentive plan payoiits. In a broad sense. these components c m be classified into f o t ~categories: s a l q anniid bonus, Long-term incentive payments, and other payments. Indirect-pay related benefits reported in proxy statements include stock gains from the exercise of previoiisly rewarded stock options or stock appreciation rights and stock holdings of directors." Financial data on Canadian firms are obtained from the Financial Post data file. Becaiise the change in C E 0 payment will be iised as the dependent variable in the regression below, those firms that have only one year of data are ignored. Excliiding 'For example, by either total assets or sales levels in fiscal yesx 1993, more than 90 percent of the 300 largest Canadian finns are traded at the Toronto Stock Exchange (TSE in short). T h e compensation data incltide 98% of TSElOO firms and 92% of TSESOO firms. 'In most cases, a CE0 is also a director. Since the disclosure niles require firms to provide only the ciment fiscal year information on executive stock gains and directors' stock ownership, the data of these two items are only available for fiscai years ending 1993 and 1994.

3.2 THE METHODOLOGY AND THE DATA

103

partial compensation due to tirnover, firm reorganization, Bscd year change and parttime service, and eiiminating the observations with misshg data after matching C E 0 compensation with £hmhancial data, the Canadian sample contains 365 firms for a total of 1183 observations.

The US Data To make the data comparable with previous US stiidies. a Forbes sample of US h

s

is iised. The CE0 compensation data are obtained from the Forbes ' anmal report of

executive compensation for fiscal year 1991 to 1994. The report inchides three compensation categories: total salary plus bonus, stock gains, and other. Total s a l q plus boniis is the same as the s i m of salary and annual boniis in the Canadian data."tock gains are the net valiie realized fiom the evercise of options or stock appreciation rights

previoiisly anwded. The other category incliides long-term incentive payments (which are reported separately in Canadian fims' proxy sçtatements) and variotis benefits. CE0 stock ownership is also reported in terms of the value of each executive's holding as a percentage of the firm's total market valiie. The financial data for US firms are obtained from the Compustat data files. The US sample contains 678 f i m s with a total of 2169 observations.

Selected Statistics This chapter focuses on CEOs' direct payments and stock ownership. Since there are only two years of data on stock options available for Canadian fims ( 1993 and 1994), and no option data in the Forbes' s i w e y on US C E 0 compensation, the pay-performance g ~ h Ontario e disclosiire niles closely W o w the US disclositre d e s that reqiùre firrns to disclosiire executive compensation in standardized tables. Before 1993, C E 0 salary and boniis are rcported as a total in the Forbes reports.

3.2 THE METHODOLOGY AND THE DATA

links associated with options and stock appreciation rights are ignoreci.''

The effect

of option-related incentives will be discussed in Section 3.3.3 The pay variables in the Canadian sample include salary, boniis, long-tem incentive pay excluding options, and other payments. These four components are combineci to form two major pay variables, sa.lary+bonus and total direct pay, which can be meastireci for each firm in the Canadian and US sarnples.

AU discussions of a Canada-US cornparison are based on these two pay

variables. Table 3.1 summarizes selected statistics of C E 0 compensation as well as some

firm variables. The

value of Canadian variables is converted into

US dollars based on

anniial average exchange rates." -411variables are in 1991 constant dollars. There exists a siibstantial difference in h m size between the Canadian sample and

the US sample. Based on median assets and sales, the average size of US firms is about 12 to 15 times as large as that of Canadian firms. The mean total pay to US CEOs is about

4 times as high as that of Canadian CEOs. Stipposing that there is a common managerial Iaboiir market for Canada and the United States, these observations are consistent with the allocation theory of control: in a market eqtdibriiun. the rnost talented execiitives occiipy positions in the largest fums and thiis receive the highest pay (Rosen, 1982). They are also consistent with the fact that the firm-size elasticity of execiitive compensation is smaller than one.

Salary and bonus are the most important compensation components in both Canndian and US firms,constituting an average of 94% of total pay for Canadian CEOs and -

--

--

-

.

-

-

1°While CEOs' wealth in stock gains from exercising previously awarded options directly depends on stock price, the relationship between stock performance and the valtie of the ciment option g a n t s remains unclear. Taking into accoimt both the valtie of options and the benefits hom exercising options, Jensen and Murphy (1990) find the change in CEO's option-related wealth is positively related to the change in shareholder wealth. Excliiding stock gains of exercising previoiisly awarded options, however, Murphy (1985) finds a negative relationship between firm performance and the value of options granteci in the current fiscal year. "To avoid introdiicing noises into the data, al1 Canadian variables remain unconverteci in the regressions, which affects the constant term in the joint-sarnpIe estimation but causes little &ange in the performance-related coefficients.

82% of total pay for US CEOs.12 It is interesting to observe that the median and mean values siiggest a left-skewed distribution for this percentage. Noticing that the distribution of firm size is right-skewed. this seems to suggest that larger firms tend to lise more long-term incentive payments and other benefits, or ernphasize a variety of incentive payments.

3.3 Incentive Strength

3.3.1 Incentives from Direct Payments The cornparison with direct payments fociises on sdary pliis bonus and total pay. iising both the arithmetic and the elasticity specifications. In the first stage of the analysis, the Canadian and US sampleç are cornbined and regressions are performed for the joint sample iising the two specifications by introdiicing a diimrny variable. CND. for Canadian firms. Table 3.2 presents the results. The £irst two coliimns present the resiilts kom the arithmetic estimation. The siun of the coefficients on changes in shareholder wealth estimates the pay-performance sensitivity for US firms. The s i m of the coefficients on the corresponding t e r m with d i m y variable C N D estimates the Canada-US difference in the sensitivity. Incliiding the coefficients on the previoiis year's performance. the

sensitivity for US firms is aboiit 8 cents change in salary m d bonus for every $1000 change in shareholder value. Of the change in total pay the sensitivity increases to

20 cents for every $1000 change in shareholder wealth. These estimates are consistent with, thoiigh higher than, those of previous stiidies. l3 The fociis here is the coefficients 12When stock gains from exercising previoiisly awarded options are incliided in total pay in the US. sample, the percentage of saiary and bonis in total pay for US. CEOs drops to 72. This is consistent with the finding of Joskow and Rose (1994) that salary and bonus averaged 82 percent of total compensation diiring 1972 to 1990, and the percentage declined by an average of 1.1 percentage points per year (p13). 13using the Forbes sarnple during 19741986, Jensen and Murphy report a sensitivity of 0.000022 and 0.000033 with respect t o saiary plus bonus and total pay respectively. With a similar sample for the period between 1986 and 1990, Schaefer identifies a sensitivity of aboiit three tirnes as large as that of

on the dummy variable for Canadian firms. The estimates of these coefficients can be

sumrnarized as the following: the sign is mixecl, the magnitude of the coefficients is relatively small, and none are significant. This result suggests no significant difference in pay-performance sensitivity between Canadian firms and US firms.

The regressions for the elasticity specification, however, are quite different, as shown in the third and fourth columns of Table 3.2.

AU the coefficients on d i i m y variable CND

are negative and most are statistically significant a t the one percent level. Incltiding both the curent and previous years' performance variables, the elasticity of salary+boniis with respect to shareholder value is 0.278 for US firm and 0.130 for Canadim. h s . For every one percent increase in shareholder wealth, US CEOs' salary and boniis increases by about

0.28 percent while the Canadian coiuiterpart increases by 0.13 percent. The difference is even larger for the elasticity of total pay. The large magnitude and high significance level of the difference in the elasticity implies that C E 0 payments are much less sensitive to

firm performance in Canadian firms than in US firms. This residt, obvioiisly, contradicts that frorn the arithmetic estimation.

If there exists a significant difference in the responsiveness of C E 0 compensation to k m performance between the two samples, it shodd be shown in both the arithmetic semitivity and elasticity. A possible reason for the inconsistency is sample heterogeneity which poses a problem when it has an asymmetric effect on the two specifications. -4s observed by eariier stiidies, while the elasticity estimator appears qiiite robiist to firm size, the arithrnetic sensitivity estimator deciines notably as firm size increases. This suggests the possibility that the Canada-US difference in pay-performance relation may not be revealed iising the arithrnetic estimation because of the large difference in b r n size Jensen and Miirphy. These resiilts are consistent wit h the observation of Joskow and Rose (1994) that CE0 compensation in the US. firms surveyed by Forbes became significantiy more sensitive to firm performance during the 1980s compared to the 1970s.

between the two samples. To clarify this issue, 1 now examine firm-size heterogeneity. F i m size is usudy proxied by total assets or saies. Noticing that the distribution of sales is less skewed than that of total assets, 1 use average sales during the years 19911994 as a proxy for hsize throiighout the discussion. Figure 3.1 illustrates the 6rm size distribution for the two samples. While about 80% of Canadian fkms in the sample have average sales below 500 million, only about 9% of the US firrns in the sample have average sales below this level. Given such a striking difference in the h - s i z e distributions between the two samples, a distortion imavoidably occiirs in the arithmetic estimation. Becaiise finri size is a continuous variable, a dummy variable d l not help control for the effect of size heterogeneity. In this chapter 1 adopt an approach that

rninimizes size heterogeneity by directly controlling for the sample that is iised. Taking into acconnt the nimber of firms and the firm-size distribiition, I choose to incliide those firms with average sales between $400 million and $8 billion; this siibsample contains

35.6% of the total Canadian sample and 78.3% of the total US sample. These fkms are fiirther divided into two subsamples for both Canadian fkms and US fim. Siibsample 1 incliides firms with average sales between $400 million and $1 billion, and siibsarnple

2 incliides firms with average saies between $1 billion and $8 billion. As shown in Table 3.3,14 the Canadian and US firms have sirnilar median and mean average firm saies in

the two siibsarnples. After controlling for size heterogeneity, 1 redid the regressions of Table 3.2 for the joined siibsamples. 1hst examined the combined subsample which includes 130 Canadian h m s and 531 US firms. The results are presented in Table 3.4. The two specifications now

provide consistent estimates of the Canada-US difference, and both confirm a weaker pay'"~istedin Table 3.3 t here is anot her dimension of sample heterogeneity, indiistry çtnictiue, which is discimeci at the end of this section.

performance W g e for Canadidian firms. The coefficients of the ciment year performance interacteà with the d t m y variable CND are all significantly negative.15 Incliiduig the effect of the previoiis year's performance, the anthmetic sensitivity of C E 0 total pay to shareholder wealth for Canadian firms is only aboiit 20% of the corresponding US value. Sunilarly, the shareholder-value elasticity of total pay in Canadian firms is aboiit 27% of the US number. These residts establish that the incentive strength of CE0 direct pay is significantly weaker for the Canadian h s in the total siibsample 1 plis 2.

1 then examine subsamples 1 and 2 separately. Table 3.5 presents the regressions for siibsample 1. The resiilts are qiialitatively sirnilar to those for the total siibsample in the sense that both the arithmetic sensitivity and the elasticity are significantly smaller in Canadian firms than in US fimis. There are some differences thoiigh. Compared with the residts in Table 3.4, the estimate of the Canada-US difference is larger and the sio~ficancelevel is generally higher in Table 3.5. This implies that there is s larger Canada-US difference in the pay-performance relation for the siibsample of smaller firms. This observation is strengthened by the regressions for siibsample 2 shown in Ta-

ble 3.6. Althoiigh al1 of the coefficients of firm performance interactecl with CND are negative in both the arithmetic estimation and the elasticity estimation.16 none is significant. Compared with the resiilts from siibsample 1, the magnitude of the coefficients for the Canada-US difference are substantially rediiced with subsample 2. On one Iland' the ovenvhelmingly negative sign of these coefficients still siiggests a weaker pay-performance Linkage in Canadian firrns. One the other hand, the low significance level and smaller I5The two coefficients in the arithmetic estimation are statistically significant a t the 5 percent level and the two in the elasticity estimation are statistically significant a t the 1 percent level. The relatively low significance level in the arithmetic estimation may reflect the effect of the rernaining size difference between Canadian firms and US. fims in the total siibsample 1 and 2, where the average U.S.firms are about twice as large the average Canadian firms. 1 6 ~ hcoefficient e on the previoiis performance with dummy variable CND in the regression of changes of total pay becomes zero. Biit the siim of the two coefficients on the performance variables with CiVD is still negative.

magnitude of the coefficients sitggests that there is no real economic différence in incentive strength for the siibsample of larger firms.

While aJl of the subsample regressions indicate that there is a weaker pay-performance relationship in Canadian Enns, the regressions for the two separate siibsamples sitggest that there is an inverse relationship between the Canada-US m e r e n c e and firm size. Using an alternative approach, I will provide further evidence for these residts and then extend them to the total sarnple.

As pointed out by Gibbons and h.Iiuphy (1992), and conf~rmedby the above regressions, the elasticity estimator changes little with firm size among large US h s . Thus, to proceed, 1assume that elasticity of CE0 pay with respect to h performance is roiighly constant within the US sample. Under this assiunption, and introdiicing the variable

SIZE to measiue fkm size and the dilmmy variable CND to denote CanadiCuih s , specification (3.2) is revised as follows:

where SIZE is eqiial to the firm's average sales diuing 1991-1994. The sim gi +q2SIZEo as a linear approximation, characterizes the relationship between firm size and the

Canada-US difference in pay-performance elasticity. A negative value of the s i m means a weaker incentive strength with Canadian hns. To check whether the Canada-US difference declines as firm size increases, 1 simply check whether

72

is significantly positive.

Estimates of (3.3) for the total sarnple are reported in Table 3.7. Colurnns 1 and

3 present coefficient estimates when only the crurent year's performance miables are regressors. In both regressions, coefficient 01 is significantly negative. Coefficient pn is positive in both regressions and significant at the one percent level for the estimate with

total pay. When S I Z E < 3 billion, which is the case with about 93% of all Canadian

h

+

s in the sarnple, the siun of 71 q2SIZE is negative. These residts veri& an earlier

observation that the pay-performance linkage is weaker in Canadian hthan in US

fkms! and that the difference declines as firms become larger. For example, for a Canadian

hwith average sales of $100 million, the performance elasticity is 0.101 for salary and bonus and 0.097 for total pay, which is about 39% and 27%, respectively, of the US counterparts. For a Canadian firm with average sales of $1 billion, these elasticities rises to 0.140 and 0.175 respectively, and become 53% and 50% of the US estimates. When the performance variables of the previous year are incliided, the qtialitative residts remain the same, as shown in coliunns 2 and 4. So far 1 have ignored the effect of indiistry heterogeneity. As shown in Table 3.3. there is a marked difference in indiistry stnictiire between the Canadian sample and the

US sarnple. Dividing 6rms into two categories, resoilrces (mining, rninerals, and oil and gas) and non-resoiirces, based on the two digit

SIC classification, Table 3.3 shows that

18.4% of Canadian firms in the sample fa11 into the resoiirce category, but only 1.3%of the US f i r m belong to the resoiirce indiistry. To capture the effect of this difference, al1 of the regressions iising the total samples were reestimated with an additional diimmy variable for resource hm.While the signs of the coefficients of the diunmy variable for resource firms are mixed and most of these coefficients are insignificant, the estimate of the Canada-US difference in incentive strength rernains essentially the same. As for the regressions with the subsamples, since the nimber of resolirce firms is very srnall in both the Canadian sarnple and the US sample, indiistry heterogeneity becomes negligible.

3.3.2 Incentives from Stock Ownership Stock ownership provides another important contribution to managerial incentives. Compared with direct compensation, the incentives effects of stock ownership are a-

guably greater. According to Jensen and Murphy, the incentive intensity of C E 0 stock ownership in terms of arithmetic pay-performance sensitivity is about eight times as high as that of direct payments. Hence, it is necessary t o examine the incentives from C E 0

stock ownership before a reasonable conclusion can be drawn on the relative intensity of managerial incentives between the Canadian and US frrrns.

The data of Canadian CEOs' stock ownership are obtained from the k m ' s proxy statements for the fiscal years ending 1993 and 1994. The stock holdings of US CEOs for the same fiscal years as the percentage of the £ k n 7 stotal market valiie are obtained from the Forbes' anniial report. The data incliide the shares held by family members but exclude the shares acquirable throtigh options. Wealth changes in CEOs' stock ownership are perfectly correlated with changes in shareholder wealth. This, the percentage of a CEO's stock shares in his firrn's total shares o i i t s t d n g has the same interpretation as the arit hrnetic pay-performance sensit ivity in regard to incentive intensity. I first look a t the sample statistics for C E 0 stock ownership percentage s h o m in Table 3.8. For the total samples, Canadian CEOs o m an average of 2.811 percent and a median of 0.161 percent of their firms' stocks, while US CEOs own

an average of 1.823 percent and a median of 0.17 percent." The close medians but miich higher mean percentage for Canadian CEOs seem to siiggest a higher stock holdings and thus a stronger incentive strength associated with CE0 stock ownership in the Canadian sample.

As before, hm-size heterogeneity affects the Canada-US cornparison of CE0 stock ownership. For this reason, the percentage owned is also calcidateci for the two subsamples. Table 3.8 shows that Canadian CEOs in siibsample 1 own an average of 1.124 '?The mean and median levels of U.S.CEOs' stock ownership are similar to, b u t smaller than, that reported by Jensen and Mitrphy (2.42% and 0.25%). O n e possible reason for t h e ciifference is that before 1991 the Forbes ' report on stock ownership inclildes options that can be exercised within 60 days.

percent (a meàian of 0.056 percent) of their firm's stock, compareci with an average of

2.271 percent (a median of 0.360 percent) owned by US CEOS. In siibsample 2, the average percentage becomes 0.833 (median percentage 0.016) for Canadian CEOs and 2.052 (meciian percentage 0.170) for US CEOs. With 6rm size being controlled, the results show that Canadian CEOs hold a much smaller percentage of their h s ' stock than their US coiinterparts do. In other words, the incentives provided by stock onmership are s m d e r in Canadian firms than in US firms. This obsenration is consistent with the earlier results for direct payments.

To obtain a more formal and systematic pictiire, 1 firther explore the relationship between CE0 stock ownership and firm size by analyzing the relation between stock holdings and f i m size. The following simple mode1 is iised For this piirpose:

where Sharesl is the percentage of C E 0 stock shares in the firm's comrnon share outstanding by the end of fiscal year t. Estimates of the parameters of (3.4) are presented in Table 3.9. Coliunns 1 and 2 report separate estimates for the Canadian and US saruples. respectively. The coefficients on log(Salest) are negative and statisticdy significant in both regressions, conkming a strong negative correlation between the percentage of C E 0 stock ownership and k m

size. Compared with the US sample, a rniich srnaller intercept in the Canadian sample siiggests a significantly smaller stock ownership of Canadian CEOs, which is verified in

-

the pooled sample regression, C o l i m 3. The coefficient on CN D log(Sa1es) in the pooled sample regression is positive. Thoiigh the significance level of the coefficient is

low, its positive sign is consistent with the earlier comparative residt for direct pay, i.e.,

the Merence between Canadian fkns and US fkms decreases in larger firms.

3.3.3 The Effect of Stock Options It is widely reco-ed

that stock options have becorne an increasingly important

component of execiitive compensation. We thris need to determine whether or not ignor-

ing options imdermines the cornparisons made above. Jensen and Murphy obtain an estimate of 0.00015 for the pay-performance sensitivity of C E 0 options for 73 Fortune 500 manufacturing 6rms diiring the period 1969-1983; they incliide c~vrentlygranted options and stock gains from exercising previoiisly awaxded options. This sensitivity is aboiit seven times as large as the sensitivity they estimate for

salary and bonis. Examiniog 147 banks over the decade of the 1980s, Hiibbard and Palia

(1995) find a sensitivity for options that is aboiit six times as large as that for s d q plus bonlis. Given that the sensitivity of total pay is roiighly twice as large a s that of salary pliis boniis,18 both of these stiidies therefore siiggest that, for large US firms. the incentive strength associated with CE0 stock options is three to foixr times as large as that associated with C E 0 total p . Siipposing that stock options play a similar role in C E 0 compensation across Canada

and the United States' incliiding options in the cornparison made earlier shoidd not change the qualitative residts concerning the Canada-US difference in managerial incentives. On the other hand, becarise the incentive strength fiom C E 0 stock ownership is

more than 10 times larger than that from direct pay in US firms, and since Canadian

CEOs hold percentages of their £hm'stocks that are, on average, less than half of that of US CEOs, even a modest difference in option-related incentives between Canadian firms and US firms is iuilikely to change oiir earlier conclusions. 18See Jensen and Miliilphy (1990)and the above results of this paper.

3.4 INCIENTIVES AND PERFORMANCE

114

To pursiie this matter, 1 also compare, for different samplos and coimtries. the gains from exercising options as a portion of C E 0 total pay (including the stock gains). There are two reasons for undertaking this cornparison. First, becaise execiitive options are non-transferable, the stock gain portion of total pay gives an estimate of the relative importance of options as a pay componentowhich in t u m reflects the relative importance of option-relateci managerial incentives Second, the data are available for both Canadian

firms and US f k m only for the fiscal years ending 1993 and 1994. As shown in Table 3.10, the ratio of stock gains hom exercising previoiisly awarded options to total C E 0 pay is smaller in Canadian firmç than in US f k n s for both the total sample and each siibsarnple. In the two siibsamples, the stock gains constitiite about 5 percent of total compensation for Canadian CEOs which is only 40% of that for US CEOs. This ciifference siiggests a notably s m d e r compensation role for stock options in Canadian firrns than in US h s . This, had we been able to incliide options in the econornetric comparison. the Canada-US difference in managerial incentive strengths woidd likely have been even laxger .

3.4 Incent ives and Performance Managerid incentive schemes are important becaiise of t heir impact on corporate performance. That is, because managerial effort increaes with incent ive strength. a stronger pay-performance linkage is e.xpected to lead systematically bet ter firm perfor-

mance. Abowd (1990) obtains evidence siipporting this argument. Examining managerial compensation for 250 large corporations diring 1981-1986, Abowd fin& that increases in the sensitivity of compensation to shareholder return enhanced corporate performance. Given the Canada-US m e r e n c e in management incentives disciissed above, I now examine whether or not US £hmhave in fact outperformed Canadian h s .

3.4 INCENTIVES AND PERFORhfANCE

A C E 0 pay scheme is expected to &ect a firm's fimdamentals. As an approximation, the average retiun to the firm's common stock during the related period is examineci. Becaise of a potential positive correlation between average retiun and retiirn variance (the riskiness), the retiini alone may not teIl the whole story. For this reason. the performance measiire of the average retiirn to the r e t i m standard deviation. a Sharpe mesure for portfolio performance ( s e Jobson and Korkie, 1981) is examined as well. 1 first compare the r n o n t w retiun. The Canadian data are obtained fiom the

Uni-

versity of Western 0nta.rio7sTSE monthly retiun file. The US data are retrieved from the CRSP stock retiirn file. Since the TSE data are only available for the period before 1992, 1 compare the firm's monthly r e t i m over the years 1987-1991. The resiilts are siimmarized in the iipper panel of Table 3.1 1. To be consistent with the disciission of last section, I fociis on the two siibsamples. As shown in the table, the average retiini to common stock is s m d e r , and the standard deviation is larger, for Canadian firms than for US firms. Conseqiiently, the ratio of the average retlun to the standard deviation is much smaller for the Canadian siibsamples. Based on mean ~~dues, this ratio for Canadian firms is 0.064 and 0.092 for siibsamples 1 and 2, respectively. which are only 15.1% and 63.0% of the corresponding US niunbers. This indicates a siibstantial difference in the systematic performances of Canadian and US hm. In tiim? this performance difference, together with the Canada-US difference in pay-performance linkage? is at least consistent with the hypothesis t hat managerial incentives &ect firm performance. 1 also examined firrns' annual stock returns during the period of 1984-1994 with

the Cornpustut data for both Canadian firms and US firms. The resiilts are s h o m in the lower panel of Table 3.11. The ciifferences in the average retiinis and the ratios of the average retuni to the standard deviation are similar to these resulting monthly data, again indicating poorer performance of Canadian firms. It is worth noting that

3.4 INCENTIVES AND PERFOI3.kIANCE

116

the standard deviation of annual retiini is very close between US 6 . m and ~ Canadian

finus. The higher ratio of average return to retiirn standard deviation for US fhns cornes mainly from a much higher average return. To eliminate the effect of ciifferences in the risk free retuni, the above cornparison is also made after the return data are adjiiçted to determine the retiun premiim r - r j , where r denotes the r e t m to the fïrm's common stock and r j denotes risk free rates. Yields on three month Treasury Bills are used as rj. Using r - r f , the Canada-US increases while the qiialitative resdts remain the same. Supposing that managerial labour is mobile between Canada and the United States, a fiirther implication of the siiperior performance of US his that a management qitality at US £kms is

&O

higher. This implication fhds siipport in the difference in pay levels

between Canadian CEOs and their US coimterparts. As s h o m in Table 3.12. diiring 1991-1994 an average Canadian C E 0 in siibsample 2 earned a total salary plus bonus that was about 59 percent of that of an average US CEO. In terrns of total direct pay, the percentage is only 46. In siibsample 1, these percentages drop to 19 and 37, respectively. The difference in the level of C E 0 pay between the two coiintries is siirprisingly large. There are, of course, indirect pay-related factors that have not been taken into account, e.g. non-monetary benefits and jo3 seciinty whch may be emphasized more in

Canadian fim. However, since direct paynents are the most important components of compensation, the difference is iinlikely to disappear even if indirect-pay related factors are incliided.

These residts are consistent with the possibility that, compareci with US firms, a l e s performance-based pay system residted in generally poorer performance by Canadian

firms. A natiiral question for fiuther research is, why haven't Canadian hadopted managerial pay schemes more similar to their US cormterparts.

3.5 Conclusion This chapter compares managerial incentives from C E 0 direct pay and stock ownership between large Canadian fkms and US tirms during the period of 1991-1994. blltch attention is paid to the effect of firm-size heterogeneity between the Canadian sample and the US sample, as this proves crucial for the cornparison of incentive strengths. With firm-size heterogeneity, inter-sample cornparisons may be rnisleading and dependent on the choice of mode1 specification. This phenornenon occurs becaiise of a different effect of

hsize on different specifications. When sample heterogeneity is adeqiiately controlled. consistent results are obtained with both the arithmetic sensitivity estimator and the elasticity estimator. This chapter shows that the incentive strength of C E 0 compensation is weaker in Canadian firm than in US firms and that the difference declines as h m size increases. For firms with average sales between $400 million and $8 billion (which is the case with

130 Canadian firms and 531 US fkms in the total sample). the sensitivity of C E 0 total pay to shareholder wealth for Canadian firms is aboiit a qiiarter of that for the US coimterparts. Canadian CEOs hold a percentage of their firms' stocks which is less than

half of the fiaction held by US CEOs.

Examining the return to firms' comrnon stocks, it is foiind that US firms oiitperformed Canadian 6.rrn.s diuing the period 1984-1994. This observation, together n i t h the fact that, during 1991-1994, an average Canadian C E 0 earned a total direct compensation that was aboiit 50% of that of an average US CEO, is consistent with there being less productive management in Canadian fim. In turn, a less performance-sensitive managerial pay systern and generdy poorer performance by Canadian firms are consistent with the hypothesis that managerial incentives affect corporate performance.

Table 3.1 Selected Statistics Canadian Firms

U.S. F i m s

Median Mean Observation

Median Mean Observation

Pav Variables ($000)

Salary+bonus Total pay (Salary+bonus)/Total (%)

274 290 100.0

365 404

94.0

1,183 1,183 1,183

917 1,115 90.1

1,136 1,584 82.3

2,169 2,169 2,169

1,180 1,180 1,180 1,183

4,258 2,533 1,942 11.8

12,250 5,189 4,123 23.5

2,160 2,155 2,166 3,169

F i m Variables (%miIlion)

Assets Sales Market value Shareholder retum (%)

284 21 1 242 9.92

2,461 1,180

795 25.4

Note: The data are from 1991 to 1994. A11 variables are annual and in 199 1 US dollars.

Table 3.2 Estimation of Pay Performance Sensitivity (total sarnple) Dependent Variables -

(Arithmetic Estimation)

(Elasticity Estimation)

Independent Variables

+Bonus)

*(Total Pay)

~log(Salary +Bonus )

alog(Tota1 P~Y)

lntercept CND

~(ShareholderWealth)

CNDxa(Shareho1der Wealth)

~(ShareholderWeaIth).,

CNDxa(Shareho1der Wealth).,

aIog(Shareho1der Wealth)

CNDxalog(Shareh lder Wealth)

alog(Shareho1der Wealth).,

CNDxalog(Shareh1derWea1th)-,

R' Observations Note: t-statistics are in parentheses. CND is a dummy variable that equals one for Canadian firms.

Table 3.3 Selected Statistics about Sarnple Heterogeneity Canadian Firms Media. Size

Mean

Num of

Size

Fims

U.S. Firms Median Mean Num of Size Size Firms

Total sam~le Total Resource Non-Resource

Total Resource Non-Resource

Total Resource Non-Resource

Tota1 Resource Non-Resource Note: 1 . Firm size is measured as average sales in 199 1 $US (million). Subsample 1 includes firms with average sales within ($400 million, $1000 million). Subsarnple 2 includes firms with average sales within ($1000 million, $8000 million). 2. Resource denotes firms of mining, minerais, oil and gas.

Table 3.4 Estimation of Pay-Performance Sensitivity (subsample 1 & 2 ) Dependent Variables --

(Arithrnetic Estimation)

-

-

-

.

(Elasticity Estimation)

Independent Variabtes

Intercept

CND ~(ShareholderWealth) CNDxa(Shareho1der Wealth) ~(ShareholderWealth)., CNDxa(Shareholder Wea1t.h)-, alog(Shareho1der Wealth) CNDxalog(Shareh1der Wealth) alog(Shareho1der Wealth)_, CNDxalog(Shareh1der Wealth).,

R' Observations Note: t-statistics are in parentheses. CND is a dummy variable that equaIs one for Canadian firms. The sample inchdes firms with average sales within ($400 miIIion, $8000 million)

Table 3.5

Estimation of Pay Performance Sensitivity (subsample 1) Dependent Variables (Arithmetic Estimation)

(Elasticity Estimation)

lndependent Variables

Intercept CND

A(Shareholder WeaIth)

CNDxa(Shareholder Wealth)

~(ShareholderWealth).,

CNDxa(Shareho1der Wea1th)-,

alog(Shareho1der Wealth)

CNDxalog(Shareh1der Wealth)

alog(Shareho1der Wealth)-,

CNDxalog(Shareh1der Wealth)-,

R' Observations Note: t-statistics are in parentheses. CND is a dummy variable that equals one for Canadian firms. Subsample I includes the fims with average sales within ($400 million, $1000 million).

Table 3.6

Estimation of Pay-Performance Sensitivity (subsample 2) Dependent Variables (Arithmetic Estimation)

(Elasticity Estimation)

Independent Variables

Intercept CND

~(Shareho lder Wealth)

CNDxa(Shareho1der Wealth)

~(ShareholderWealth)_,

CNDxa(Shareho1der Wealth).,

alog(Shareholder Wealth)

CNDxalog(Sharehlder Wealth)

alog(Shareholder Wealth),,

CNDxalog(Shareh1der Wealth)-,

R' Observations Note: t-statistics are in parentheses. CND is a dummy variable that equals one for Canadian fins. Subsarnple 2 include firms with average sales within ($1000 million, $8000 million)

Table 3.7 n i e US-Canada Differential and Firm Size (total sample) Dependent Variables Independent Variables

alog(Sa1ary +Bonus) (1

(2)

alog(Tota1 Pay)

(3)

(4)

Intercept CND CND x SIZE

alog(Shareho1der Wealth)

CNDxalog(Shareholder Wealth) CNDxSIZE xalog(Shareho1der Wealth) alog(Shareholder WeaIth).,

CND xalog(Shareh1der Wea1th)-, CNDx SIZE xalog(Shareho1der Wealth).,

R'

Observations Note: t-statistics are in parentheses. CND is a dummy variable that equals one for Canadian fims. SIZE is measured as firm's average sales in %million.

Table 3.8 CE0 Stock Ownership as the Percentage of Shares Outstanding Median

Mean

(%)

(W

Observation

Canadian CEOs Total sample Subsample 1 Subsarnple 2

Totai sample Subsarnple 1 Subsample 2 Note: The data are for fiscal year 1993 and 1994. Subsample 1 includes f i m s with average sales within ($400 million, §! 1000 million). Subsample 2 includes firms with average sales within (% 1000 million, $8000 miIIion).

Table 3.9 Regressions of C E 0 Stock Ownership Against Firm Size Dependent Variable: log(Shares in %) Independent Variables

Canada)

(US)

( Canada and US)

Intercept

CND log(Sales) CNDxlog(Sa1es)

R2

Observations Note: t-statistics are in parentheses. CND is a dummy variable that equals one for Canadian firms.

Table 3.10 The Ratio of Stock Gains fiom Exercising Previously Granted Options Over Total CE0 Pay US Firms

Canadian Fims The Ratio Observ.

The Ratio

Observ.

Total: sample Subsample 1 Subsarnple 2

Note: The data are for fiscal year 1993 and 1994. Subsample 1 inctudes fims with average sales within ($400 million, $1000 million). Subsample 2 indudes firms with average sales within ($1000 million, $8000 million). Total CE0 pay indudes direct payments and stock gains.

Table 3.1 1 Cornparison of Firm's Systematic Performance Ave Retum P!4) (Median) (Mean)

Monthly Rcturn (1987-1991)

Annuai

Std Deviation

(Median) (Mean)

Ave Retl Std Dev (Median) (Mean)

Observations

Çanadian Fims TotaI sarnple Submplc 1 Subsample 2

0.48 0.70 0.76

-1 2 3 -3.96

0.1 07 0.095 0.084

0.253 0.225 0.194

0.050 0.098 0.1 12

0.043 0.064 0.092

355 53 54

U S . Fims Total sample Subsampte 1 Subsample 2

1.19 1.1 1 1.16

1.44 1.37 1.44

0.091 0.095 0.091

0.102 O. 105 0.102

0.150 0.130 0.145

0.147 0.141 0.146

581 80 378

Çanadian F i r m ~ Total sample Subsarnple 1 Subsample 2

9.83 9.47 7.41

19.41 10.61 8.43

0.331 0.332 0.265

0.573 0.362 0.325

0.321 0.313 0.321

0.295 0.305 0.308

309 56 59

U.S. Firms TotaI sample Subsample I Subsarnple 2

13-00 15.65 12.13

16.35 19.89 16.33

0.254 0.286 0.250

0.357 0.385 0.373

0.505 0.544 0.493

0.517 0.565 0.504

635 91

-2.53

Return ( 1984-1994)

JO2

Note: 1. US monthly return is retneved from the CRSP stock return file. Canadian monthly retum is obtained from the return file of the University of Western Ontario's TSE database. Annud return is calculated fiom the COMPUSTAT files for both U.S. and Canadian firms. 2. Ave Ret and Std Dev are average return and standard deviation of the firm's stock retum respectively. 3. Subsample 1 includes firms with average sales within (5400 million, $1000 million). Subsample 2 includes fÏms with average sales within ($1000 million, $8000 million).

Table 3.12 Cornparison of CE0 Pay Levels -

Canadian CE0 Pay (Median)

(Mean)

-

U.S.CE0 Pay (Median)

(Mean)

Total sample Subsample 1 Subsample 2 Total Pav Total sample Subsample 1 Subsample 2

Note: The data are from 199 1 to 1994. AI1 variables are annual and in 199 1 US dollars (thousand).

Figure 3.1 Distribution of Firms

- Canadian -US

O

10

20

30 40 50 60 Average Sales ($1 00 million)

70

1

80

9

References Abowd, John M., "Does Performance Based Managerial Compensation Affect Corporate Performance?" Indmtnal and Labor Relations Review, 43, 1990, 52-73s. Chaykowski, Richard P. and Brian Lewis, Compensation Practice and Outcornes in Canada and the United States, IRC press, Queen's University, 1995. Ciscel, David H. and Thomas M. Carroll, "The Deterrninants of Emcutive Salaries: An Econometric Siuey," Review of Economic and Statzstzcs, 1980, 7-13. Cosh, A., 'The Remuneration of Chief Execiitives in the United Kingdom," Economic Journal, 85, 1975, 75-94 Coiighlan, Anne T. and Ronald M. Schmidt, ('Exectitive Compensation, Managerial nimover, and Firm Performance," Journal of Accounting and Economics. 7' 1985. 67-54.

Garen, John E., "E'ceciitive Compensation and Principal-Agent Theorf

.Journal of

Political Economy, 102, 1994, 1175-1199. Gibbons, Robert and Kevin bltirphy, "Optimal Incentive Contracts in the Presence of Career Concerns: Theory and Evidence," Journal of Political Economy, 100: 1992.

468-505. Gibbons, Robert and Kevin hIurphy, "Relative Performance E d u a t i o n for Chief Execiitive Officers," Industrial and Labor Relations Review, 43, 1990, 30-51s. Grossman, Stiuiford J. and Oliver D. Hart, "An Anidysis of the Principal-Agent Prob-

lem," Econometrica, 5 1, 1983, 7-46.

Hart, Oliver and Bengt H o h t r o m , "The Theory of Contracts," In Advances in Economic Theory FiRh World Congress, ed. by Tnunan F. Bewley, Cambridge: Carnbridge University Press, 1987. Haiibrich, Joseph G., "Risk Aversion, Performance Pay, and the Principal-Agent Prob-

lem," Journal of Political Economy, 102. 1994, 258-276.

Holmstrom, Bengt, "Moral Hazard and Observability," Bell Journal of Economics, 10, 1979, 7491. Hubbard, R. Glenn and Darius Palia, "Exectitive Pay and Performance Evidence from the US Banking Industry," Journal of Financial Economics, 39, 1995, 105-130. Jensen, bfichael C. and Kevin J. Miirphy, "Performance Pay and Topblanagement Incentives," Journal of Political Economy, 98, 1990, 225-264. Jobson, J.

D. and Bob M. Korkie, "Performance Hypothesis Testing with the Sharpe

and Treynor Measiiles," Journal of Finance, Vol. XXXVI. No. 4, 1981, 889-908. Joskow, Paul and Nancy Rose, "CE0 Pay and Firm Performance: Dynarnics, Aspunetries, and Alternative bfeasiires of Performance," NBER Working Paper No. 1976,

1994.

Joskow, Paiil, Nancy Rose, and Andrea Shepard, "Regdatory Constraints on CE0 Compensation, " with comments and discussions. Brookings Papers on Economic Actzvity

-

Micmeconomics, 1993, 1-72.

Kaplan, Steven N, "Top Execiitive Rewards and Firm Performance: -4 Cornparison of Japan and the United States," Journal of Polztical Economy, 102, 1994, 510-546. Kato, Takao and Mark Rockel, "Experiences, Credentials, and Compensation in the Japanese and U.S. Managerial Labor Markets: Evidence from New Micro Data," Journal of the Japanese and International Economics, 6, 1992: 30-51. Leonard, Jonathan S, "Execiitive Pay and Firm Performance," Industrial and Labor Relations Review, 43, 1990, 13-29s. Lewellen, Wilber G. and Blaine Huntsman, "blanagerial Pay and Corporate Performance," American Economic Review, 60, 1970, 710-20. Masson, Robert, "Execiitive Motivations, Earnings, m d Consequent Equi ty Perfor-

mance," Journal of Political Economy, 79, 1971, 1278-92. Mirrlees, J. A., "The Optirnd Structure of Incentives and Aiithority within An Organization," Bell Journal of Econornics, 7, 1976, 105-31.

Miirphy, Kevin J., "Corporate Performance and Managerid Remuneration, An Empirical Analysis," Journal of Accoimting and Economics, 7. 1985, 11-43. Rh-phy, Kevin J ., "Incent ives, Learning, and Compensation: A Theoretical and Empiricd Investigation of Managerial Labor Contracts," Rand Journal of Economics. 17(1), 1986, 69-77.

Rosen, Sherwin, "Contracts and the Market for Execiitives" with cornments by Bengt Holmstrom, in Contract Economics, ed. Lars Werin and Hans Wijkander, Massachrisetts: Basil Blackwell Ltd, 1992.

Ross, Stephen A., "The Economic Theory of Agency: The Principal's Problem," American Economic Review (Papers and Pruceedzngs), 63, 1973, 135- l39. Schaefer, Scott, T h e Dependence of C E 0 Pay-Performance Sensitivity on the Value of the Firm," kIaniiscript, Keilogg Graduate School of Management, Northwestern University, 1995. Smith, CLifford W. and Ross L. Watts, ('The investment opportiuiity set and corporate financing, dividend and compensation policies," Journal of Financial Economics. 32, 1992, 263-392.

IMAGE EVALUATION TEST TARGET (QA-3)

APPLIEO 4 IMAGE. lnc

-.---- Fa: ---

1653 East Main Street Rochester, NY 14609 USA Phone: 716/482-0300 716/288-5989

O 1993. AOplied Image. Inc.. All Rights Rescnved