Agency Costs and Efficient Contracting: Evidence from the Financial Services Sector *

 

 

 

 

Eli Talmor**

London Business School

 

and

 

James S. Wallace

Claremont Graduate University

 

 

 

 

April 2005

 

 

 

 

 

 

 

* We are thankful to Kuntara Pukthuanthong for capable research assistance and to Reddie Kotha, Gilad Livne, Dennis Oswald, Laura Starks, Ana Vidolovska, an anonymous referee and especially Peter Pope (Managing Editor) for very valuable suggestions. We are also grateful to seminar participants at the London Business School, London School of Economics, University of Chicago, Claremont-McKenna College and University of Exeter for helpful comments.

** Address correspondence to: Professor Eli Talmor, London School of Business, Regent's Park, London  NW1 4SA, United Kingdom; Tel: +44 (0)20 7262 5050 (office), +44 (0)20 7724 7875 (fax); email: etalmor@london.edu


Agency Costs and Efficient Contracting: Evidence from the

Financial Services Sector

 

 

Abstract

 

Financial research uses contracting theory to explain cross-sectional differences in compensation policy.  We broaden the analysis by incorporating details of corporate governance. We study financial sector data that provide a relatively homogenous setting to explore the relation between executive compensation, economic determinants, firm performance, ownership, and corporate board characteristics.  The analysis supports the hypothesis that compensation contracts are used to mitigate agency costs.  We also find evidence that the boards of directors provide a monitoring function and that executive equity holdings and a strong board each appear to be a substitutes for incentive remuneration and hence instrumental in aligning incentives and mitigating potential conflicts.  The evidence, taken together, is consistent with an efficient contracting theory.

 

Keywords:  Efficient contracting, executive compensation, corporate governance, financial services, agency theory.

 

 


 
Agency Costs and Efficient Contracting: Evidence from the Financial Services Sector

 

1.       Introduction

Recent financial scandals on both sides of the Atlantic have energized the ongoing debate over the separation between management and ownership in the modern corporation.[1]  Critics complain that corporate governance has failed the shareholder.  They cite the seemingly outlandish levels of CEO compensation, along with the apparent disconnection between executive compensation and firm performance.  These same critics claim that management entrenchment has led to their control of corporate boards, the very group that is charged with monitoring and rewarding management’s performance.

Researchers have exerted significant effort to tie principle-agent theory to firm characteristics in order to assess compensation practices (e.g., Garen [1994] and Lambert and Larcker [1987]).  In an important paper, Smith and Watts [1992] examine explanations for various corporate choices, among them compensation policy.  They find evidence suggesting contracting theory best describes cross-sectional variation in observed compensation policy.  Their study and the subsequent literature focus on economic factors, particularly firm size, investment opportunities and the regulatory environment.  Most of these studies abstract from corporate governance considerations.  We broaden this inquiry by examining additional potential determinants of compensation policy.  Specifically, we use contracting theory to explain the linkage between compensation policy and not only economic determinants, but also ownership and corporate governance determinants. We study the financial services sector by examining manually collected data on commercial banks, insurance companies, and brokerage and other non-depository institutions.  By limiting our study to one relatively homogeneous industry, we provide a reasonable control for unspecified factors that might affect compensation.

The recent spate of financial scandals has brought with it a public debate regarding such normative questions as the proper level and forms of executive compensation, and the ability of corporate boards to fulfill their fiduciary obligation to the shareholders.  Conyon and Murphy (2000) compared CEO compensation in the U.S and the U.K. and found U.S. CEOs to earn total compensation nearly twice their U.K. counterparts, largely due to a greater use of stock options.  While we avoid any normative prescriptions in this research paper, we do provide positive evidence to inform this debate.  In particular we find strong support for efficient contracting theory.  We interpret our results as evidence that compensation contracts are used as a means to mitigate potential agency costs.  We find evidence that the board of directors provide a monitoring function.  In addition, we find evidence that executive equity holdings and a strong board each appear to be a substitutes with incentive compensation for mitigating agency costs.

The remainder of the paper is organized as follows.  In Section 2 we review the literature underpinning the theory guiding our empirical analysis.  In section 3 we discuss our hypotheses developed from this theory.  We discuss our data collection procedure and our choice of research methodology in Section 4.  Section 5 presents our main findings.  Finally, a summary is provided in Section 6.

 

2.       Theories of Executive Remuneration

2.1       Agency Theory

Economic theory of executive pay has focused on the design of optimal compensation schemes to align the interests of hired managers and shareholders.  Agency theory has identified several factors by which these interests may differ; including the level of effort exerted by the manager and problems resulting from the unobservabilty of the agent’s relevant skills.  The design of optimal compensation contracts essentially trades-off between different incentive problems and risk-sharing considerations.

Research has also been directed to the identification of proper performance standards for evaluation and compensation. Both accounting and market return measures have been considered in the literature for top executive compensation, however no conclusion has been reached regarding which type is more appropriate.  From the shareholders’ perspective, return is generated from stock price changes.  However stock prices are subject to many events not directly controllable by management.[2]  Earnings-based incentives are often employed in compensation contracts because they are more controllable by management and therefore shield managers from much of the stock market fluctuations.  Earnings-based measures, however, lack a forward focus and suffer from lack of timeliness issues.  It is therefore common to see both market and earnings measures included in compensation contracts.  Indeed, the empirical literature has identified both a strong linkage between accounting measures of return and top executive compensation (e.g., Lewellen and Huntsman [1970], Sloan [1993] and Joskow and Rose [1994]), and a significant relation between CEO compensation and the stock market rate of return (e.g., Coughlan and Schmidt [1985], and Murphy [1985]). 

Agency theory tells us that the moral hazard problem will be larger where the manager’s actions are more difficult to observe.  In addition, we expect the conflict to be larger where the manager has a greater influence on firm performance.  Smith and Watts [1992] predict that CEOs in firms with larger investment opportunities are expected to have more skills and accordingly receive a higher level of pay. In addition, firms with larger investment opportunities are more difficult to monitor and therefore are more likely to link compensation to indicators of firm performance through incentive plans.[3]Prior empirical studies also report an association between the degree of regulation and manager compensation (Smith and Watts [1992], Joskow, Rose and Shepard [1993]).  Smith and Watts [1992] note that regulation restricts the manager’s discretion and therefore reduces the manager’s marginal product.  In the context of banking, the effect of regulation is examined in Crawford, Ezzell, Miles [1995] and Hubbard and Palia [1995].  Both studies report stronger pay-performance sensitivities following the move during the 1980s to deregulate banking services, in particular the 1980 Depository Institution Deregulation and Monetary Control Act, and legislation by nearly all states to allow interstate banking.  Thus, more highly regulated firms should pay lower compensation.  Moreover, since the regulation restricts the manager’s investment opportunity set, incentive compensation should be lower as regulation increases. 

 

2.2       Corporate Governance

The need for corporate governance arises since managerial employment contracts cannot fully resolve the agency problems from the separation of ownership and control.  Incomplete state verification and prohibitive costs make it unrealistic for shareholders to map the firm strategic choices and other managerial actions, thereby disallowing a sole reliance on a pure contracting specification to align interests.  Corporate governance instruments that are used to augment the managerial employment contract include an effective board of directors, the presence of large outside share blockholders, institutional share ownership, and the possibility of takeover.

Of the different corporate governance mechanisms, “the board is the shareholders’ first line of defense against incompetent management” (Weisbach [1988], p. 431).  This is particularly true in the financial sector where the high degree of regulation to protect depositors and insurance buyers, coupled with remaining barriers on ownership transactions limit the use of market-based monitoring devices relative to other industries.[4]  The implementation of board governance remains, however, a source of concern. Inside directors (i.e., directors who are also officers of the firm) are rarely in a position to challenge the CEO in the boardroom, and certainly are not expected to play a detrimental factor when setting the CEO compensation.  Outside directors, however, may be concerned with the company compensation structure in the interest of shareholders to solidify a reputation capital as competent board members. Weisbach [1988] provides empirical evidence that supports the view that insiders are detrimental to board independence. The National Association of Corporate Directors (NACD), in their Blue Ribbon Commission on director professionalism state, “Board independence is crucial to ensure that the board effectively carries out its mission and responsibilities, and fairly holds management accountable to shareholders.” (NACD 1996, p. 9).  The NACD goes on to recommend that boards require that independent directors fill the substantive majority of board seats.  In addition to being employed by the firm, relations that can compromise independence include director interlocks, and significant consulting or employment relations.

It is suggested that board effectiveness may depend on the director’s commitment and ability.  With regard to commitment, the NACD calls attention to the limitations of the calendar and to the limitations of the mind, which restrict the number of companies for which a director can maintain current knowledge (NACD 1996, p. 11).  The NACD further favors “a strong director evaluation process to assure the board members retain the necessary energy, enthusiasm, commitment, and creativity to forestall habitual or simply reactive – and therefore less effective – director participation.” (NACD 1996, p. 13).  Among the NACD recommendations is a mandatory retirement age.

The CEO’s power base widens when the CEO also serves as the chairman of the board.  Board size may also be related to board effectiveness (Jensen [1993], Yermack [1996]).  Jensen [1993] argues for smaller boards in a view he refers to as overcrowding of the board.  This attitude concerning the relative lower effectiveness of large board is also commonly expressed in the popular press.

 

2.3       Ownership Structure

In addition to board characteristics both external and executive ownership may also play an important monitoring role in corporate governance.  Whereas typical shareholders, because of their relatively small ownership interest, lack the incentive to actively monitor management, this may not be the case with large blockholders (Finkelstein and Hambrick [1988], and Core, Holthausen, and Larcker [1999]).  Increased monitoring should reduce any potential agency conflicts, thus lowering the need for equity incentives.  This is consistent with the finding in Mehran [1995] of a negative relation between equity-based compensation and blockholder ownership.

Managerial stock ownership is quite possibly the most complex determinant of compensation.  Morck, Shleifer and Vishny [1988] hypothesize that the relation between firm performance and the degree of managerial share ownership is non-monotonic.  At low levels managerial share ownership provides a better congruence of interest with outside shareholders.  As their share ownership increases, managers gain a tighter control and may engage in non-value maximizing activities.  However, at still higher ownership levels, such behavior has too high a personal cost and a closer realignment of objectives with outside shareholders is once again achieved.

Prior empirical literature has found equity’s incentive alignment effect to dominate, leading to a negative relation between executive stock ownership and the level of manager compensation (e.g. Core, Holthausen, and Larcker [1999]).  Mehran [1995] provides empirical evidence that equity based compensation is negatively related to CEO ownership, and therefore CEO equity ownership and incentive pay are substitutes for incentive alignment.

 

3.  Hypotheses development

Each of our hypotheses that follow is based on an overarching hypothesis that compensation contracts are used to mitigate agency costs.  Our first hypotheses deal with the relation between compensation and firm performance.  As noted above, theory and prior literature predicts a positive relation between compensation and firm performance.  Our study will consider two measures of firm performance, both the accounting measure of return on equity (ROE) and total shareholder return (RET) as determinants of compensation.  We predict a positive relation between firm performance and the level of total compensation.  The predicted association between firm performance and the incentive compensation is less clear.  However, since these equity incentives are intended to motivate managers toward improved firm performance, we predict a positive relation between firm performance and incentive-based compensation.

H1A     Ceteris paribus, a positive association will exist between firm performance and the level of CEO compensation.

 

H1B     Ceteris paribus, a positive association will exist between firm performance and the percentage of CEO incentive compensation.

 

The next three sets of hypotheses concern firm characteristics other than firm performance.  The agency literature noted above follows from agency theory that postulates larger moral hazard problems when monitoring becomes more difficult.  In this study we consider several situations that potentially affect the ability to monitor executive’s actions, company size, growth, and outcome variability.  We predict a positive relation between firm size, proxied by the natural log of assets (ASSETS) and the level of CEO compensation.  It has been argued that it is more difficult to monitor management in larger firms (Eaton and Rosen [1983], Smith and Watts [1982], and Sloan [1993]). Another important measurement of managerial discretion is the growth of operations.  We measures realized growth as actual sales growth over the prior three years (SALES3YR).  We predict each of these measures of managerial discretion will be positively associated with both the level of CEO compensation and equity holdings.

High discretion also suggests a greater variability of outcomes and therefore a riskier environment. We measure outcome variability as the standard deviation of return on equity (sROE) measured over the preceding five years.  Given risk aversion, we predict a positive association between the risk/outcome variability measure and the level of compensation.  In addition, since monitoring is more difficult under higher uncertainty, we also predict a positive relation between sROE and equity holdings.

H2A     Ceteris paribus, a positive association will exist between firm size and the level of CEO compensation.

 

H2B     Ceteris paribus, a positive association will exist between firm size and the percentage of CEO incentive compensation.

 

H3A     Ceteris paribus, a positive association will exist between firm growth/investment opportunity and the level of CEO compensation.

 

H3B     Ceteris paribus, a positive association will exist between firm growth/investment opportunity and the percentage of CEO incentive compensation.

 

H4A     Ceteris paribus, a positive association will exist between firm variability and the level of CEO compensation.

 

H4B     Ceteris paribus, a positive association will exist between firm variability and the percentage of CEO incentive compensation.

 

The next two hypotheses consider stock ownership, both external and internal.  As noted above in our literature review, theory predicts a negative relation between both the level of compensation and the incentive compensation mix and the potential monitoring by large external blockholders.  We note all cases where there exists an external party that holds at least a 5% share ownership and denote this variable BLOCKS.

H5A     Ceteris paribus, there will be a negative association between the existence of a large external blockholder and the level of CEO compensation.

 

H5B     Ceteris paribus, there will be a negative association between the existence of a large external blockholder and the percentage of CEO incentive compensation.

 

Much of the prior literature has included only actual stock ownership to proxy for the incentive effects of executive equity ownership and ignored stock options (e.g., Core, Holthausen, and Larcker [1999]).  We follow the procedure outlined in Core and Guay [1999, 2002], where the sensitivity of the managers’ option portfolio to a percentage change in the underlying stock price can be estimated.  We combine the sensitivity of the manager’s option holdings to their stock holdings, including beneficial ownership.  We define the variable OWNER as the total combined (stock plus options) sensitivity to a 1% change in stock price.  Our predictions are based on prior findings, noted above, that have found equity’s incentive alignment effect to dominate, leading to a negative relation between OWNER and the level of manager compensation along with empirical evidence that equity based compensation is negatively related to CEO ownership.

H6A     Ceteris paribus, a negative association will exist between CEO share ownership and the level of CEO compensation.

 

H6B     Ceteris paribus, a negative association will exist between CEO share ownership and the percentage of CEO incentive compensation.

 

Although lessening in extent in recent years, the degree of regulation varies among the three sub-groups within the financial sector.  While the past few decades have seen a trend in the U.S. toward deregulation away from the Glass-Steagall Act (e.g., the Gramm-Leach Biley Act of 1999), the banking sub-group of the 1990s still faced regulation concerning what services could be provided, as well as ownership restrictions.  Similar, yet to a lesser degree than in the banking sub-group, insurance companies operate under the watchful eyes of regulators.  Of the three sub-groups, brokerages and other non-depositories operate more like non-regulated industries.[5]  CEOs in more regulated industries typically have less discretion in their strategic choices such as types of products and entering new businesses.  We predict that both the level of total compensation and incentive compensation across the financial sector subgroups will be negatively related to the degree of regulation.  We create a variable (REGULATE) that is given the value 1 for the relatively most regulated banking sub-group, 2 for the insurance companies, and 3 for the relatively least regulated brokerage and non-depositories firms.

H7A     Ceteris paribus, a negative association will exist between the degree of regulation and the level of CEO compensation.

 

H7B     Ceteris paribus, a negative association will exist between degree of regulation and the percentage of CEO incentive compensation.

 

Our next hypotheses concern the relation between boards of directors’ strength and executive compensation.  In the prior section we discussed many characteristics of boards that have been theorized to impact board strength.  We next discuss how we make operational our measure of board strength.

We consider board strength to be a function of two general attributes, independence and efficiency.  Regarding impendence, we denote the variable INSIDE as the ratio of executive directors to the total number of directors.  Interlocked directors, denoted LOCK, are defined as inside directors who sit on the boards of outside directors’ firms.  Board members who are not full-time employees but affiliated with the company in another way, denoted GRAY, includes such individuals as consultants, lawyers and investment bankers or others who have a business relation with the firm, as well as directors with family ties to a company employee, usually the CEO.  Gray directors may be less independent than other outsider directors because of their family ties and business relations.[6]  Each of these potential impairments of board independence, insiders, interlocked directors, and gray directors, are predicted to be associated with a lower level of monitoring and hence a higher level of CEO compensation.  Lower levels of monitoring can potentially be mitigated through other means, such as incentive compensation.  We therefore predict sources of reduced board independence will be associated with increased use of incentive compensation.

Regarding efficiency, we have selected a threshold number of three boards to define a busy outside director (denoted BUSY).  To test the effect of aging on board effectiveness we denote OLD as the number of outside directors over age 65.  We define DUAL as a binary variable that indicates if a CEO duality exists.  We define the total number of board members as TOTAL. 

As with the potential impairments of board independence, each of the potential impairments to board effectiveness, BUSY, OLD, DUAL, and TOTAL, are predicted to be associated with a lower level of monitoring and a higher level of CEO compensation.  As expressed above, the lower level of monitoring can potentially be mitigated through compensation contracts with a larger degree of equity-based incentives.  We therefore predict sources of reduced board effectiveness will be associated with increased equity holdings.[7]

Because the degree of board strength can result from a large number of factors, we create an overall measure of board strength from a composite of these seven individual board characteristics.[8]  Our procedure for the construction of the board strength variable (BOARD) appears in the appendix.  BOARD is constructed such that a higher value represents a potentially more independent and effective board and therefore a theoretically stronger board.  Finally, we partition our sample into three groups ranked by the strength of their respective boards, WEAK, MIDDLE, and STRONG_BOARDS.

H8A     Ceteris paribus, a negative association will exist between Board strength and the level of CEO compensation.

 

H8B     Ceteris paribus, a negative association will exist between Board strength and the percentage of CEO incentive compensation.

 

The above hypotheses have all considered how various predictors affect the level of compensation.  These hypotheses are motivated by the outcry expressed over the relation between CEO compensation and that of the common workers, what many label outlandish.  One of the outcomes of these outcries is Internal Revenue Code Section 162(m) that limits the deductibility of non-performance-based compensation to $1 million.  While the principle intent of this law is to reign in the level of compensation, the secondary intent is to create a greater pay to performance relation.  Because the deductibility of only fixed compensation is limited, incentives are created to increase the amount of performance-based compensation relative to fixed compensation.

( Insert Table 1 here)

            Consider the case of Walt Disney Co. and its CEO Michael Eisner.  Eisner has enjoyed annual compensation during the 1990s as high as $200 million in 1996, however outrage over his compensation was rarely heard while Disney enjoyed strong stock market returns.  Disney’s stock has plummeted the last few years and we have seen a corresponding anger over Eisner’s compensation.  Interestingly, a lot more attention recently has also been focused on Disney’s perceived weak corporate board.  It seems that it is not only the level of compensation that is of concern, but it is also the pay for performance relation, or the sensitivity of the CEOs compensation to the firm’s performance.  This brings us to our last set of hypotheses. 

 

 

H9A     Ceteris paribus, CEO total compensation sensitivity will be positively associated with Board.

 

H9B     Ceteris paribus, CEO incentive compensation sensitivity will be positively associated with Board.

 

For convenience, we have included in Table 2 a listing of the variables discussed above along with our predictions of their effect on both executive compensation level and CEO incentive compensation.

(Insert Table 2 here)

 

4.                  Data and Research Methods

The data for the study includes all the 160 financial institutions for which executive compensation data is available in Standard and Poor’s ExecuComp database.  The sector is split into three subgroups using two-digit SIC codes: depository institutions (commercial banks and thrifts; SIC 6000-6099), insurance companies (SIC 6300-6399), and brokerages and other non-depository institutions (SIC 6100-6299).  The study uses data for the years 1992-1997.

The ExecuComp database includes all the compensation items from the annual proxy statement (Schedule 14a).  It lists separately all major forms of cash compensation: salary, bonus, payout from long-term incentive plans, and other annual compensation (such as perquisites, payments to cover taxes, etc.).  It also includes disclosed information regarding stock-based compensation: restricted stock awards, stock options grants and stock appreciation rights (option grants and restricted share ownership data are adjusted for stock splits occurring during the sample period).[9]  In addition, we gathered data on block ownership and board memberships directly from Schedule 14a.  Company financials are obtained from Standard and Poor's Compustat.

Our observations make up a panel of data with each executive/year comprising a row of data in the panel.  Each executive can be present in multiple rows, depending on the number of years the individual was CEO.  The columns in the panel represent the various compensation and predictor variables.  Panel data such as this presents two challenges to any analysis.  First, whether or not any executive’s compensation is observed is dependent on whether they remained in the CEO position.  Hence, there is a distinct survivorship bias in the data.  Second, providing the executive does appear in multiple rows because they did survive, these observations are not independent of each other.  Therefore the independent identically distributed errors assumption needed for OLS is violated and resulting parameter estimates may be biased.

Prior research has shown the need to correct for the potential selection bias through the use of a Heckman selection model (Renneboog and Trojanowski 2003).  The Heckman selection model assumes the following underlying regression equation (Heckman 1976, 1979):

yi = xjb + u1j

Unfortunately the dependant variable yi is not always observed.  Instead, it is only observed assuming the following selection equation:

zjg + u2j > 0

where:

            u1 ~ N(0,s)

            u2 ~ N(0,s)

            corr(u1, u2) = r

            Unless r = 0, the standard regression techniques, when applied to the underlying equation, yield biased results, whereas the Heckman selection model provides consistent, asymptotically efficient estimates of the model parameters (Stata 2003, p. 63).

            The second potential bias results from the violation of the independence assumption.  We utilize the cluster option within the Heckman model in order to obtain robust standard errors since the Heckman model supports the Huber/White/sandwich estimator of variance. (Stata 2003 p. 66).

            Our underlying model that we test is as follows (omitting year and firm subscripts):[10]

We assume that Comp is observed based on the following selection equation:

We assume that Comp is observed based on the following selection equation:

g0 + g1RET + g2Regulate x Board  + e2 > 0

where

Comp

=

either the natural log of the level of total compensation or incentive compensation.

ROE

=

return on equity.

Assets

=

the natural log of total assets.

s(ROE)   

=

the standard deviation of ROE.

Sales3yr

=

the three-year least square annual growth rate of sales.

Blocks

=

number of 5% or greater blocks of stock owned by outsiders.

Owner

=

sensitivity of CEO stock and option holding to a 1% change in stock price.

Bank   

=

an indicator variable set equal to 1 if the firm is a bank.

Brokerage

=

an indicator variable set equal to 1 if the firm is a brokerage or other non-depository.

Weak_Board

=

an indicator variable set equal to 1 if the firm is a brokerage or other non-depository.

Strong_Board

=

an indicator variable set equal to 1 if the firm has a strong board of directors.

Year

=

an indicator variable for each of the sample years 1992 through 1997.

Regulate x Board

=

The interaction of Regulate, a proxy for the relative regulation being faced by the firm’s industry, and Board, a composite board strength measure.

 

Several studies argue that it is not so much the level of pay, but the incentive effects of what the manager receives, that motivates managers (e.g., Jensen and Murphy [1990a]).   We report results for both a variable representing the level of CEO compensation (LEVEL) along with a variable constructed representing incentive compensation (INCENT).[11]

We choose to study the sensitivity of CEO compensation to the firm’s performance utilizing a framework motivated by Murthy (1999) where the dependent variable is natural log of the changes in annual compensation.  [MSOffice1] Our empirical model is the following:

 

where DComp is equal to either the natural log of the change in total or incentive compensation.

We again assume that DComp is observed based on the following selection equation:

g0 + g1RET + g2Regulate x Board  + e2 > 0

As mentioned above, we made use of the Standard and Poor's ExecuComp database in to obtain our compensation data.  LEVEL is the sum of the CEO’s annual salary, bonus, other annual pay, long-term incentive payout, restricted stock granted, and granted stock options. [12]   Incentive compensation (INCENT) is a subset of total compensation, comprised of bonus, long term incentive payout, restricted stock granted, and granted stock options.  In order to lessen the effect of scale differences, we transform our compensation variables to a natural log and use the transformed variables as dependent variables (Barro and Barro [1990], Sloan [1993]).

 

5.         Analysis of Reported Results

Table 3 displays corporate and compensation summary statistics for the financial sector and its three sub-groups.  Selected operational characteristics for all the groups are displayed in Panel A of Table 3.  All figures are cross-sectionally pooled across the sample period, thus representing aggregate statistics for the entire period.  Among the two size measures, there are no significant differences between the three groups based on sales, however banks are significantly larger based on assets. 

(Insert Table 3 here)

Among performance metrics, banks displayed significantly greater net income, however this may be due in part to size.  Brokerages and other non-depository institutions display the largest ROE.  Banks display the greatest shareholder return and growth in sales.

Panels B and C report summary statistics for the components of CEO compensation.  There is not a significant difference among the three sectors with respect to the level of salary.  This may in part be due to tax considerations since the firms are subject to Section 162(m), $1 million dollar deductibility limit on non-performance-based pay.  Clearly, in terms of total compensation level, the mean of brokerages and other non-depository institutions, 5.4 million dollars, is much higher than for banks and insurance firms (a ‘mere’ $3.5 million and $3.1 million, respectively).  This finding is consistent with contracting theory where the lower regulated sub-group displays the highest level of compensation.  Direct comparisons between sectors may be misleading, however, because of average size differences.  In terms of the percentage mix of pay, the breakdown between cash CEO compensation (salary, bonus and long-term incentive plan payouts (LTIP)) vs. stock-based compensation (options and restricted stock) is higher for banks and insurance firms and less for brokerages and other non-depository institutions.  The lesser-regulated brokerages and other non-depositories use less salary (24%) and more annual bonuses (46%) than the other two more regulated sectors, again consistent with contracting theory.

Panel D provides descriptive statistics of governance and external ownership blocks in the financial sector.  Three characteristics are noticeably different in bank board composition compared to boards of insurance and brokerages firms: First, total board members is significantly larger in banks (16 vs. 11).  Second, the fraction of officer representation on the board is smaller (mean of 16% compared to 36% in brokerages).  Third, the percentage of outside directors with interlocking directorships (i.e., an inside officer of the firm serves on that outside director’s Board) is much high for banks than for either of the two other sub-groups.

Depositories are also distinguished from the other financial institutions with respect to the external ownership variable.  On average there is only one outside block in banks, compared to two blocks for the two other sub-groups.

We next turn to the determinants of CEO compensation. Table 4 provides the results for our Heckman selection models looking at the level of CEO compensation.  We first consider the level of total compensation.  Total compensation is significantly associated with the both measures of firm performance, both the accounting measure ROE and the market measure RET.  This result is consistent with the literature, which has typically identified a strong linkage between firm performance and top executive pay.  Somewhat surprisingly, the level of significance is much higher for ROE than for RET.  This result may be partially explained by contracting theory that emphasizes one should be held accountable only for items under ones control.  While it can be argued that one-year shareholder return is at least partially under the control of the CEO, many other considerations factor into share prices.  The financial performance measure of Return on Equity, in contrast, is clearly more under the control of the CEO.

(Insert Table 4 here)

As predicted by contracting theory, larger firms (ASSETS) are also associated with a higher level of CEO compensation, apparently compensating them for their greater marginal product.[13]  Supporting our prediction, and consistent with theory that based on risk sharing higher volatility should be associated with a higher levels of compensation, s(ROE) is significant in the positive direction.  Also consistent with contracting theory, the proxy for the firm growth, SALES3YR, is positively associated with the level of CEO compensation.  This result is supportive of contracting theory’s assertion that increased managerial discretion is associated with higher levels of compensation.

Surprisingly, external share ownership (BLOCKS) did not have a significant effect on the level of CEO compensation.  At least for our sample of firm years, we do not find evidence that the external markets exert an influence of executive compensation through its monitoring role.

The CEO’s sensitivity to changes in the firm’s stock price, OWNER, proved statistically significant in the predicted negative direction.  CEOs whose equity holdings have a higher sensitivity to changes in the firm’s stock price appear to act more like owners and accept lower levels of compensation.  While we do not test for their unobservable wealth change, it may be that their lower compensation is offset by higher capital gains through share appreciation.

The next two variables represent the level of regulatory oversight.  The sample includes three categories of financial institutions, with regulatory oversight greatest in banks, next greatest in insurance companies, and least in brokerages.  Indicator variables are included for banks and for brokerages, with insurance firms absent.  Therefore the results of the two included groups are relative insurance companies.  As predicted, banks have the lowest level of compensation, significantly lower than insurance, and brokerages have the highest level of compensation, significantly greater than insurance.

Finally, we consider the corporate governance variables.  We have partitioned the sample firms into three groups, based on our measure of board strength (BOARD).  Recall that Board is a composite variable computed from seven individual variables proxying for board independence and efficiency.  We formed indicator variables representing weak, middle, and strong boards.  Our model includes variables presenting weak and strong boards, with the middle level absent.  Therefore the results for weak and strong boards are each relative to the middle level of board strength.  As predicted, stronger boards are associated with lower compensation after controlling for economic and external ownership compensation determinants, while weaker boards as associated with higher levels of CEO compensation.  This can be interpreted as evidence that the board of directors plays an important role in monitoring and rewarding the CEO. 

The second column in Table 4 displays the results of our model examining the CEO incentive-based compensation.  As predicted, both measures of firm performance display a significantly positive relation with incentive compensation.  Contracting theory posits that incentive compensation can mitigate agency problems when monitoring is more difficult.  We therefore predicted complex firms (ASSETS) with greater variability s(ROE), and greater growth opportunities (SALES3YR) will be more likely to use incentives to overcome the moral hazard conflict, leading to positive relations with INCENT.  We find support for this theory with each proxy variable. 

As was the case with LEVEL, external share ownership (BLOCKS) fails to displays a significant association with CEO equity-based incentives.  In contrast with external ownership, and as we noted above with LEVEL, the CEO’s sensitivity to changes in the firm’s stock price, OWNER, proved statistically significant in the predicted negative direction.  The incentive effect created by the CEOs equity holdings may act as a substitute for incentive compensation.    

Also consistent with our predictions, and with the results reported with LEVEL, the more heavily regulated category of banks are associated with lower CEO incentive compensation, with the least regulated sub-category, brokerages are associated with a greater level of incentive compensation.  Finally, as we predict, weak boards are associated with greater levels of incentive compensation, while stronger boards are associated with lower levels of incentive compensation.  These results on board strength can be interpreted as evidence that the board and equity-based incentives also serve as substitutes for controlling CEOs.

Our final set of models look at CEO pay to performance sensitivities.  Table 5 analyses the sensitivity of total CEO compensation to firm performance, while Table 6 considers incentive pay to performance sensitivity.  Each table is partitioned based on the level of board strength.  The variable of interest is the performance metric (ROE).  We predict that stronger boards will do a better job of insuring greater pay to performance sensitivity.  Our results provide support for this prediction, with strong boards revealing a significant positive pay to performance relationship.  Interestingly, firms with weak or moderate strength boards fail to report significant pay to performance sensitivity for their CEOs. 

(Insert Tables 5 and 6 here)

Table 6 results mirror those in the previous Table 5.  One would expect a stronger pay to performance relationship between incentive compensation and performance than between total compensation and performance since it is the incentive component of total compensation that is likely tied most closely with a performance metric.  The results in Table 6 bear out these expectations, with each level of board strength showing a higher significance level on the variable (ROE) than that shown in Table 5.  As with Table 5, only strong boards display a significant pay to performance relationship.        

 
6.       Summary

Our study is motivated by the recent high profile corporate financial scandals and is intended to inform the debate regarding the agency costs resulting from the separation between management and ownership within the modern corporation.  In this paper we perform a comprehensive study of the relation between executive compensation, firm performance, and corporate governance determinants.  Using data from the period 1992-1997, we examine the financial sector comprised of three sub-sectors: commercial banks, insurance companies, and brokerages and other non-depositories.

Several characteristics of our research distinguish it from prior work.  The majority of the prior studies in the financial services sector have focused on commercial banks.  Barro and Barro [1990] analyze the pay-for-performance relation for a sample of large commercial banks.  They find compensation is affected by firm performance, both accounting and stock returns.  In addition, they report that the sensitivity of the relation diminishes with the experience of the CEO.  Several papers look at the impact of regulatory changes within the industry on management compensation policy.  Crawford, Ezzell and Miles [1995] and Hubbard and Palia [1995] confirm that executive pay-performance sensitivity increases during the 1980s following deregulation of commercial banks (e.g., the 1980 Depository Institution Deregulation and Monetary Control Act and legislation by nearly all states to allow interstate banking).  Houston and James [1995] examine whether the form of compensation contracts in commercial banks promotes more risk taking.  The authors find lower use of stock-based compensation in banks than in other industries.  They interpret this evidence as being consistent with contracting theory but inconsistent with the incentive for increased risk taking resulting from depository insurance.[14]  For a sample of saving and loan institutions, Hermalin and Wallace [2001] show that allowing heterogeneity in the model specification results in a stronger correlation between firm performance on managerial compensation than that found in previous studies (notably, Jensen and Murphy [1990b]).[15] 

With regard to the broader literature on executive compensation, most prior research limits the analysis of compensation determinants to primarily economic variables (e.g., firm size, growth, risk) or manager specific variables (e.g., tenure).  We include additional variables that are believed to be associated with variation in compensation.  Specifically, we examine corporate governance structure (i.e., board characteristics) and external market monitoring (i.e., large blockholders).  Unlike most prior studies that only consider the level of executive compensation, we examine both total compensation and incentive effects of management’s equity ownership.  Regarding the latter incentive effects, we consider not only actual share ownership, we also incorporate option holdings.  In this regard we measure the combined effect of the executive’s stock holdings and the sensitivity of their option portfolio to changes in the underlying stock.

We find general support for our predictions.  We first look at the determinants of the level of CEO compensation.  Consistent with prior research, higher levels of the firm performance appear to be positively associated with higher pay.  As predicted by contracting theory, compensation appears higher in situations where the executive is harder to monitor.  Larger firms are also associated with a higher level of CEO compensation, apparently compensating them for their greater marginal product.  Supporting our prediction, and consistent with theory that based on risk sharing, higher volatility is associated with a higher level of compensation.  Also consistent with contracting theory that posits firms with larger investment opportunities are more difficult to monitor, we find a positive relation between sales growth and the level of CEO compensation.  With regard to monitoring, a strong Boards of Directors are negatively associated with the level of CEO compensation.  This provides evidence that they serve as a means to monitor executives. 

We next report findings with regard to executive incentive compensation.  These results are much like those explaining total compensation.  Consistent with contracting theory, larger firms, firms with greater variability, and firms with greater growth, are associated with greater CEO incentive compensation.  Finally we find a negative association between board strength and CEO incentive compensation.  We interpret this as evidence that board strength and equity incentives are substitutes for mitigating agency costs.  These findings, taken together, support an efficient contracting argument.

We further analyze executive pay to performance sensitivities since it is not be only the level of compensation that has caused such a concern among executive pay critics, but also a lack of pay to performance sensitivity.  We find support that stronger boards are associated with significant pay to performance sensitivity, however such a relationship fails to be seen in firms with weak or moderate strength boards.

Due to the need for hand collecting many of our governance variables, our study was limited only to one industry. While the exclusive use of the financial sector is potentially a limitation, it has also potential strengths.  Firstly, as noted above, the use of a single industry affords us a manageable sample.  With this sample we are able to gather the necessary data to provide a unified test encompassing theory and associated variables from many separate studies that looked at individual aspects of the executive compensation question.  Using a single industry also provides a sample of firms with relatively homogeneous operational characteristics, thus offering a reasonable control of unspecified factors.  We also do not know of any a priori reasons why this industry should differ from the general population of firms with regard to executive pay.  Finally, by isolating a characteristic that differs within the sample (i.e., regulatory environment), we are able to study hypotheses related to job complexity and corporate governance by testing across the three sub-groups within the industry.

 

Appendix: Constructing a composite board strength variable.

The variable (BOARD) represents a composite variable measuring overall board strength.  We give equal weight to each characteristic in the overall BOARD measure, however we also recognize the substitutability of measures within the board.  We accomplish this through the following procedure:

1.        Create a variable called INDEPENDENCE by summing the two mutually exclusive percentage variables INSIDE and GRAY.

2.        Partition the sample observations into quartiles based on INDEPENDENCE and assign a score of 1 to observations in the top quartile, a score of 0 to observations in the middle two quartiles, and a score of –1 to observations in the bottom quartile.  Higher scores represent greater board independence and therefore a theoretically stronger board.  Using the same strategy, the observations are partitioned on LOCK.

3.        Since the four remaining effectiveness measures are not mutually exclusive and are not of the same units, we cannot simply sum them to together.  Instead, we again partition the observations into quartiles based on each of the four effectiveness measures.  As in step 2, we assign scores based on quartile, however we assign a score of 1 to the upper quartile and –1 to the bottom quartile.  Higher numbers represent greater effectiveness and therefore a theoretically stronger board.

4.        Finally, we add together the score for both the independence variables and the effectiveness variables to create an overall board strength measure.  To recognize that these individual characteristics may be substitutes, we again partition the observations, this time based on the computed strength score.  We assign a score of 1 to observations in the top quartile, a score of 0 to observations in the middle two quartiles, and a score of –1 to observations in the bottom quartile where the positive value represents a theoretically stronger board.[16]


References:

Almazan, A. and J. Suarez (2003). “Entrenchment and severance pay in optimal governance structures,” Journal of Finance 58, pp. 519-547.

 

Antle, R. and A. Smith (1986). “An Empirical Investigation of the Relative Performance Evaluation of Corporate Executives,” Journal of Accounting Research 24, Spring, pp. 1-39.

 

Baber, W.R., S.N. Janakiraman, and S.H. Kang (1996). “Investment Opportunities and the Structure of Executive Compensation,” Journal of Accounting and Economics 21, pp. 297-318.

 

Baker, G.P. and B.J. Hall (2004). “CEO Incentives and Firm Size,” Journal of Labor Economics 22, pp. 767-798

 

Berle, A. and G. Means (1932). The Modern Corporation and private Property. Publisher

 

Barro, J.R. and R.J. Barro (1990).  “Pay, Performance and Turnover in Bank CEOs,” Journal of Labor Economics 8, October, pp. 448-481.

 

Baysinger, B.D. and H.N. Butler (1985). Corporate Governance and the Board of Directors: Performance Effects of Changes in Board Composition,” Journal of Law, Economics and Organization 1, Spring, pp. 101-124.

 

Bushman, R., Q. Chen, E. Engel, and A. Smith (1999). “The Sensitivity of Corporate Governance Systems to the Timeliness of Accounting Earnings,” Working paper, University of Chicago.

 

Ciscel, D.H. and T.M. Carroll (1980).  “The Determinants of Executive Salaries: An Econometric Survey,” Review of Economics and Statistics 62, pp. 7-13.

 

Conyon, M. and K.J. Murphy (2000). “The Prince and the Pauper? CEO pay in the United States and United Kingdom,” The Economic Journal 110, pp. 640-671.

 

Coughlan, A. and R.M. Schmidt (1985). “Executive Compensation, Managerial Turnover and Firm Performance: An Empirical Investigation,” Journal of Accounting and Economics 7, pp. 43-66.

 

Core, J.E. and W. Guay (1999). “The Use of Equity Grants to Manage Optimal Equity Incentive Levels,” Journal of Accounting and Economics 28, pp. 151-184.

 

Core, J.E. and W. Guay (2002). “Estimating the Incentive Effects of Executive Stock Option Portfolios,” Journal of Accounting Research 40, pp. 618-630.

 

Core, J.E., R.W. Holthausen, and D.F. Larcker (1999). “Corporate Governance, Chief Executive Officer Compensation, and Firm Performance,” Journal of Financial Economics 51, pp. 371-406.

 

Crawford, A.J., J.R. Ezzell, and J.A. Miles (1995). “Bank CEO Pay-performance Relations and the Effects of Deregulation,” Journal of Business 68, pp. 231-256.

 

Diamond, P. (1998). “Managerial Incentives: On the Near Linearity of Optimal Compensation,” Journal of Political Economy 106, pp. 931-957.

 

Eaton, J. and H. Rosen (1983) “Agency, Delayed Compensation and the Structure of Executive Renumeration,” Journal of Finance 38, pp. 1489-1505.

 

Fama, E.F. and M.C. Jensen (1983). “The Separation of Ownership and Control,” Journal of Law and Economics 26, pp. 301-325.

 

Fields, L.P. and D.R. Fraser (1999). “On the Compensation Implications of Commercial Bank entry into Investment Banking,” Journal of Banking and Finance 23, pp. 1261-1276.

 

Finkelstein, S. and D.C. Hambrick (1988). “Chief Executive Compensation: A Synthesis and Reconciliation,” Strategic Management Journal 9, pp. 543-558.

 

Garen, J.E. (1994). “Excessive Compensation and Principal-Agent Theory,” Journal of Political Economy 102, pp. 1175-1199.

 

Gaver, J. and K. Gaver (1993). Additional Evidence on the Association between the Investment Opportunity Set and Corporate Financing, Dividend, and Compensation Policies, Journal of Accounting and Economics 16, pp. 125-160.

 

Gibson, Dunn & Crutcher LLP (1999). The Gramm-Leach-Bliley Act, P.L. 106-102, Financial Services Modernization, Working Summary No. 4.

 

Hermalin, B.E. and N.E. Wallace (2001). “Firm Performance and Executive compensation in the Savings and Loan Industry,” Journal of Financial Economics 61, pp. 139-170.

 

Heckman, J. (1976). “The Common Structure of Statistical Models of Truncation, Sample Selection, and Limited Dependent Variables, and a Simple Estimator for Such Models,” Annals of Economic and Social Measurement 5, pp. 475-492.

 

Heckman, J. (1979). “Sample Selection Bias as an Estimation Error,” Ecomometrica 47, pp. 153-161.

 

Houston, J. F. and C. James (1995). “CEO Compensation and Bank Risk: Is Compensation in Banking Structured to Promote Risk Taking?” Journal of Monetary Economics 36, pp. 405-431.

 

Hubbard R.G. and D. Palia (1995). “Executive Pay and Performance: Evidence from the U.S. Banking Industry,” Journal of Financial Economics 39, pp. 105-130.

 

Jensen, M.C. (1993). “The Modern Industrial Revolution, Exit, and the Failure of Internal Control Systems,” Journal of Finance 48, pp. 831-857.

 

Jensen, M.C. and K.J. Murphy (1990a). “CEO Incentives- It’s Not How Much You Pay, but How,” Journal of Applied Corporate Finance Vol, pp. 36-49.

 

Jensen, M.C. and K.J. Murphy (1990b). “Performance Pay and Top-Management Incentives,” Journal of Political Economy 98, pp. 225-264.

 

John, K., A. Saunders, and L.W. Senbet (2000). “A Theory of Bank Regulation and Management Compensation, Review of Financial Studies 13, pp. 95-125.

 

Joskow, P.L. and N.L. Rose (1994). “CEO Pay and Firm Performance: Dynamics, Asymmetries, and Alternative Performance Measures,” MIT working paper, December.

 

Joskow, P.L., N.L. Rose, and A. Shepard (1993). “Regulatory Constraints on CEO Compensation,” Brookings Papers on Economic Activity – Microeconomics, 1, pp. 1-58.

 

Kane, E. J. (2002). “Using Deferred Compensation to Strengthen the Ethics of Financial Regulation,” NBER Working Paper W8399.

 

Ke, B., K. Petroni, and A. Safieddine (1999).  “Ownership concentration and sensitivity of executive pay to accounting performance measures: Evidence from publicly and privately-held insurance companies,” Journal of Accounting & Economics 28, pp. 185-209.

 

Lambert, R.A. and D.F. Larcker (1987). “An Analysis of the Use of Accounting and Market Measures of Performance in Executive Compensation Contracts,” Journal of Accounting Research 25, Supplement, pp. 85-125.

 

Leonard, J.S. (1990). “Executive Pay and Firm Performance,” Industrial and Labor Relations Review 43, pp.11s-29s.

 

Lewellen, W.G. and B. Huntsman (1970). “Managerial Pay and Corporate Performance,” American Economic Review 60, September, pp. 710-720.

 

Lewellen, W.G., C. Loderer, and K. Martin (1987). “Executive Compensation and Executive Incentive Problems: An Empirical Analysis,” Journal of Accounting and Economics 9, pp. 287-310.

 

Mayers, D. and C. Smith (1992). “Executive Compensation in the Life Insurance Industry,” Journal of Business 65, pp. 51-74.

 

Mehran, H. (1995). “Executive Compensation Structure, Ownership, and Firm Performance,” Journal of Financial Economics 38, pp. 163-184.

 

Milliron, J. (2000). “Board of Director Alignment and the Design of Executive Compensation Contracts,” PhD Dissertation, University of Chicago.

 

Morck, R., A. Shleifer, and R.W. Vishny (1988). “Management Ownership and Market Valuation--An Empirical Analysis,” Journal of Financial Economics 20, pp. 293-315.

 

Mueller, D.C. and S.L. Yun (1997). “Managerial Discretion and Managerial Compensation,” International Journal of Industrial Organization 15, pp. 441-454. 

 

Murphy, K. J. (1985). “Corporate Performance and Managerial Remuneration: An Empirical Analysis,” Journal of Accounting and Economics 7, pp.11-42.

 

National Association of Corporate Directors, (1996). “NACD Blue Ribbon Commission Report on Director Professionalism.”

 

Renneboog, L. and G. Trowanowski (2003). “The Managerial Labor Market and the Governance Role of Shareholder Control Structures in the UK,” Working paper, European Corporate Governance Institute.

 

Schaefer, S. (1998). “The Dependence of Pay-Performance Sensitivity on the Size of the Firm,” Review of Economics and Statistics 80, pp. 436-443.

 

Silverman, G. (2002) “Level Playing Field Still Elusive,” Financial Times, February 22.

 

Skeel, D.A. (1999). “The Market Revolution in Bank and Insurance Firm Governance: its Logic and Limits,” Washington University Law Quarterly 7, pp. 433-459.

 

Sloan, R.G. (1993). “Accounting Earnings and Top Executive Compensation,” Journal of Accounting and Economics 16, pp. 55-100.

 

Smith, C. and R. Watts (1982). “Incentive and Tax Effects of U.S. Executive Compensation Plans,” Australian Journal of Management 7, pp. 139-157.

 

Smith, C. and R. Watts (1992). “The Investment Opportunity Set and Corporate Financing, Dividends, and Compensation Policies,” Journal of Financial Economics 32, pp. 263-292.

 

Weisbach, M.S. (1988). “Outside Directors and CEO Turnover,” Journal of Financial Economics 20, pp. 431-460.

 

Yermack, D., (1996). “Higher market valuation for firms with a small board of directors,” Journal of Financial Economics 40, pp. 185-211.


Table 1

Walt Disney Co. firm performance and compensation of its CEO Michael Eisner

 

 

 

 

Year      SALARY      BONUS   Total Comp      ROE      Return

 

1993      750.00       0.00       759.67    13.791      4.731

1994      750.00    7268.81     10666.93    21.073      3.298

1995      750.00    8024.71     14778.11    22.701     49.034

1996      750.00    7900.00    202185.14    10.679     10.996

1997      750.00    9900.00     10653.82    11.783     28.293

1998      764.42    5000.00      5768.24    10.089     -5.048

1999      750.00       0.00       753.82     6.442      2.663

2000      813.46    8500.00     33546.57     4.082     48.224

2001     1000.00       0.00      1004.02     0.513    -50.967

2002     1000.00    5000.00      6092.89     5.360    -17.866
Table 2

Variable definitions and predicted effect on compensation.

 

 

 

Variable

Acronym

Definition

Predicted effect on the level of compensation

Predicted effect on the percentage of incentive compensation

Return on equity

ROE

Net income divided by average total equity

+

+

Shareholder return

RET

Dividends plus stock appreciation divided by beginning of the year share price (adjusted for dividends)

 

+

 

+

Natural log of assets

ASSETS

The natural log of total assets

+

+

Standard deviation of ROE

s(ROE)

Standard deviation of return on equity over trailing five years

+

+

3 year growth in sales

SALES3YR

Three year least square annual growth rate of sales

+

+

CEO ownership

OWNER

Sensitivity of CEO stock and option holding to a 1% change in stock price.

 

-

 

-

5% shareholder blocks

BLOCKS

Number of 5% or greater blocks of stock owned by outsiders

-

+/-

Regulation

REGULATE

An indicator variable given the value 3 for banks, 2 for insurance companies, and 1 for brokerages and other non-depositories.

 

-

 

 

-

 

Bank

BANK

An indicator variable with the value of 1 if the observation is from a Bank.  Banks represent the highest regulated of the three financial sector groups.  Banks are defined as firms with SIC from 6000-6099

 

-

 

-

Insurance

INSURANCE

An indicator variable with the value of 1 if the observation is from an insurance company.  Insurance companies represent the middle level of regulation of the three financial sector groups.  Insurance firms are defined as firms with SIC from 6300-6399

 

 

Brokerage and other Non-depository

BROKERAGE

An indicator variable with the value of 1 if the observation is from a Brokerage and other Non-depository.  Brokerage and other Non-depository represent the least regulated of the three financial sector groups.  Brokerages are defined as firms with SIC from 6200-6299 and other non-depositories are defined as firms with SIC from 6100-6199

 

 

+

 

 

+

Board Strength

BOARD

A composite board strength variable made up of the below seven variables.  BOARD is constructed such that a larger value represents a theoretically stronger Board

 

 

-

 

 

-

Weak Boards

WEAK

The lowest third of the observations ranked on board strength

 

+

 

+

Strong Boards

STRONG

The top third of the observations ranked on board strength

 

-

 

-

Interaction of regulation and board

REGULATE x BOARD

The interaction of regulation on board strength

 

-

 

-

Insiders on the Board

INSIDE

Percentage of the Board’s Directors that are executives of the Company

+

+

Director Interlocks

LOCK

Percentage of outside directors where an inside officer of the firm serves on that outside director’s Board

 

+

 

+

Gray Board members

GRAY

Percentage of the Board’s Directors that are either family members of the CEO or have contracts with the firm

 

+

 

+

Busy Board members

BUSY

Percentage of the Board’s Directors that are outside directors and sit on at least two other Boards

 

+

 

+

Old Board members

OLD

Percentage of the Board’s Directors that are outside directors and over 65

 

+

 

+

Joint CEO and Chairmanship

DUAL

Percentage of CEOs that are also the Chairman of the Board

+

+

Total Board size

TOTAL

Total number of Board members

+

+

Total compensation

LEVEL

Salary, other annual, annual bonus, LTIP, restricted stock granted, and stock options granted

Incentive compensation

INCENT

Incentive-based compensation consisting of annual bonus, restricted stock grant, and stock options granted

Total compensation sensitivity

DLEVEL

The annual change in total compensation

Incentive compensation sensitivity

DINCENT

The annual change in incentive compensation

 


 Table 3

Descriptive statistics on firm, and CEO characteristics, data for 76 banks, 59 insurance firms, 25 brokerages and other non-depository institutions.

 

Panel A:  Firm Characteristics:

 

Banks

Insurance

Other

 

Mean

Median

Mean

Median

Mean

Median

Sales (M$)

4,020

2,214

4,651

2,466

4,824

1,867

Assets (M$)

48,097**

28,945

20,673**

7,138

45,973

9,605

Net Income (M$)

525.9**

270.2

356.3**

213.5

351.67

193.31

Return on Equity (ROE)

15.40

16.07

13.78**

13.85

19.86**

19.16

Share Return (RET)

32.57*

30.84

25.53**

25.44

31.39

27.20

3 Year Growth in Sales (SALES3Yr)

10.16**

9.29

13.94*

10.38

13.32

12.84

 

 

Panel B:  Components of CEO Compensation

 

Banks

Insurance

Other

 

Mean

Median

Mean

Median

Mean

Median

Salary (000$)

846.99

786.64

804.36

738.99

866.03

734.71

Bonus (000$)

933.79

501.44

640.16**

433.50

2,548**

1,513

LTIP (000$)

229.20

0.00

265.52

0.00

182.14

0.00

Restricted Stock (000$)

381.41

0.00

116.83**

0.00

625.09*

0.00

Options (000$)

1,099

566.40

1,232

586.15

1,204

209.00

Total

3,490

 

3059**

 

5,425**

 

 

Panel C:  Compensation Percentages

 

Banks

Insurance

Other

Salary

36.02

38.19*

28.30**

Bonus

26.52

21.54**

42.59**

LTIP

5.79

6.64

2.67**

Restricted Stock

7.07*

3.44**

8.50

Options

24.60

30.19**

17.94**

 

100.00

100.00

100.00

 


Table 3 (Continued)

 

Panel D:  Governance Characteristics

 

Banks

Insurance

Other

 

Mean

Median

Mean

Median

Mean

Median

DUAL (%)

90.33**

100.00

77.39**

100.00

86.59

100.00

Total Board Members (TOTAL)

15.79**

16.00

10.95**

11.00

12.57

11.50

Insiders (INSIDE) (% of Total)

15.58**

13.33

21.19

16.67

36.15**

30.00

Gray (GRAY) (% of Total)

4.53

0.00

4.84

0.00

2.79**

0.00

Busy (BUSY) (% of Total)

46.11

50.00

44.44

46.15

38.70*

37.50

Old (OLD) (% of Total)

23.76*

21.05

21.48

20.00

20.84

20.00

Interlock (LOCK) (% of Total)

15.00**

0.00

6.09*

0.00

3.66**

0.00

Outside Blocks (BLOCKS)

1.02**

1.00

2.00**

2.00

1.88**

2.00

 

Variable definitions: See Table 1

 

** Significant at the 1% level

  * Significant at the 5% level

 

The significance tests are based on two-sample t-tests of the mean difference where each group is compared to All Financial firms with the respective group of firms deleted. 


 Table 4

Determinants of CEO compensation.  Heckman selection model; z-statistics in parentheses. Year indicator variables included but not shown.

 

 

 

All Financial Services Firms

Independent Variable

LEVEL

INCENT

INTERCEPT

3.658

(13.85)***

3.570

(14.69)***

ROE

0.025

(4.26)***

.027

(5.48)***

RET

0.001

(1.28)*

0.002

(1.88)**

ASSETS

0.355

(18.45)***

0.325

(15.55)***

s(ROE)

0.030

(5.23)***

0.020

(4.21)***

SALES3YR

0.009

(2.91)***

0.004

(1.82)**

BLOCKS

0.023

(0.34)

0.046

(0.81)

OWNER

-1.45e-5

(-16.93)***

-1.1e-5

(-13.58)***

BANK

-0.250

-4.05)***

-0.333

(-4.65)***

BROKERAGE

0.439

(6.11)***

0.372

(5.62)***

WEAK_BOARD

0.127

(2.20)**

0.155

(2.66)***

STRONG_ BOARD

-0.133

 (-1.84)**

-0.046

(-0.74)

Selection Variables

 

 

INTERCEPT

1.807

(9.12)***

1.625

(2.34)***

RET

0.010

(3.35)***

0.001

(0.14)

REGULATE x BOARD

0.008

(0.21)

0.383

(0.95)

N

525

485

Wald test of independent equations chi2  (p-value)

29,220

(0.00)

1.3e5

(0.00)

 

Variable definitions: See Table 1

 

*** Significant at the 1% level, one-tailed

  ** Significant at the 5% level, one-tailed

    * Significant at the 10% level, one-tailed

 


Table 5

Determinants of CEO total compensation sensitivity.  Heckman selection model; z-statistics in parentheses.  Year indicator variables included but not shown.

 

 

Board Strength

Independent Variable

WEAK

MODERATE

STRONG

INTERCEPT

0.379

(1.06)

0.956

(0.36)

-0.106

(-0.22)

ROE

-0.008

(-1.22)

0.004

(0.88)

0.018

(2.47)***

ASSETS

0.023

(0.54)

0.013

(0.62)

0.016

(0.32)

s(ROE)

-0.030

(-3.06)***

-0.009

(-0.95)

0.006

(0.60)

SALES3YR

-0.012

(-2.28)**

-0.001

(-0.07)

0.005

(0.96)

BLOCKS

-0.044

(-0.43)

-0.065

(-0.94)

0.023

(0.17)

OWNER

-4.42e-5

(-3.34)***

-8.31e-7

(-1.07)

-3.30e-7

(-0.23)

BANK

0.183

(1.01)

-0.118

(-0.88)

0.718

(3.43)***

BROKERAGE

0.025

(0.21)

-0.035

(-0.42)

0.084

(0.59)

Selection Variables

 

 

 

INTERCEPT

1.079

(2.33)***

1.706

(7.39)***

1.121

(3.10)***

RET

0.007

(1.38)*

-0.001

(-0.59)

-0.002

(-0.67)

REGULATE x BOARD

-0.313

(-1.82)**

-0.272

(-5.84)***

-0.113

(-2.21)**

N

113

283

150

Wald test of independent equations chi2 (p-value)

7.17

(.007)

0.50

(.481)

8.38

(.004)

 

Variable definitions: See Table 1

 

*** Significant at the 1% level, one-tailed

  ** Significant at the 5% level, one-tailed

    * Significant at the 10% level, one-tailed


Table 6

Determinants of CEO incentive compensation sensitivity.  Heckman selection model; z-statistics in parentheses.  Year indicator variables included but not shown.

 

 

Board Strength

Independent Variable

WEAK

MODERATE

STRONG

INTERCEPT

0.242

(0.79)

0.173

(1.04)

-0.128

(-0.48)

ROE

0.003

(0.44)

0.003

(0.92)

0.012

(3.72)***

ASSETS

0.088

(0.25)

-.002

(-0.14)

-0.176

(-0.80)

s(ROE)

-0.009

(-1.05)

0.004

(0.55)

-0.099

(-1.60)*

SALES3YR

-0.005

(-1.19)

-0.002

(-1.30)

-0.002

(-0.51)

BLOCKS

0.014

(0.18)

-0.008

(-0.16)

-0.116

(-2.07)**

OWNER

-5.21e-5

(-6.78)***

4.56e-7

(1.03)

-8.87e-7

(-1.13)

BANK

0.026

(0.20)

-0.088

(-0.87)

0.228

(1.64)**

BROKERAGE

-0.053

(-0.46)

-0.150

(-2.92)***

0.231

(4.00)***

Selection Variables

 

 

 

INTERCEPT

0.997

(2.41)***

1.801

(8.10)***

1.127

(3.63)***

RET

0.005

(0.99)

-0.002

(-0.57)

-0.002

(-0.57)

REGULATE x BOARD

-0.254

(-1.84)**

-0.284

(-6.23)***

-0.135

(-2.87)**

N

113

283

150

Wald test of independent equations chi2 (p-value)

3.04

(.081)

0.41

(.523)

63.18

(.000)

 

Variable definitions: See Table 1

 

*** Significant at the 1% level, one-tailed

  ** Significant at the 5% level, one-tailed

    * Significant at the 10% level, one-tailed


 Endnotes:



[1] Berle and Means (1932) was perhaps the first to raise the concern that is now categorized as the agency costs of having corporations run by self-interested managers removed from the corporation’s owners.

[2] It is possible to devise a performance measure that controls for general market movements.  However, most executive compensation studies find that relative performance does not affect compensation (Antle and Smith [1986], Murphy [1985], and Barro and Barro [1990].  Also, from an analytical standpoint, Diamond [1998] argues that it is better to ignore relative performance.

[3] For further empirical evidence on the link between the investment opportunities set and compensation, see Gaver and Gaver [1993], Baber, Janakiraman and Kang [1996] and Mueller and Yun [1997].

[4]  The role of bank regulatory agencies is to ensure bank safety and soundness, not to monitor managers on behalf of shareholders.  The only banking regulation that applies to the structure of the board of directors is the FDIC Improvement Act of 1991 that was triggered by the savings and loan and bank crises of the 1980s.  This Act requires audit committees of bank boards be composed of non-officer directors.

[5] See Skeel [1999] and Gibson, Dunn, and Crutcher LLP [1999] for a discussion of regulation in the financial services sector.  A recent Financial Times article discusses how regulation reforms have yet to level the playing field between commercial banks, investment banks, and insurance companies (Silverman [2002]).

[6] We classify a gray director who holds a block of five or more percent of shares as an outside director.  Our reasoning is that a sizable ownership interest is more significant than considerations for past or affiliated business relations. We do not, however, apply this rule to family directors.

[7] Fama and Jensen [1983] provide a counter-argument regarding busy board members when they argue that outside board members who hold multiple directorships have greater incentives to monitor corporate decisions because of their reputation capital as decision experts.  In addition, Baysinger and Butler [1985] provide a counter-argument with regard to total board size when they argue that corporate boards have a variety of responsibilities in addition to monitoring management.  Hence, a larger board may offer a more diverse set of talents to deal with these responsibilities, thereby increasing the board’s overall effectiveness

[8] Examples of studies that have created aggregate measures of board strength include Bushman, Chen, Engel and Smith [1999] and Milliron [2000].

[9] We note that compensation reported in the ExecuComp database for executives who served as CEOs for part of a year is not distinguished from those who served a full year and is thus distorted.  Clearly, if a CEO was appointed in October, the disclosed compensation is seriously biased downward.  On the other hand, turnovers are occasionally associated with upward biases since the compensation at that time may include unusual items (such as golden parachutes) of extraordinary magnitude.  We identify all cases of CEO succession and remove the company year from our sample data.

[10] One possible approach would be to include REGULATE and BOARD as defined above as one of our predictor variables in explaining executive compensation.  We choose instead to include indicator variables representing for relative regulation (e.g., BANKS for most regulated) and board strength.  This allows us to see whether the degree of regulation and the degree of board strength matters, but whether the effect is at both the high and low end of the relative scale.  In addition, prior research has demonstrated that there is an interaction between certain of the variables that we hypothesize as determinants of executive compensation (Almazan and Suarez 2003).  Within the current study, consider that when boards are strong but regulation is weak, stockholders will take steps to compensate for the weaker regulation.  We would like to control for this suggested interaction between governance and regulation by providing an interaction variable of board strength and regulation (REGULATE x BOARD).  Unfortunately there is a very high degree of correlation between the interacted variable and BOARD.  This precludes us from being able to include both the main effects and the interactive effects in the same model.  We chose to include the main effects with the knowledge that the main effects on the board strength variables are also picking up the interactive effects.  For parsimony, we include the interacted variable rather than the main effects in the selection model.

[11] Lewellen, Loderer, and Martin [1987] and Mehran [1995] examine the structure and design of corporate executive compensation packages.

[12] The value of stock options granted is based on a modified Black-Scholes valuation.

[13] The positive association between size and CEO compensation is one of the most consistent findings across compensation studies.  See, for example, Ciscel and Carroll [1980], Leonard [1990], Schaefer [1998], and Baker and Hall [2004].

[14] John, Saunders and Senbet [2000] have analytically linked bank capital regulation to management compensation and derive pricing of FDIC insurance premium that mitigate bank risk shifting incentives.  Along similar lines, see Kane (2002) on the motivation for regulators to impose incentive-based deferred compensation for CEO’s of deposit insured institutions.

[15] Relatively few papers examine financial institutions other than commercial banks. Ke, Petroni and Safieddine [1999] and Mayers and Smith (1992) study the insurance industry, Fields and Fraser (1999) examine the pay-performance in bank holding companies with securities activities around their introduction period and compare it to a handful sample of investment banks).  In contrast to these prior studies, we consider the three segments of the financial industry in a unified test in order to capture cross-sectional differences.

[16] The above procedure assumes an equal weighting on each component of overall board strength.  Because the true weights are not observable, nor do we have a priori information as to their actual weights, we test the sensitivity of the results to our assumption of equal weights.  We separately reran the tests with each component of BOARD receiving a weight twice that of the other components.  The results with regard to BOARD remained robust.


 [MSOffice1]Corrected after draft sent to Eli