The firm determinants of CEO compensation in the U.S.
banking sector
Amsterdam Business School
Name Laura Alberts
Number 10190090
BSc Economics & Business Specialization Finance & Organization Supervisor Martin Koudstaal Completion 20-2-2015
Abstract
This research is conducted to find the determinants of CEO compensation The dataset consists of 144 U.S. banks in the period 1992-2013 with a total of 1208 observations. Agency theory predicts that performance measures should be used to determine the variable wage of the CEO, to align the CEO’s interests with the interests of the shareholders.. However, the existing literature does not find one concluding answer on the prediction that the performance measures are used as determinants for CEO compensation. The results are that base salary is positively related to Return on Assets (ROA) and after the crisis this relation is less strong. Total Shareholder Return (TSR) is also negatively related to base salary. Cash compensation is positively related to Return on Assets (ROA), and stronger related than after the crisis. Total Shareholder Return (TSR) is less strongly related to cash compensation after the crisis. Equity-based compensation is positively related by Return on Assets (ROA) and negatively related to Total Shareholder Return (TSR). After the crisis, ROA is less positively related and TSR less negatively related. Total compensation also has a positive relation with Return on Assets (ROA), after the crisis this relationship is not as strong as before the crisis. The performance measures TSR and ROA should be used as a measurement for variable compensation to incentivize and monitor the CEO. Principals should optimize the contract by letting the variable compensation depend more on the performance measures to reward the CEO.
Table of Contents
1. Introduction ... 2
2. Literature review ... 2
2.1 Agency Theory ... 3
2.2 Prior research: Literature review ... 4
3. Hypothesis, Methodology and Data ... 8
3.1. Hypothesis and Methodology ... 9
3.2. Data and descriptive statistics ... 12
4. Empirical result ... 14
5. Conclusion and discussion ... 18
1. Introduction
Since the influential work of Jensen and Meckling (1976), CEO compensation has been studied a lot. However the outcomes of these studies have not been identical. According to Chen, Steiner and Whyte (2005), during the period of 1992-2000 due to deregulation, banks increased the stock option-based executive compensation which induces risk-taking. During the period of 1993-2003 the increase of compensation for executives has gone further than what could be explained by the increase in the market (Bebchuck & Grinstein, 2014). During market booms there are lots of opportunities to run off to a better paid CEO-function. Therefore the offered compensation package will be high in these times. During market busts, these compensation packages are lower (Hubbard, 2005). Gregg, Machin and Szymanski (1993) stressed that boards are generally failing in their attempt to design remuneration schemes which properly align the interest of the executive with that of the shareholder. Several researches have shown a positive relationship between firm performance and total compensation of the CEOs in the U.S. The main difference between banks and companies is that banks have to work within the constraints set by regulators (Chen, Steiner and Whyte, 2005). The debt-ratio is also higher at banks than at companies. Not only should CEO compensation based on incentives to act on behalf of the shareholders but also on debt holders (John & Qian, 2003). There has been a lot of research on this matter. However, the existing literature does not provide a clear answer on the pay for performance relationship. Not only do articles use different data, they also use different performance measures which causes the overall conclusion to be unclear. In this paper the research question is: What are the determinants of CEO pay and firm performance in the U.S. banking industry and has the financial crisis influenced this relationship? The results are that base salary is positively related to Return on Assets (ROA) and after the crisis it is less positively related. Total Shareholder Return (TSR) is negatively related to base salary. Cash compensation is positively related to Return on Assets (ROA), after the crisis it is less strongly related. Total Shareholder Return (TSR) is negatively related to cash compensation after the crisis. Equity-based compensation is positively related by
Return on Assets (ROA) and negatively related to Total Shareholder Return (TSR).
compensation also has a positive relation with Return on Assets (ROA), after the crisis this relationship is weaker. Overall, the performance measures TSR and ROA should be used as a measurement for variable compensation to incentivize and monitor the CEO. Principals should optimize the contract by letting the variable compensation depend more on the performance measures to reward the CEO. In this paper the period after the crisis is studied as well as the effect after the crisis of performance measures on the compensation components. This has not been studied in previous literature and therefore this paper is an addition to the existing literature. This paper is organised as follows, first there is a theoretical background including the Agency theory and a literature review. Then the methodology, hypothesis and the data and descriptive statistics will follow. Third, the empirical results and the robustness check, and last the conclusion and discussion.
2. Theoretical background
2.1 Agency Theory
A. Smith (1776) is perhaps the first one who wrote about the agency theory in his book. He stated that one could not expect that the director of a private company, watches over the investors’ money as if it was his own.
The agency theory is defined by Jensen and Meckling (1976) as “a contract under which one or more persons, the principal(s), engage another person, the agent, to perform some service on their behalf which involves delegating some decision making authority to the agent” (p.310).
The principal-agent model contains three parts, available surplus, conflict of interest and asymmetric information (Hendrikse, 2003). If there is no available surplus, it means that the principal(s), in this case the shareholders, are paying the agent, in this case the CEO, the costs of the effort he puts into the task. If there is an available surplus, it means that the CEO is paid more that the costs of his effort if the shareholders are willing to. Conflict of interest means that the CEO will act on his own best interests instead of the best interests of the shareholders, which is
maximizing firm value, and these interests are not always aligned. Asymmetric information can be divided into two parts: hidden action and hidden characteristics Hidden action means that means that only the CEO knows the amount of effort he puts into his tasks. The shareholders cannot perfectly observe the CEO’s effort but
they can observe the firm value. The CEO is then able to choose an effort level that is lower than the effort level which maximizes firm value. This is also called moral hazard. Hidden characteristics means that the shareholders have incomplete
information about the CEO’s characteristics and motives. Only certain types of CEOs will sign the contract. This is also called adverse selection.
There are two ways to align the interests of the shareholders and the CEO. First, the principal can monitor the agent. Due to monitoring, the agent is more likely to act on behalf of the principal, so the principal controls the agent’s behaviour (Eisenhardt, 1989 p.59). In this case the board of directors can monitor the CEO to control the effort he puts into his tasks (Fama & Jensen, 1983 p.311).
Second, the principal can bind the agent by a contract, based on observable measures to maximize the firm value (Hendrikse, 2003, p. 97). Due to dispersed ownership there is no shareholder who holds the majority of the shares and therefore no shareholder who has the power to influence the executives decisions. These shareholders all hold a diversified portfolio and are therefore risk-neutral. Usually CEOs are risk-averse and this difference in the amount of risk-taking will lead to residual costs.
The theory predicts that pay and performance should be positively related to each other. Therefore the hypothesis of this investigation is that variable pay and firm performance are positively related.
2.2 Prior research: Literature review
There has been a lot of research in CEO compensation. However, in the literature, there is contradicting evidence on this matter. One of the foundational papers on this matter is the article written by Jensen and Murphy (1990).
Jensen and Murphy (1990) suggest that equity-based compensation rather than cash compensation gives managers the correct incentive to maximize firm value. But there is little empirical evidence on whether corporations whose executive
compensation is more equity-based actually perform better. They do find a small positive relation between CEO pay and firm performance. With every $1000,- change in shareholder wealth, CEO wealth will change with $3,25. This is a positive link but it is small for what one would expect from the role of the CEO. Also, the elasticity of CEO compensation and market firm value is 0,1, which is smaller than expected.
Their dataset consists of USA firms over a period of 50 years (1930-1980). The performance measure used in their article is shareholder wealth and increase in shareholder wealth.
Contrary to the results of Jensen & Murphy’s paper (1990), Hall and Liebman (1998) have used a panel dataset of large publicly trading U.S. firms over the years 1980-1994 to analyse the CEO holdings in stock and options. They do not investigate the determinants of CEO compensation but the responsiveness of pay for
performance. They find that the sensitivity of CEO compensation and firm performance is strongly driven by the holdings of stock and options and that this sensitivity has risen from 1.2 to 3,9 over the over the period of 1980-1994. This increase is due to the fact that total CEO compensation has risen by, 209% on average, mostly granted by stock and options. The performance measure they have used is the firm’s stock market value, which is stock price of the firm. Hall and Liebman (1997) confirm that base salary and cash bonus have a low sensitivity with respect to firm performance. They address the problem of pay for performance relationship. The reason for CEO pay relating to performance is to align the CEO’s incentives with those of the shareholders. With a CEO having stock and options he/she acts on behalf of the shareholders. But a highly correlated relationship of pay and performance may induce risk aversion, which will lead to not investing in
positive high risk, high return projects which would be beneficial for the wealth of the shareholders.
Also Mehran (1995) confirms the relationship between equity-based
compensation and firm performance. In his investigation, he focuses on the structure rather than the level of compensation. The data includes 153 randomly selected manufacturing firms conducted over the period of 2 years, 1979-1980. He has measured performance with Return on Equity (ROE) and Tobins’ q and his control variables are growth opportunities, leverage ratio, business risk and firm size. His findings are that the percentage of equity-based compensation of the CEO is positively related to firm performance. Also, the percentage of equity held by the CEO is positively related to firm performance.
Bebchuck and Grinstein (2014) found an increase in total CEO compensation of publicly trading firms (S&P500, Mid-Cap400 and Small-Cap600) during the period of 1993-2003 which increased more than the market could explain. This means that the increase in CEO compensation was higher than the increase in firm size, firm
performance and industry mix. To measure the growth in executive pay they have used Return on Assets (ROA) and past returns, the have used sales as size as a control variable.
Leonard (1990) has used a survey of 439 large U.S. corporations over a time-period 1980-1985. He investigated the effect of CEO compensation on the Return on
Equity (ROE). There is a small positive relation in executive pay (base plus bonus)
and profit. Leonard has found an elasticity of 0,112 between executive pay and sales. An increase in sales has hardly an effect on executive pay. However, this study has shown that long-term incentive plans are increasing the Return on Equity, which is in alignment with the shareholders’ wealth.
Main (1991) conducted an investigation using a sample of 241 of the largest industrial companies of Great Britain on the remuneration top executives received in 1985. He has studied the relationship between executive compensation and
shareholder wealth. The performance measures he used are normal-, abnormal, and total return to shareholders. The used control variable is size which consists of three parts: sales, assets and employees. The result is that the executive compensation is not correlated with size but is correlated with shareholder wealth.
McKnight and Tomkins (1999) have used a sample of 109 publicly trading firms of the United Kingdom with a period of 1991-1995. The have used Total
Shareholder Return (TSR), change in earnings per share (∆EPS) and change in turn
over. As control variable they have used size, measured by assets and turnover. Their research concludes that, total CEO-compensation is determined primarily by size and growth of the firm, rather than on economic performance of the firm. This is also predicted from previous studies. They also found out that there is a positive relation between performance and both short-term and long-term packages of
CEO-compensation.
Douclouliagos, Haman and Askary(2007) have conducted a study similar to the one of McKnight and Tomkins (1999) over a period of 1992-2005 within the Australian banking industry. The performance measures they have used are: Total
Shareholder Return (TSR), Return on Equity (ROE), Return on Assets (ROA), change
in earnings per share (∆EPS ) and the control variables are size and age squared. The result is that age and size have a negative effect on total compensation and TSR, ROE, ROA and ∆EPS have a positive effect on total compensation.
compensation in the banking industry and investigate whether there is a difference in CEO compensation in the banking industry and other industries in the period 1980-1990. The used performance measures are market return and Return on Assets (ROA). Their results of their examination are that there is no difference in the pay for
performance relationship between CEO’s in de banking industry and CEO’s in other industries. However they found that CEO’s in de banking industry receive less cash compensation and equity-based compensation and hold fewer stock than CEO’s in other industries. This result is mainly because of the difference in investment
opportunities but it is also because banks have different firm characteristics such debt ratio.
John and Qian (2003) conduct an investigation on the sensitivity on CEO-compensation and firm performance using a dataset of 1992-2000 of 163 commercial banks with the performance measure shareholder value(TSR) and control variables: market-value of equity(size), risk and debt ratio. Their analysis concludes that the debt ratio of the banks in the banking industry is significantly higher than in the manufacturing industry (88% debt ratio versus 33%) and that pay for performance sensitivity is lower. Which is contradicting the paper of Houston and James (1995) which has not found a difference in the pay for performance relationship between banking and other industries between 1980 and 1990. John and Qian (2003) suggest that CEO compensation should not only align with the incentives of the shareholders but also with the incentives of the debt holders.
Brick, Palmon and Wald (2006) build on the paper of Jensen and Murphy, with a dataset in the period 1992-2001 in the manufacturing industry. They have used a lot of performances measures, amongst them: Return on Assets (ROA), cash flow risk and stock return. They used the logarithm of sales and the logarithm of
employees as size control variable. Their conclusions are that additional
compensation is weakly but positive related to firm performance and that excess compensation of a CEO is associated with underperformance of the firm.
Hubbard & Palia (2005) have examined the CEO compensation in the banking industry and the effect of deregulating in the market for corporate control. They have used data of 147 banks from 1980-1989 and the performance measure: shareholders’ wealth. The shareholders’ wealth is determined by the stock return during the year time the amount of stock outstanding times the stock price at the beginning of the year. They have found higher levels of pay in competitive corporate control markets
(those in which interstate banking is permitted) than markets where interstate banking is not allowed. The have found a stronger pay-performance relation in deregulated interstate banking markets.
3. Methodology and Data
As mentioned in agency theory, to align the interests of the shareholders and the CEO, a contract will be made. In this contract measures will be added that can be monitored and improve shareholders wealth. This reduces agency costs and if CEO is able to increase shareholders wealth, he will get rewarded for this in his compensation package.
Payment Structure
CEO-compensation often consists of the following parts: base salary, cash bonus, restricted stock granted, stock options, long-term incentive plans and all other total. These elements can be retrieved from Execucomp.
Base salary
This is a predetermined part of the CEO’s salary which is paid every month to the CEO. The amount of base salary is determined by the skillset of the CEO and the size and previous growth of the firm.
Cash bonus
This is a bonus that the CEO receives at the end of the year. This is based on
performance measures during the year. It serves as a short-term incentive to the CEO to reach predetermined targets which are coupled with the performance measures.
Restricted stock granted
These are stocks of the company itself and granted to the CEO. However, because they are restricted these are not tradable until the conditions are fulfilled. When these conditions are fully satisfied the stock is not restricted anymore. By this restriction, the company creates a long-term incentive for the CEO who wants to meet this conditions which are beneficial for the shareholders.
Stock options
An option is the right to buy or sell a share at a certain price. The CEO receives call options which induces that if the share price increases he can call his options. When the share price exceeds the call price, the CEO can exercise his option which will lead to immediate profit. By this the CEO will take more risk to increase the share price, which reduces his risk-averseness and will be more aligned with the shareholders’ interests as mentioned above.
Long-term incentive plans
This is meant as a long-term incentive for the CEO. If the CEO meets certain
conditions in the future the long-term incentives are paid out. This is also an incentive to stay at the company, because if the CEO leaves before the conditions are met, we will not get any pay-out from these long-term incentives. In this study, equity compensation is the total of restricted stock granted, stock options and long-term incentive plans.
All other total
This includes all other income such as pension plans, tax gross ups, insurance premiums and personal benefits.
3.1. Methodology and Hypothesis
In this paper the four following models tested with ordinary least squares (OLS) regression:
Model 1:
ln(base salary) = α + β1TSR +β2TSR*crisis+ β
3ROA+β4ROA*crisis + β5ln(Assets)
+ β6Age+βyear + ε Model 2:
ln(bonus compensation) = α + β1TSR +β2TSR*crisis+ β
3ROA+β4ROA*crisis + β5ln(Assets)
+ β6Age+βyear + ε Model 3:
ln(equity compensation) = α + β1TSR +β2TSR*crisis+ β
3ROA+β4ROA*crisis + β5ln(Assets)
Model 4:
ln(total compensation) = α + β1TSR +β2TSR*crisis+ β
3ROA+β4ROA*crisis + β5ln(Assets)
+ β6Age+βyear + ε
A great majority of the existing papers use the natural logarithm of
CEO-compensation, therefore it used in this paper as well. The performance measures used in this paper are Total Shareholder Return (TSR), Return on Equity (ROE), Return on
Assets (ROA). In the papers of Douclouliagos et al(2007), The control variables used
are SIZE (assets) and AGE. To test whether after the crisis the performance measure has a different influence on the compensation a crisis dummy is added.
The Total Shareholder Return is used by Main (1991), McKnight and
Tomkins(1999), John and Qian (2003) and Douclouliagos et al(2007). The reason for implementing this performance measure is to see if the CEO compensation is
correlated to the shareholders best interest. It is calculated as follows: 𝑃𝑡−𝑃𝑡−1+𝐷𝑖𝑣𝑖𝑑𝑒𝑛𝑑𝑃 𝑡
𝑡
where pt is the share price in year t and pt-1 is the share price in year t-1. Also all three
of these articles are testing for banks.
Return on Equity (ROE) is a performance measure which is calculated by the
net income of the bank divided by its equity. This gives shareholders an insight in how much net income is generated by every share they have invested. This measure is based on the book value of the equity. This performance measure is also used by Mehran (1995), Leonard (1990), and Douclouliagos et al(2007).
Return on Assets (ROA) is the return on the book value of all assets. Net
income is divided by the accounting value of the assets. It shows the overall profitability of the company. Increase in return on assets is also in line with the shareholders’ best interest. Brick, Palmon and Wald (2006), Bebchuck and Grinstein (2014) , Douclouliagos et al(2007) and Houston and James (1995) have used ROA in their papers. However due to multicollinearity between Return on Assets (ROA) and
Return on Equity (ROE). Muliticollinearity mans that to dependent variables are
highly correlated with each other and explain the independent variable in the same way. The correlation of Return on Assets and Return on Equity is equal to 0.7355, see table 1. Therefore, Return on Equity is omitted from the model. Thus ROA uses the book value and TSR uses the market value.
Correlation matrix
ROE ROA TSR Assets Age
ROE 1 ROA 0.7355 1 TSR 0.0133 0.1017 1 Assets 0.0171 0.0052 -0.0092 1 Age -0.0139 0.0109 -0.003 -0.01 1
The control variable SIZE is used because if a bank is big one can expect that the CEO has to make more complex decisions and has more responsibilities towards stakeholders. Mehran (1995) McKnight and Tomkins (1999) and Douclouliagos, et al(2007) all have used the logarithm of the book value of assets as size. Bebchuck and Grinstein (2014) and Leonard (1990) use sales as size, whereas Main (1991) uses sales, the book value of assets and amount of employees as size. In this paper the logarithm of the book value of assets is used because a low amount of sales does not mean that the CEO has less important decisions to make.
The second and control variable used is Age. This is a commonly used control variable (Houston and James (1995), Leonard (1990), Douclouliagos et al(2007)) and therefore added to control the CEO.
To test whether the crisis is influencing the CEO’s compensation, a dummy is added. However, there is no existing literature on the crisis and CEO compensation. Main (1991), McKnight and Tomkins (1999), Douclouliagoset al(2007) and John and Qian (2003) have found a positive relationship between total shareholder return and CEO compensation. It is therefore expected that β1 > 0.
Brick, Palmon and Wald (2006) have not found a significant relationship between compensation and Return on Assets (ROA), Bebchuck and Grinstein (2014) have found a small but positive relation as well as Douclouliagos et al(2007) and Houston and James(1995) did. Thus the hypothesis is that β3 > 0.
McKnight and Tomkins(1999) do find a positive relation between assets and compensation. However, Mehran (1995) has not found a significant relationship and Douclouliagos et al(2007) found a negative relationship between assets and
compensation.
Leonard(1990) has found a positive relationship between age and base and bonus compensation. On the contrary, Douclouliagos et al(2007) have found a negative relationship between age and compensation.
All the existing literature has studied periods before the crisis. In this paper, however, also the period after the crisis is studied as well as the effect after the crisis of
performance measures on the compensation components. The agency theory predicts that performance measures used are depending on the firm performance. Thus, during a crisis CEO compensation should be much lower in variable compensation due to poorer bank performance. This has not been studied before in existing literature and therefore this paper is an additional value to the existing literature.
3.2. Data and descriptive statistics
The data used for this research is retrieved from Compustat, Execucomp and CRSP, see table 2. The SIC-code, which is a sector code, used is 6020. This code contains all given data on commercial U.S. banks. Out of execucomp the items selected are: base salary, cash bonus, equity compensation, total compensation and CEO-age. In compustat the items which are selected are: dividend, net income, and book value of assets are selected. In CRSP the item share price is selected. In order to obtain the
Return on Assets (ROA), net income is divided by the book value of assets. To obtain
the variable Total Shareholder Return (TSR), the share price difference over a year plus the dividends received is divided by the share price at the beginning of the year.
Table 2: variables retrieved from the databases
Variable Description Database
Dependent variable
Base salary Fixed salary Execucomp
Bonus Cash bonus Execucomp
Equity compensation Restricted stock, options, long term incentive plan Execucomp Total compensation Base salary, cash bonus, equity and all others Execucomp
Independent variable
TSR Total shareholder return consisting of:
Share price CRSP
Dividend Compustat
ROA Return on assets Compustat
Control variable
Firm size Book value of assets Compustat
CEO age Age of the CEO in the fiscal year Execucomp
The panel dataset is originally retrieved from the databases is a sample consisting of 1208 observations including 144 commercial banks with the SIC-code 6020 over a
period of 1992-2013. This code is also used by John & Qian (2003). In Execucomp, the minimum allowed year is 1992. The table in Appendix A shows the data set periods in the existing literature. All papers have used a timeframe equal to or smaller than 12 years, except for Jensen & Murphy (1990). They have used a data set over 30 years. There is no existing literature on data after 2005. Therefore it is chosen to measure a longer period than in previous studies of 21 years, until 2013, the latest year in which all information is available. Besides this also, the effect of the crisis has been taken into account, which has not been done before. As shown in table 3, the total average compensation is $4,049,325,- over the period 1992-2013. The average base salary is $758,130,-, the bonus $532,293,- and equity based compensation is $2,748,901,- .
Table 3: Summary statistics
Variable Mean Std. Dev. Min Max
Total compensation 4049325 5375107 296750 59546780
Base salary 758130 401924 1 5600
Bonus 532293 1119506 0 14500
Equity bonus 2758901 4609363 0 52046780
In figure 1, the distribution of the CEO-compensation is shown. One can see that the base salary is rigid and does not change much over time. However, in the economic contraction, the credit crunch in 2008, there is almost no cash bonus paid to the CEO’s. This is also in line with shareholders wealth which decreased during the crisis. Also the equity compensation shifts with the economic environment. This is also in line with the shareholders wealth. In appendix B all the parts of the
In appendix C, the regression estimates of year dummies and compensation are shown. The years 1994-1995 have a negative relationship with base, equity and total compensation. This might be caused by the Mexican peso crisis, which also influence the American markets. From 1997 to 2006, these years have a positive effect on cash compensation, probably because of the market boom, the internet bubble. For base compensation 2008, for bonus compensation 2008 and 2009 and for equity and total compensation 2008-2010, these years have a significant negative effect on
compensation due to the crisis.
4. Empirical result
The purpose of this research is to investigate whether CEO-compensation is related to firm performance over the period 1992-2013 in the banking sector. Firm performance is measured by Total Shareholder Return (TSR), Return on Equity(ROE), Return on
Assets (ROA). The control variables are the natural logarithm of SIZE, measured in
assets and the AGE of the CEO.
Total Shareholder Return (TSR) has a small negative effect on base
compensation (10% significance) and equity compensation (5% significance). This result is contrary to the hypothesis which expected a positive relationship and the existing literature which found a positive relationship between CEO pay and firm
0 1000000 2000000 3000000 4000000 5000000 6000000 7000000 Am o un t o f do lla rs ( $ ) Year
Graph 1:Distribution of total compensation
The division of base compensation, bonus compensation and equitycompensation per year
Equity compensation Bonus compensation Base Salary
performance. However, after the crisis TSR has a stronger effect on base
compensation (10% significance) and equity compensation (10%) and a smaller effect on cash compensation. Thus, TSR has a positive effect on the base compensation during the crisis and after the crisis even stronger. Return on Assets (ROA) has a positive effect on all components of compensation, base salary is significant at the 10% level and the other components at the 1% significance level. This is also predicted by the theory and literature. After the crisis, ROA has a greater positive relation to base compensation (1% significance) but a smaller effect on cash compensation (5% significance), equity compensation(5% significance) and total compensation (1% significance). ASSETS has a small but positive effect on all components of compensation and total compensation (1% significance). AGE has a small positive effect, but close to zero on base compensation (1% significance) equity compensation (10% significance) and total compensation (1% significance). 30% of the change in base compensation is explained by the model, for the change in cash compensation it is 41%, for the change in equity compensation it is 42% and for the change total compensation it is 60%.
Robustness checks
The compensation data is winsorized in Stata. Winsorizing is a way to deal with outliers. Instead of omitting outliers from the data, the command winsor is used to clean the data. By winsorizing data one percent of the highest and lowest data points are adjusted to the next highest and lowest data points. In table 6 the summary of the winsorized statistics are shown. Compared to table 3, the minimum and maximum values of the compensation variables are less extreme. This also reduces the standard deviation, which causes the distribution to be less tailed e.g. there are less data points in tails of the normal distribution.
Table 4: Regression estimates for pay and performance
Regress compensation TSR TSRcrisis ROAcrisis Assets Age
Independent variables Dependent variables (ln) Base compensation Cash compensation Equity
compensation Total compensation
TSR -0.11* 0.17 -0.35** -0.11 (-1.89) (1.36) (-2.37) (-1.55) TSRcrisis 0.11* -0.72*** 0.26* 0.08 (1.79) (-3.98) (1.66) (1.1) ROA 3.30* 64.34*** 31.76*** 28.93*** (1.69) (10.61) (6.24) (11.63) ROAcrisis 8.65*** -22.06** -13.44** -19.03*** (3.74) (-2.56) (-2.24) (-6.48) Assets (ln) 0.21*** 0.50*** 0.75*** 0.52*** (20.05) (16.69) (27.58) (39.32) Age 0.01*** 0.00 0.01* 0.01*** (4.2) (-0.33) (1.82) (2.7) Constant 2.49*** -2.82 -6.39* -1.57*** (12.37) (-12.21) (-6.13) Year dummies included
YES YES YES YES
Adjusted R squared 0.30 0.41 0.42 0.60
***,** and * is significance at 1%, 5% and 10% respectively T-statistic in the parentheses
Table 6: Summary statistics
Variable Mean Standard
deviation Min Max
Total compensation 3950540 4757628 381313 2686467
Base salary 754038 367264 250000 2750000
Bonus 504727 915505 0 5475000
Equity bonus 2673445 4052150 8898 23723670
In table 7 the regression estimates for pay and performance are shown. These results confirm the results in table 5, where the outliers are not winsorized. This does not hold for the performance measure Return on Assets (ROA) as measure for base compensation. Because of the outliers there was a positive effect, but the winsorized data cannot confirm this. This means that the outliers only influence the determinant
Return on Assets (ROA) as measure of the change in base compensation. All the other
regressions show that the outliers do not influence the determinants of CEO compensation in the models used in this paper.
Table 7:Regression estimates for pay and performance winsorized.
Regress compensation TSR TSRcrisis ROA crisis assets Age
Independent variables Dependent variables (ln) Base compensation Cash compensation Equity compensation Total compensation TSR -0.10*** 0.17 -0.35** -0.11 (-2.79) (1.33) (-2.43) (-1.5) TSRcrisis 0.12*** -0.71*** 0.26* 0.08 (3.12) (-3.97) (1.7) (1.03) ROA 0.29 64.70*** 29.25*** 27.81*** (0.24) (10.73) (5.85) (11.31) ROAcrisis 2.67* -22.72*** -11.21* -18.10*** (1.84) (-2.65) (-1.9) (-6.23) Assets (ln) 0.20*** 0.49*** 0.75*** 0.52*** (30.37) (16.22) (27.95) (39.47) Age 0.01*** 0.00 0.01* 0.01*** (8.22) (-0.25) (1.75) (2.71) Constant 2.59*** -2.58*** -6.34*** -1.50*** (20.49) (-4.65) (-12.3) (-5.9)
Year dummies included YES YES YES YES
adjusted R-squared 0.47 0.40 0.42 0.60
***,** and * is significance at 1%, 5% and 10% respectively T-statistic in the parentheses
5. Conclusion and discussion
To align the CEO’s interests with the best interests of the shareholders, the designed contract is organized to incentivize and monitor the CEO through performance measures. This research is conducted to investigate whether the performance measures Total Shareholder Return (TSR), Return on Assets (ROA) determine the components of CEO compensation. These parts are: base salary, cash compensation and equity compensation. The performance measures which are used in the model are:
Total Shareholder Return (TSR), Return on Assets (ROA). According to the results of
this paper, the performance measure Total Shareholder Return (TSR) should not have been used over the period 1992-2013 because this has a negative effect on base
compensation which will not incentivize the CEO in the right way. However, after the crisis, TSR has a greater effect on base compensation and therefore it should be used in monitoring and rewarding the CEO in the future. The same holds for Return on
Assets (ROA), which is not a significant determinant for base compensation over the
whole period, but does have a positive effect on base compensation after the crisis. For the change in cash compensation, Total Shareholder Return (TSR) is not a significant determinant, After the crisis, TSR has a smaller effect on the change in cash compensation. Return on Assets (ROA) does have a positive effect on the change in cash compensation. After the crisis this is a weaker positive effect. For equity compensation TSR has a negative effect on the change in equity compensation during the period 1992-2013 but a positive effect after the crisis. With the determinant ROA it is the same as with cash compensation. Over the whole period it has a positive effect and after the crisis, this effect is less strong on the change in equity
compensation. This latter part also holds for total compensation. However, TSR is not a significant determinant for total compensation. Concluding, the performance
measures TSR and ROA should be used as a measurement for variable (cash and equity) compensation to incentivize and monitor the CEO.
There are some limitations to this research, first of all these relationships between performance measures and CEO-compensation are correlated, but this does not mean they are causally related (one causes the other). The model is thus
endogenous, there is a relationship between compensation and the error term. In the dataset are 1208 observations of 144 U.S. commercial banks over a period from 1992
to 2013. However, these banks do not all exist during the whole period. It might be that the case that some banks went bankrupt and that the CEO-compensation a year before deviates from what a financially healthy bank would pay-out.
Also principals should let depend the variable compensation more on the performance measures to reward the CEO. Besides this coming research on this matter should focus on testing for causality and determining a set of performance measures which should be used in all research on this matter.
References
Bebchuck, L & Grinstein, Y (2014). The growth of executive pay, Oxford review of
economic policy, 21(2), pp. 283-303
Brick, I.E., Palmon, O. & Wald, J.K. (2006) CEO compensation, director compensation, and firm performance: Evidence of cronyism? Journal of
Corporate Finance, 12, pp. 403-423
Chen, C.R., Steiner, T.L. & Whyte A.M. (2005) Does stock option-based executive compensation induce risk-taking? An analysis of the banking industry.
Journal of Corporate Finance, 30, pp. 915-945
Douclouliagos H., Haman , J. & Askary, S. (2007) Directors' remuneration and performance in Australian banking.Corporate governance : an international review 15(6), pp. 1363-1383
Fama, E.F. & Jensen M. C. (1983) Seperation of Ownership and Control. Journal of
Law and Economics (27) pp. 1-32
Gregg, P., Machin, S. and Szymanski, S. (1993). The disappearing relationship between executives’ pay and corporate performance. British Journal of
Industrial Relations, 31 pp. 1-9
Hall, B.J. and Liebman J.B. (1998) Are CEO’s really paid like bureaucrats? Quarterly
Journal of Economics, 113(3), pp. 653-691
Hendrikse, G. (2003) Economics & Management of Organisations, McGrawhill. Houston, J.F. & James, C. (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. & Palia, D. (1995) Executive pay and performance Evidence from the U.S. banking industry. Journal of Financial Economics, 39, pp. 105-130 Hubbard, R. G. (2005) Pay without performance: A market equilibrium critique. The
Journal of Corporate Law, 30(4), pp.717-720
Jensen, M. C. & Meckling W.H.. (1976) Theory of the firm: Managerial behaviour, Agency costs and Ownership structure. Journal of Financial Economics, 3(4), pp. 305-360
Jensen, M. C. & Murphy K. J. (1990) Performance pay and top-management incentives. Journal of Political Economy, 98(2), pp. 225-264
John, K. & Qian, (2003) Y. Incentive features in CEO compensation in the banking Industry. Economic Policy Review, 9 (1), pp. 109-121
Leonard, J. S. (1990) Executive Pay and Firm Performance. Industrial and Labour
Relations Review, 43(3) ,pp. 135-295
Main, B.G.M. (1991) Top executive pay and performance. Managerial and Decision
Economics, 12(3), pp. 219-229
McKnight, P. J., Tomkins, C. (1999) Top executive pay in the United Kingdom: A corporate governance dilemma. International Journal of the Economics of
Business, 6 (2) pp. 223-243
Mehran, H (1995) Executive compensation structure, ownership, and firm performance. Journal of a Financial Economics, 38, pp. 163-184
Smith, A (1776). An Inquiry into the Nature and Causes of the Wealth of Nations, a reprinted with comments by Cannan (ed.) (1937), New York: Modern Library
S umm ar y e x ist ing li ter at ur e An ove rvie w o f the e x ist ing li ter atu re includin g : data set pe riod, se ctor, pe rfor manc e mea sur e a nd countr y Author s Da ta se t pe riod S ec tor P erf or manc e mea su re s C ountr y Douc loul iag os et al (2 00 7) 1992 -2005 B ankin g TSR , ROE, R OA Austra li a John a nd Qia n (20 03) 1992 -2000 B ankin g TSR U.S.A Houston a nd J ames ( 1995) 1980 -1990 B ankin g Mar ke t re tu rn, ROA U.S.A B ric k, P alm on a nd W ald (2006 ) 1992 -2001 Manu fa cturin g R OA , Cashf low r isk, s tock r eturn U.S.A Hubba rd a nd P ali a (2005 ) 1980 -1989 Manu fa cturin g S ha re holder w ea lt h U.S.A Mehr an (19 95 ) 1979 -1980 Manu fa cturin g R OE & Tobin 's Q U.S.A Main ( 1991) 1985 Manu fa cturin g TSR U.K . B ebc hu ck a nd Gr inst ein (2014) 1993 -2003 S & P 500 Mi d -C ap400 a n d S mall -C ap600 R OA , pa st re turn U.S.A Je nsen & Murph y ( 1990 ) 1930 -1980 Tr adin g S ha re holder w ea lt h U.S.A McK nig ht and Tomki ns (1999) 1991 -1995 Tr adin g TSR , ∆E PS , ∆T ur nove r U.K . Ha ll a nd L ie bman ( 1997 ) 1980 -1994 Tr adin g S tock pr ice s U.S.A L eon ard ( 199 0) 1980 -1985 Tr adin g R OE U.S.A Appendix A
Appendix B
Base salary (in thousands of dollars($))
Year Mean Std.dev Max Min Median
1992 733.6037 66.3634 866.039 659.772 675 1993 692.3341 26.6076 972.5 500 709.579 1994 692.3341 26.6076 972.5 500 709.579 1995 572.3234 33.2763 1505 140 562.532 1996 608.1395 31.8396 1515 150 599 1997 682.6443 33.6757 1500 200 690 1998 712.9496 34.3991 1500 376.61 700 1999 724.1219 37.8371 2000 272.016 700 2000 719.8311 41.5148 2000 265 690 2001 736.8889 48.4869 2500 249.996 700 2002 777.6421 54.0258 3000 262.75 750 2003 820.6694 56.3292 3000 300 793.75 2004 831.9286 55.8006 2461.53 249.615 794.64 2005 806.5392 39.4567 2000 259.014 824 2006 751.3986 31.456 1200 241.667 761.853 2007 696.2931 27.4725 1200 305.23 678 2008 660.8423 26.602 1200 172.404 650 2009 783.8328 78.7626 5600 238.75 670.833 2010 930.0524 83.0005 3239.85 0.001 750 2011 870.7889 59.364 3141.67 283.884 786.041 2012 860.7146 50.9817 2800 335 763.029 2013 815.5408 43.1768 2800 238.846 760.154 Total 758.1309 11.5641 5600 0.001 700
Bonus compensation (in thousands of dollars($))
Year Mean Std.dev Max Min Median
1992 516.6667 72.6483 650 400 500 1993 742.2977 130.543 2364.4 189.461 485.284 1994 742.2977 130.543 2364.4 189.461 485.284 1995 454.4762 64.0833 2187 0 387.3 1996 570.4695 88.752 3000 0 342.574 1997 812.2692 106.855 3000 0 581 1998 802.7229 119.424 3000 0 580.116 1999 831.6693 132.933 4600 0 530.255 2000 761.0993 137.128 5475 0 450.692 2001 798.4625 121.442 5000 0 525 2002 982.0439 165.495 7000 0 625.24 2003 1367.758 238.028 7600 0 713.152 2004 1332.969 229.671 7500 0 745.355 2005 1264.19 233.037 8400 0 760 2006 426.3382 224.181 13000 0 0 2007 275.8735 204.775 14500 0 0 2008 19.98142 7.84341 450 0 0 2009 54.27009 19.1802 975 0 0 2010 128.8878 70.0512 5000 0 0 2011 104.7625 64.7225 4500 0 0 2012 85.13908 26.8241 1300 0 0 2013 96.06026 26.1628 925 0 0 Total 532.2933 32.2102 14500 0 164.635
Equity compensation in thousands of dollars($))
Year Mean Std.dev Max Min Median
1992 1016.105 449.12 1913.84 541.241 593.231 1993 1358.858 313.251 4797.91 0 868.443 1994 1358.858 313.251 4797.91 0 868.443 1995 893.6844 140.737 3494.7 4.62 566.674 1996 1694.773 296.557 8486.15 29.776 795.441 1997 2246.043 405.141 15132.2 4.75 1288.57 1998 3375.045 554.182 13659.6 60.885 1951.42 1999 3611.45 764.375 32519 19.26 1699.91 2000 4400.801 991.427 30105.9 0 1538 2001 4466.691 1121.71 52046.8 5.12 1648.29 2002 3255.985 633.224 30364.4 2 1561 2003 3558.235 785.535 39200 2 1294.48 2004 3038.53 529.239 20775.7 48.698 1601.68 2005 3275.994 583.789 18138.3 13.689 1490.64 2006 4532.87 899.793 31884.2 22.865 1896.14 2007 2900.481 587.803 23723.7 8.898 935.224 2008 2057.407 497.467 34716.1 8.594 834.523 2009 1518.626 310.44 13775.3 19.98 468.428 2010 2009.759 392.092 15129.2 3.618 660.029 2011 2537.135 421.937 17143.3 13.981 957.171 2012 2558.316 407.048 17170 42.563 1225.42 2013 2763.405 374.361 16520.4 7.562 1650.1 Total 2758.901 132.62 52046.8 0 1150.73
Total compensation (in thousands of dollars($))
Year Mean Std.dev Max Min Median
1992 2266.375 581.83 3429.88 1668.23 1701.01 1993 2793.49 413.847 7862.31 923.798 2218.34 1994 2477.774 356.859 11209.3 508.218 1594.07 1995 1920.484 200.055 5525.67 542.666 1507.88 1996 2873.382 364.472 11719.7 707.410 1984.77 1997 3740.956 496.037 18332.2 562.072 2822.63 1998 4890.717 627.692 14344.6 587.385 3565.31 1999 5167.241 848.569 34611.3 665.491 3151.08 2000 5881.732 1050.92 32393 370.967 2814.64 2001 6002.042 1241.03 59546.8 360.224 3005.59 2002 5015.671 811.883 39364.4 296.750 2841 2003 5746.663 994.138 46700 571.42 2801.53 2004 5203.428 749.466 29270.7 446.437 3206.32 2005 5346.723 777.896 22338.8 464.670 3124.22 2006 5710.607 1031.96 41153.4 332.865 2527.64 2007 3872.647 683.06 28856.3 484.331 1796.65 2008 2738.230 508.074 35716.1 309.745 1463.63 2009 2356.729 373.52 18756.2 368.061 1198.65 2010 3068.699 461.398 20776.3 313.822 1611.99 2011 3512.687 488.437 23059.9 315.375 2014.84 2012 3504.169 433.154 19320 381.313 2250.36 2013 3675.006 400.692 19320.4 531.569 2439.25 Total 4049.325 154.651 59546.8 296.75 2210.28
Appendix C
Regression estimates for pay and year dummies
Independent variables Dependent variables (ln)
Year dummies Total compensation Base compensation Cash compensation Equity compensation
1992 -0.21 (-0.38) -0.03 (-0.1) 0.74 (1.01) -0.47 (-0.5) 1993 -0.1 (-0.4) -0.09 (-0.63) 0.9** (2.28) -0.56 (-1.3) 1994 -0.38** (-2.13) -0.24** (-2.23) 0.51 (1.54) -0.96*** (-3.14) 1995 -0.53*** (-2.98) -0.34*** (-3.17) 0.4 (1.19) -1.18*** (-3.91) 1996 -0.24 (-1.43) -0.28*** (-2.65) 0.5 (1.54) -0.53* (-1.82) 1997 0.08 (0.45) -0.15 (-1.36) 0.96*** (2.88) -0.14 (-0.45) 1998 0.26 (1.43) -0.10 (-0.92) 0.88*** (2.63) 0.25 (0.83) 1999 0.23 (1.37) -0.10 (-0.95) 0.83** (2.53) 0.14 (0.49) 2000 0.23 (1.34) -0.12 (-1.14) 0.93*** (2.79) 0.11 (0.36) 2001 0.26 (1.54) -0.12 (-1.13) 0.95*** (2.91) 0.11 (0.4) 2002 0.18 (1.08) -0.06 (-0.6) 1.08*** (3.31) -0.11 (-0.4) 2003 0.25 (1.49) -0.01 (-0.05) 1.29*** (3.98) -0.06 (-0.22) 2004 0.23 (1.4) 0.01 (0.08) 1.16*** (3.59) -0.03 (-0.1) 2005 0.23 (1.37) 0.01 (0.05) 1.19*** (3.62) -0.1 (-0.34) 2006 0.11 (0.67) -0.07 (-0.69) 1.06*** (2.59) 0.07 (0.24) 2007 -0.25 (-1.59) -0.14 (-1.43) 0.23 (0.57) -0.39 (-1.48) 2008 -0.50*** (-3.28) -0.20** (-2.19) -1.33*** (-3.04) -0.69*** (-2.68) 2009 -0.61*** (-3.95) -0.12 (-1.29) -0.86** (-2.01) -1.09*** (-4.15) 2010 -0.40*** (-2.59) -0.15 (-1.59) 0.09 (0.2) -0.86*** (-3.28) 2011 -0.20 (-1.27) 0.03 (0.26) -0.42 (-0.97) -0.35 (-1.31) 2012 -0.11 (-0.69) 0.04 (0.37) -0.16 (-0.39) -0.15 (-0.57) Constant 7.87*** (72.07) 6.62*** (99.14) 5.49*** (19.53) 7.21*** (38.59) ***,** and * is significance at 1%, 5% and 10% respectively
