Compensation and Bank Performance
Bachelor Thesis
University of Amsterdam Faculty of Economics and Business
Economics and Finance Supervisor: Razvan Vlahu
Steve Nugteren 10466843
Abstract
This paper examines the relationship between CEO pay and the performance of banks for American listed companies in the time-period 2006-2014. The CEO pay is separated into determinants which are salary, cash-bonus, stocks and stock options. As performance
measure, the accounting-measures Return on Equity (ROE) and Return on Assets (ROA) are chosen. The research controls for bank size, leverage and the age of the CEO.
The results of the regressions suggest that there is no relationship between total compensation and bank performance. This suggests that the agency problem cannot be solved by
compensation pay. However, mixed evidence is found that may imply a positive relationship between stock options and bank performance.
Statement of Originality
This document is written by Steve Nugteren, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its reference have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Content
1 Introduction ………... p. 5 2 Literature review ……… p. 6 2.1 Theoretical background ……….. p. 6 2.2 Relevant literature ……….. p. 9 3 Research methodology ……….. p. 10 3.1 The model ………... p. 10 3.2 The variables ……….. p. 12 3.3 Data ……… p. 15 4 Results ……....………... p. 16 5 Conclusions ……....………... p. 21 6 References ..……....………... p. 23 7 Appendix 1 .……....………... p. 251 Introduction
There are continuous debates among employers, employees, regulators and the press about the level, structure and role of CEO compensation (Tosi et al., 2000). Whereas the level of executive pay has risen sharply in the last twenty years, influenced by the increased equity-based compensation (Chen et al., 2006), the level of controversy has too. As the financial crisis of 2008-2009 unfolded, politicians and policymakers of the United States took actions to limit the compensation of firm executives (DeYoung et al., 2013). The critics against the compensation pay suggest that Chief Executive Officer (CEO) pay is not related to performance and it is therefore frequently excessive (Hubbard and Palia, 1995). The remuneration level of top executives would especially be too high in times of poor financial conditions and results. Proponents of compensation pay argue that there is a relationship between pay and performance (Hall and Liebman, 1998). As a result, higher CEO pay would imply better firm performance.
The main objective of this bachelor thesis is to examine if there is a relationship between CEO pay and the performance of banks in the United States during the time-span 2006-2014. Therefore this thesis will answer the following research question:
Does a relationship between CEO compensation pay and the performance of banks exist?
In order to give an answer to this research question, both literature and quantitative research will be conducted. The purpose of the literature research is to get a better understanding of the theories on CEO compensation and to look back at previous research on this topic. The empirical research will examine the relationship between CEO compensation pay and the performance of the American banks during the period 2006-2014.
2 Literature Review
2.1 Theoretical Background
There has been much research into compensation pay. However, the majority of these investigations focus on the performance of firms across different industries, instead of the banking sector (e.g. Jensen and Murphy, 1990; Tosi et al., 2000). The compensation structure of the banking industry differs significantly from the structure of other industries (Houston and James, 1995). Banks operate in a different business and regulatory environment than nonbanking firms. Evidence from Smith and Watts (1992) suggests that compensation is less responsive to firm performance in unregulated industries. Therefore, the banking sector cannot be compared to other industries. The United States is an interesting case to examine, as the CEOs earn relatively more than CEOs of other nations. To illustrate, CEOs in America earn a 45% higher cash compensation and 190% higher total compensation compared to the United Kingdom (Conyon and Murphy, 2000). Moreover, most recent attention for CEO pay in America is focused on the banking sector (Lehman Brothers, Merrill Lynch).
Before we look deeper into compensation pay the agency problem will be introduced. Jensen and Merckle (1976) define an agency relationship 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. In an ideal situation, the agent will always follow the objectives of the principal. This is often not the case, as agents tend to act to their own interests and diverge from the objectives of the principal. The agency problem arised since the interests of the principals and managers are not perfectly convergent (Bebchuck et al., 2003). In response to this situation a principal can limit divergences from his own interest by limiting the activities of the agent. Moreover, he could pay the agent to guarantee that he will not take certain actions that could harm the principal or to ensure he is compensated if he does take such actions. The downside of paying the agent is that this can be very costly (Bebchuck et al., 2003). The reduction in welfare as a result of this divergence is called the residual costs. Other costs are expenses to monitor the actions of the agent. It is impossible to perfectly monitor his actions, due to information asymmetry. The asymmetry is sided in favor of the agent, who has more information than the principle. Therfore, the agent has more information than the principle. The agent can use this information to his own advantage and pursue his own objectives instead. The costs of monitoring the agent and the residual loss combined are called the agency costs.
A classic example of this agent problem is the conflict of interest between shareholders and the CEO (Jensen and Murphy, 1990). In this case the CEO is the agent and the shareholders are the principles. Shareholders want CEOs to act in their interest. Jensen and Murphy argue this is whenever expected return on an action exceeds the expected costs. However, a CEO does not always act in the interest of the shareholders. Whenever a CEO follows his own interest instead of the objectives of the shareholders, he suffers from an agency problem (Bebchuck and Fried, 2004). CEOs compare their own private gain and cost from pursuing a particular activity. The wealth of shareholders is not interesting to them if it does not benefit them. What is beneficial to CEOs, is increasing the size of their firm (Tosi et al., 2000). Increasing the size of the firm may lead to more pay, power and prestige. Virtual all empirical studies find a positive relation between firm size and pay (e.g. Jensen and Murphy, 1990). Yet increasing the size may not be best for the shareholders return and it could be undesirable. Shareholders can solve the agency problem by aligning own interests to the interests of the CEO. One way to align this interest is by using compensation policy (Jensen and Murphy, 1990). This approach is called the optimal contracting approach. The agency theory supposes a positive relationship between compensation pay for executives and the performance of the firm. By connecting the compensation policy of a manager to a performance measure that is aligned with the objectives of the shareholders, the conflict of interest can be reduced. Since 1976, Jensen and Meckling have analyzed the problem of managerial power and discretion as an agency problem. According to the optimal contracting approach managers pay arrangements as a (partial) remedy to the agency problem. Therefore boards are assumed to design compensation schemes to provide managers with efficient incentives to maximize the shareholder’s value (Bebchuck and Fried, 2004). There are other researchers that argue (e.g. Bebchuck and Fried, 2003) compensation pay does not solve the agency problem. It is appropriate to look for an alternative theory, that has a different way of looking at the relationship between the compensation pay for executives and the performance of firms.
According to Bebchuck and Fried (2004), compensation policy can only solve the problem if shareholders possess complete information regarding all activities of the CEO and of all the investment opportunities the firm faces. If so, a contract could be designed specifying and enforcing the managerial action to be taken in every possible situation. This is not the case, considering managerial actions and investment opportunities are not perfectly observable. As a result, there is no contract that would perfectly align the interests of the manager and
shareholders (Bebchuck et al., 2003). Bebchuck et al. have an influential alternative approach to the optimal contracting approach, which is called the managerial power theory. According to this theory, the missing link between executive pay and the performance of firms is power imbalance between the executives and shareholders (Tosi et al., 2000). Executive compensation is not only viewed as a potential instrument to solve the agency problem, but it can be seen as part of the agency problem itself (Bebchuck and Fried., 2004). Whereas the board is assumed to design compensation schemes with efficient incentives for the managers according to the optimal contracting approach, managerial power theory assumes compensation arrangements approved by boards often deviate from this. The board designs a compensation package that does not maximize shareholder’s value (Core et al., 1998). This is due to board directors that are subject to influence by the management, are sympathetic to management or are just simply not capable of overseeing compensation (Bebchuck et al., 2002). Consequently, executives can receive excessive pay. The excessive pay received by CEOs is called rents and plays an important role in the managerial power theory (Bebchuck et al., 2002). The greater the power of the CEO, the higher the rents will tend to be. To camouflage or facilitate the extraction of these rents, managerial power can lead to the use of inefficient pay structures that even weaken or distort incentives, which reduce shareholders value even further (Bebchuck et al., 2002). So in this theory, we should not see executive compensation as a tool to align the interest between managers and stockholders. Concluding, the executive compensation is now seen as part of the agency problem.
In this scenario, the sub-optimal contract is one that minimizes the agency costs. An optimal contract could only exist if there were no agency costs. For risk-neutral CEOs the optimal contract would be a one-to-one correspondence between the value of a firm and CEO pay (Hall and Liebman, 1998). Such contract would essentially sell the firm to the CEO. This contract would have a sharing rate of one, resulting in the correct incentives on every margin for CEOs. This may be reasonable for small firms but it would not be appropriate for the large and publicly traded companies, such as US banks. There are two reasons for this. First, most CEOs will not have the capital to purchase a substantial fraction of a large company. This is the financial constraint (Hall and Liebman, 1998). Second is that CEO’s tend to be risk averse (Garen, 1994). In the optimal contracting approach a trade-off exists between incentives and risk-sharing (Fama and Jensen, 1983). If the bench-mark of one-to-one holds, there would be large swings in CEO pay. An project with a large positive NPV that would be optimal from the perspective of shareholders, could be avoided due to a risk-averse CEO. To
compensate CEOs for this high risk, they will need to be rewarded with compensational pay. Compensation pay can be used to make a CEO more risk-seeking. More risk-seeking CEOs can take on riskier projects with a positive NPV, which benefits shareholders.
2.2 Relevant literature
This research paper tests whether and to what degree CEO compensation and firm performance are related. In order to answer this question, it is important to look at relevant previous research and the main findings of these studies.
In 1990, Jensen and Murphy published the first influential paper on the relationship between firm performance and executive compensation. They found statistical evidence that suggested that CEO wealth changes $3.25 dollar for every $1000 change in shareholder wealth, also called the pay-performance sensitivity (PPS). PPS measures the effect of total compensation in $ on an increase in shareholders wealth of $1000. This is known as the Jensen-Murphy measure (JM-measure). Jensen and Murphy also found that firm size is an important determinant in measuring PPS. CEOs at larger firms tend to own less stock and have less compensation-based incentives than CEOs in smaller firms. In their sample of 2213 CEOs from 1974 to 1986 in 1295 US corporations, they found the pay-performance sensitivity for the larger top half was $1.85 dollar per $1000 compared to $8.05 per $1000 in the lower half. Their results were inconsistent with the implications of the agency models of optimal contracting. The empirical relation between the pay of executives and firm performance was positive and statistically significant, but small for an occupation were incentive pay is expected to play an important role. Concluding there is a lack of strong pay-for-performance incentives for CEO.
Hubbard and Palia (1995) examined CEO pay in the American banking industry and the effect of deregulating the market for corporate control. With a sample using panel data from 1980-1989 on 147 US banks , they find a stronger pay-performance relation in deregulated banking markets. The results are consistent with the idea that restricting pay levels of chief executive officers reduces the effectiveness of a well-functioning labor market and its associated pay structure. Finally, they find evidence that the size of the bank is positively related to the level of compensation.
Hall and Liebman (1998) have found a strong relationship between firm performance and executive compensation. This relationship is generated mostly by changes in the value of
CEO holdings of stock and stock options. Moreover, they found that both the level of CEO compensation and the PPS have risen dramatically since 1980, especially because of increases in stock options. Their dataset covered the years from 1980 to 1994 and contained 792 companies in the US. They conclude that the wealth of CEOs are strongly related to the wealth of the companies they manage.
3 Research Methodology
3.1 The model
The statistical technique that is used to perform this research is ordinary least squares (OLS) regression. OLS regression estimates the minimal sum of the squared residuals and it attempts to find the function which most closely approximates the data. OLS is used more often in empirical studies on executive compensation (e.g. Hall and Liebman, 1998). Multiple regression analysis is necessary as this study will use more than one variable. The multiple regression analysis can be formulated as follows:
𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖 = 𝛼𝛼 + 𝛽𝛽1𝐿𝐿𝐿𝐿𝑋𝑋1+ 𝛽𝛽2𝐿𝐿𝐿𝐿𝑋𝑋2+ 𝛽𝛽3𝐿𝐿𝐿𝐿 𝑋𝑋3+ 𝛽𝛽4𝐿𝐿𝐿𝐿 𝑋𝑋4 (1)
𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖 = 𝛼𝛼 + 𝛽𝛽1𝐿𝐿𝐿𝐿𝑋𝑋1+ 𝛽𝛽2𝐿𝐿𝐿𝐿𝑋𝑋2 + 𝛽𝛽3𝐿𝐿𝐿𝐿 𝑋𝑋3+ 𝛽𝛽4𝐿𝐿𝐿𝐿 𝑋𝑋4 (2)
Where X1 = Salary; X2 = Bonus; X3 = Value of Stock awarded according to the Financial
Accounting Standards and X4 = Value of options granted. The values are expressed in natural
logarithm to adjust for non-normality.
Yit is the dependent variable for period t with observation i. Xit are independent variables for
period t with observation i, α is the constant term, β1 to β7 are the coefficients for the
independent variables and εit is the error term of observation i in period t.
To examine the pay for performance relationship, this paper will not only regress the determinants base salary, bonus plans, stocks and stock options, but also regress Return on Equity (ROE) and Return on Assets (ROA) on total compensation. These regressions can be formulated as follows:
𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖 = 𝛼𝛼 + 𝛽𝛽1𝐿𝐿𝐿𝐿 𝑋𝑋8 + 𝜀𝜀𝑖𝑖𝑖𝑖 (3)
Subsequently, this study will investigate what the impact is of adding control variables Size,
Leverage and Age to the determinants of salary, cash-bonus, stocks and stock options. This
regression is formulated as follows:
𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖= 𝛼𝛼 + 𝛽𝛽1𝐿𝐿𝐿𝐿𝑋𝑋1+ 𝛽𝛽2𝐿𝐿𝐿𝐿𝑋𝑋2+ 𝛽𝛽3𝐿𝐿𝐿𝐿 𝑋𝑋3+ 𝛽𝛽4𝐿𝐿𝐿𝐿 𝑋𝑋4+ 𝛽𝛽5𝐿𝐿𝐿𝐿 𝑋𝑋5+ 𝛽𝛽6 𝑋𝑋6+ 𝛽𝛽7 𝑋𝑋7+ 𝜀𝜀𝑖𝑖𝑖𝑖 (5)
𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖 = 𝛼𝛼 + 𝛽𝛽1𝐿𝐿𝐿𝐿𝑋𝑋1+ 𝛽𝛽2𝐿𝐿𝐿𝐿𝑋𝑋2+ 𝛽𝛽3𝐿𝐿𝐿𝐿 𝑋𝑋3+ 𝛽𝛽4𝐿𝐿𝐿𝐿 𝑋𝑋4+ 𝛽𝛽5𝐿𝐿𝐿𝐿 𝑋𝑋5+ 𝛽𝛽6 𝑋𝑋6+ 𝛽𝛽7 𝑋𝑋7+ 𝜀𝜀𝑖𝑖𝑖𝑖 (6)
Where X5 = Size; X6 = Leverage, X7 = Age of CEO and X8 = Total Compensation.
In addition, ROE and ROA will be regressed on total compensation with the control variables included, which are formulated as follows:
𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖 = 𝛼𝛼 + 𝛽𝛽1𝐿𝐿𝐿𝐿 𝑋𝑋8 + 𝛽𝛽5𝐿𝐿𝐿𝐿 𝑋𝑋5+ 𝛽𝛽6 𝑋𝑋6+ 𝛽𝛽7 𝑋𝑋7 𝜀𝜀𝑖𝑖𝑖𝑖 (7)
𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖 = 𝛼𝛼 + 𝛽𝛽1𝐿𝐿𝐿𝐿 𝑋𝑋8 + 𝛽𝛽5𝐿𝐿𝐿𝐿 𝑋𝑋5+ 𝛽𝛽6 𝑋𝑋6+ 𝛽𝛽7 𝑋𝑋7 𝜀𝜀𝑖𝑖𝑖𝑖 (8)
Finally, a robustness check will be performed. This is done by conducting a lag-test, in order to see if the results using this instrument are still consistent. The previous regression equations assume relationships between the pay of the CEO and the performance of banks, where CEO pay in year t is related to bank performance in the same year. It is possible that CEO compensation is determined by the performance of the bank in year t-1. This can be formulated as follows:
𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖 = 𝛼𝛼 + 𝛽𝛽1𝐿𝐿𝐿𝐿𝑋𝑋1(𝑖𝑖−1)+ 𝛽𝛽2𝐿𝐿𝐿𝐿𝑋𝑋2(𝑖𝑖−1)+ 𝛽𝛽3𝐿𝐿𝐿𝐿 𝑋𝑋3(𝑖𝑖−1)+ 𝛽𝛽4𝐿𝐿𝐿𝐿 𝑋𝑋4(𝑖𝑖−1)+ 𝛽𝛽5𝐿𝐿𝐿𝐿 𝑋𝑋5(𝑖𝑖−1)+
𝛽𝛽6 𝑋𝑋6(𝑖𝑖−1)+ 𝛽𝛽7 𝑋𝑋7(𝑖𝑖−1)+ 𝜀𝜀𝑖𝑖(𝑖𝑖−1) (9)
𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖 = 𝛼𝛼 + 𝛽𝛽1𝐿𝐿𝐿𝐿𝑋𝑋1(𝑖𝑖−1)+ 𝛽𝛽2𝐿𝐿𝐿𝐿𝑋𝑋2(𝑖𝑖−1)+ 𝛽𝛽3𝐿𝐿𝐿𝐿 𝑋𝑋3(𝑖𝑖−1)+ 𝛽𝛽4𝐿𝐿𝐿𝐿 𝑋𝑋4(𝑖𝑖−1)+ 𝛽𝛽5𝐿𝐿𝐿𝐿 𝑋𝑋5(𝑖𝑖−1)+
𝛽𝛽6 𝑋𝑋6(𝑖𝑖−1)+ 𝛽𝛽7 𝑋𝑋7(𝑖𝑖−1)+ 𝜀𝜀𝑖𝑖(𝑖𝑖−1) (10)
All variables from model 5 & 6 apply here, they are only lagged by one year.
Multiple banks are observed for more than one time period. So this empirical model uses panel data, also called cross-sectional time series data. A panel dataset should have data on N cases, covering T time periods, for a total of N x T observations. This panel set is unbalanced, meaning that the number of time series differ. Previous research on CEO compensation determination for econometric models is used more often (e.g. Hubbard and Palia, 1995; Hall and Liebman, 1998).
3.2 The variables
This section will look deeper into the variables. First, the independent variable Pay will be analyzed, second the dependent variable Performance.
There are many mechanism by which compensation policy provides value increasing incentives to improve the performances of CEOs (Hubbard and Palia, 1995). The CEO compensation can be classified between base salary, bonus, stocks, stock options, pensions and other compensation. Stocks and stock options are equity-based compensation. Previous research has shown that the pay-performance relationship can differ between the components of CEO pay (Hubbard and Palia. 1995). In this paper we will only look at Salary (1), Bonus Plans (2), Stocks (3) and Stock Options (4).
(1) Salary
This is the fixed base salary paid to a CEO during the year in dollars. This salary can change each year because of inflation or based on previous performance. Base salary is usually a small portion of the total compensation.
(2) Cash-Bonus
Bonus, or the contract that provides incremental (typically) cash compensation to the base salary, is one of the determinants of compensation pay. It is the yearly variable cash payment. Together with salary, it is called cash compensation. The cash-bonus is often based on accounting measures and CEOs receives bonuses after reaching specific targets designed by the board (Holthausen et al., 1995). On average, annual bonus is approximately 20% of total CEO compensation. Therefore, cash-bonus will be used for this research.
(3) Stocks
Stocks are, together with stock options, called equity-based compensation. There are multiple empirical researches that found evidence that suggest a positive relationship between equity-based compensation and firm performance (e.g. Jensen and Murphy, 1990; Mehran, 1995). When a CEO owns more stocks, his interest is more linked to the interest of the shareholders (Jensen and Meckling, 1976). The more stocks a CEO owns, the more his compensation is related to the stockholder return of the company. These stocks are valued at the time of issuing.
(4) Stock Options
Stock options give the right to purchase stock of the company at a fixed price at a certain time period. A CEO can make profit on these options by increasing the exercise price of the option to a higher level than the pre-set price. The difference between stocks and stock options is that options create asymmetric incentives. The option will become either worthless or it can increase without limits. This way, CEOs cannot lose more than the options, but it can gain without limits. The enormous growth in top U.S. executives’ compensation has resulted largerly from stock option awards (Yermack, 1995). In comparison to industrial firms, the use of stock option-based compensation has become more widespread in the banking industry (Chen et al., 2006). Moreover, the percentage of stock option-based compensation relative to the total compensation has increased. Stock options induces risk-taking in the banking industry (Chen et al., 2006), therefore CEOs can become less risk-averse by using stock options. Stock options that are received are valued based on the rules of the Financial Accounting Standards Board.
In this research paper it is assumed that if CEOs receive more pay on performance related determinants, such as bonuses, options and stock options, banks should perform better. Overall, a positive relationship between CEO pay and bank performance is hypothesized. Salary is measured as the actual paid fixed salary in one year. This is a fixed compensation determinant which should not influence a CEO to perform better. Based on these assumptions, the following hypotheses can be formulated:
(H1) Total compensation is positively related to bank performance
(H2) Salary is not related to bank performance
(H3) Cash-bonus is positively related to bank performance
(H4) Stocks are positively related to bank performance
(H5) Stock options are positively related to bank performance
The performance measure should be strongly linked to CEO compensation for practical relevance. In this empirical research, accounting-based measures Return on Equity (ROE) and Return on Assets (ROA) are chosen. There are more empirical studies that have chosen these measures (e.g. Hubbard and Palia, 1995; Tosi et al., 2000). Tosi et al. (2000) find in their analysis that ROE and ROA have the largest explanatory power in explaining CEO pay
of the accounting-based measures (2000). Return on Equity (ROE) measures how the company generated profit with the money shareholders invested. It is the amount of net income returned as a percentage of the shareholders equity. Return on Assets (ROA) measures how efficient management is using its total assets to generate profits.
Factors that influence the performance banks should be controlled for. This current paper controls for three variables that could influence the relationship between compensation pay and bank performance. These are Size , Leverage and the Age of the CEO.
The first control variable is company size. Although there are different ways in measuring the size of a company, this paper measures firm size by the book value of total assets. Measuring the firm size by total assets has been done in multiple empirical research on executive pay (e.g. Mehran, 1994). Other empirical research uses firm sales (e.g. Core et al., 1998) or market value of equity. Virtual all empirical studies have found a positive relation between firm size and pay (e.g. Jensen and Murphy, 1990). According to Tosi et al. (2000), CEOs are more interested in increasing firm size than maximizing profits, because doing so may lead to more pay, power and prestige for the CEO. This could even result in negative returns to the shareholders. Moreover, CEOs can exert more influence over firm size than performance and therefore CEOs would prefer to use firm size as the criterion for compensation purposes. Greater size would legitimize higher CEO pay by justifying for a size premium. An increase in size would also result in greater organizational complexity, therefore a CEO requires more human capital to run the business. Furthermore, bigger firms have more hierarchical layers and therefore more pay at the top. Finally, as CEOs are risk-averse, they can reduce risk by linking compensation to company size.
Leverage is the second control variable and it is measured by dividing total debt by total assets. Previous empirical research has found a negative relationship between firm debt and performance (e.g. Agrawal and Knoeber, 1996). According to Ortiz-Molina (2007), debt can mitigate the agency problems between shareholders and the CEO by inducing lenders to monitor, reducing the free cash flow available to managers. The threat of bankruptcy forces managers to focus on value maximization. Thus, higher debt and high-powered incentives lead to a lower pay-performance sensitivity in more levered firms.
Finally, the third control variable is age. This research assumes a positive relationship between the age of a CEO and the bank performance. A CEO can gain knowledge and
therefore human capital as he becomes older. As a result, it could mean that more knowledge would imply better performance.
3.3 Data
This paragraph describes the main features of the dataset. All data have been analyzed with the software program Stata. Simple OLS regressions were performed with panel data. This will be tested for a sample of 28 publicly listed American banks during the time period between 2006 and 2014 (9 years). Two banks were eliminated from the regression beforehand, as they missed out on information for stock options and ROE. The banks in this sample are all listed and contain a random sample of large and smaller banks in the United States.
All data for salary, cash-bonus, stocks, stock-options, total compensation and age of the CEO were retrieved from ‘Execucomp’, found on WRDS. The data for performance measures ROE and ROA are collected from Datastream, as well as the control variables size and leverage. Size was measured in total assets, leverage in total debt divided by total assets. For a robustness check, a lag-test was performed. The independent variables used for this test were lagged by one year in Stata. To look at which American listed banks were included in this sample, see table 5 in the appendix.
The highest-earning CEO in this sample is the CEO of Goldman Sachs, with a total compensation of $53,965,418 in 2007. The average total compensation in this sample was $6,115,980. CEOs earned on average $2,307,118 of their compensation in ownership of stocks and $1,158,684 in stock options. Salary and bonus had an average of $983,508 and $656.930 respectively. This leaves out $1,049,470 of total compensation, which is earned with other compensation. Ownership of stocks was the biggest determinant of total compensation pay with a share of 37,48 %. The performance measures ROE and ROA had an average of 4,61 % and an average of 0.84 % respectively. During the crisis, it is clear banks had the lowest (or negative) ROE and ROA rates. ROE had an average of -5.3% and -6.73% in 2008 and 2009, ROA had rates of -0.07% and 0.17% in 2008 and 2009. The highest rates were found for both performance measures in 2006, with 14.94% and 1.72% for ROE and ROA respectively. The average age of CEO’s was 57.2, with a median of 58 years.
Table 1: Summary of statistics on the data used for regressions in all the models. The number
Min) and maximum (Max) are all included in this table for salary, bonus, stocks, stock options, total compensation, size, leverage, age, ROE and ROA.
4 Results
This chapter will present the results of the aforementioned regressions. There are 10 models that will be discussed, for each performance measure 5. The objective of this research is to examine the relationship between the CEO pay and the performance of US banks. Table 2 shows the results of the first four models, which do not include the control variables. Table 3 presents the results of the second 4 models, where control variables are now included. The last two models show the results of the lag-test which is conducted as a robustness check.
Table 2: This table gives an overview of the coefficients obtained from the regression of the
models 1, 2, 3 and 4. Model 1 & 2 include the determinants salary, bonus plans and stock and stock options. The elements are expressed in natural logarithm. The performance measures used for the regression are Return on Equity and Return on Assets. In model 3 & 4, Return on Equity and Return on Assets is regressed separately on total compensation, which is also expressed in natural logarithm. The number above gives the coefficient and the number below gives the corresponding standard error term. Explanatory power (R2) and the number of observations are reported in the lower part of each model. *, **, *** denotes significance at the 1%, 5% and 10% levels respectively.
Variable Obs Mean Median Std. Dev Min Max
Salary 233 983,508 917,352 482,653 0 3,144,823 Bonus 233 656,930 0 2,849,695 0 26,985,470 Stocks 233 2,307,118 1,025,004 3,680,930 0 28,830,000 Stock Options 230 1,158,684 0 3,781,444 0 42,110,830 Total Comp 231 6,155,980 3,615,986 7,548,555 553,244 53,965,420 Size 233 2.78 * 108 2.12 * 107 6.18 * 108 2,030,775 2.42 * 109 Leverage 233 18.761 15.2 18.761 2.224 55.874 Age 231 57.212 58 7.362 34 73 ROE 233 4.612 8.27 16.292 -128.07 34.59 ROA 220 0.841 0.99 1.41 -14.8 3.99
In model 1 base salary, cash-bonus and stocks are not significant for the performance measure ROE. This would mean that there is no relationship between pay in salary, bonus and stocks and bank performance. Stock options are statistically significant, with a positive relationship at a level of 5 % (p=0.046). This would suggest that banks that pay more (less) in stock options, perform better (worse). The determinants at model 1 are also tested for ROA, in model 2. Again, base salary, cash-bonus and stocks are not found significant. Stock options are significant at a level of 5% (p=0.043). Both performance measures find stock options positively related with bank performance. Model 3 only looks at the influence of total compensation on ROE. There is no statistical significance of a relationship. In model 4 this is done for ROA. Again, there was no significance found, which would suggest that for both performance measures ROE and ROA there is no relationship between total compensation and bank performance.
In order to check if other factors could have influenced the impact on performance, control variables will be included in the next regressions. These control variables are size, leverage and age of the CEO.
Variable Model 1 Model 2 Model 3 Model 4
Intercept (α) 9.192 (7.606) -0.692 (1.436) -0.932 (6.189) 0.599 (0.066) (1) ln(Salary) -1.144 (1.132) -0.065 (0.104) (2) ln(Bonus) 0.102 (0.442) -0.019 (0.039) (3) ln(Stocks) 0.227 (0.367) -0.022 (0.033) (4) ln(StockOptions) 0.615 (0.308)** 0.056 (0.028)** (5) ln(Total Compensation) 0.728 (0.734) 0.034 (0.066) Observations 230 217 231 218 R2 0.0339 0.0231 0.0065 0.0018
Table 3: In this table, compensation is measured by base salary, cash-bonus, stocks and stock
options. Performance is measured in ROE (5 and 7) and ROA (6 and 8). Size, leverage and the age of the CEO are included as control variables. Size is measured by the natural logarithm of total assets. *, **, *** denotes significance at the 1%, 5%, 10% levels respectively.
Model 5 shows that the determinants salary, cash-bonus and stocks are not significant. This would imply that there is no evidence for a relationship between these determinants and bank performance. The determinant stock options is once more significant, with a p-value of 0.036. There is a positive relationship, which means that banks that pay more (less) stock options perform better (worse). Size is not significant, which suggests that a bigger firm does not necessarily lead to a better performance. In addition, the results do not show evidence that an
Variable Model 5 Model 6 Model 7 Model 8
Intercept (α) -10.629 (19.132) -0.692 (1.436) -20.199 (18.407) -0.844 (1.371) (1) ln(Salary) -1.563 (1.149) -0.078 (0.106) (2) ln(Bonus) 0.226 (0.452) -0.021 (0.039) (3) ln(Stocks) 0.142 (0.379) -0.028 0.035 (4) ln(StockOptions) 0.649 (0.309)** 0.063 (0.028)** (5) ln(Size) 0.492 (0.943) 0.036 (0.070) 0.683 (0.939) 0.044 (0.069) (6) Leverage -0.119 (0.121) 0.007 (0.009) -0.0115 (0.118) 0.005 (0.009) (7) Age of CEO 0.287 (0.176) 0.023 (0.014)* 0.191 (0.176) 0.014 (0.014) (8) ln( Total Compensation) 0.526 (0.756) 0.005 (0.068) Observations 229 216 229 216 R2 0.065 0.0469 0.0305 0.0156
increase in leverage leads to worse performance. Finally, older CEOs do not result in better performance. Model 6 finds similar results, with no significant results for salary, cash-bonus and stocks, but with a significance for stock options (p=0.026). Model 1, 2, 5 and 6 have found statistical significance for stock options, which would suggest that firms that pay more stock options perform better. Size is not significant for the performance and neither is leverage. There has been found small significance for the age of the CEO (p=0.097). It is a positive relationship and it suggests that older (younger) CEOs perform better (worse). In model 7 and 8 the control variables size, leverage and age have been added to total compensation. Model 7 & 8 do not find any statistical significance for total compensation, which is corresponding to the output shown in table 2. Model 7 finds no evidence for a relationship between size, leverage or age and bank performance. Moreover, model 8 does not find any significance at all.
In order to estimate the sensitivity of the results, this paper performs a robustness check (lag-test).
Table 4: In this table, compensation is measured by base salary, cash-bonus, stocks and stock
options. These determinant are all lagged by 1 year. Performance is measured in ROE (9) and ROA (10). Size, leverage and the age of the CEO are included as control variables, which are lagged by 1 year. *, **, *** denotes significance at the 1%, 5%, 10% levels respectively.
Variable Model 9 Model 10
Intercept (α) -10.629 (19.132)* -1.557 (1.544) (1) ln(Salary) -0.715 (1.124) -0.004 (0.125) (2) ln(Bonus) 0.433 (0.479) -0.039 (0.045) (3) ln(Stocks) -0.868 (0.384)** -0.071 0.037* (4) ln(StockOptions) 0.144 (0.323) 0.028 (0.032) (5) ln(Size) 2.086 (1.027)** 0.081 (0.078) (6) Leverage -0.478 (0.131)*** -0.008 (0.010)
In model 9, salary, cash-bonus and stock options show no statistical significance. Stock options showed statistical significance in the previous regression. This is different now the determinants are lagged by 1 year. Ownership of stock shows a negative relationship with bank performance (p=0.024). This would mean that banks who pay more (less) in stock, perform less (better). The size is positively related to performance, with a significance of 5% (p=0.042), meaning that size has a positive influence on the performance. Leverage was found statistically significant at a 1% level (p<0.000), which implies that leverage is negatively related to firm performance. This appears to have some consensus with previous research performance (e.g. Agrawal and Knoeber, 1996). There has been found small statistical significance (10% level) for a positive relationship between age and firm performance. This would imply that an older CEO would positively influence the performance of banks.
This research paper formulates 5 hypotheses and it tries to confirm or reject the hypotheses in order to draw cautious conclusions from it. There was no model which found evidence that suggested that total compensation had a relationship on the performance of American banks, so there is no support for hypothesis 1. These results are inconsistent with the agency models of optimal contracting. This is in line with previous research of Jensen and Murphy (1990). They found a small significance, but it was not enough to be consistent with the agency models of optimal contracting. However, it differs from the results of Hubbard and Palia (1995), who found a strong relationship between executive pay and firm performance. The second hypothesis assumes there is no relationship between salary and the performance of banks. There was no statistical significant relationship found in any of the models. Therefore, there is no reason to reject hypothesis 2. Cash-bonus plans were assumed to be positively related to bank performance, but no statistical data from the models supported this hypothesis. The determinant stock ownership found a negative relationship with bank performance in model 9, with a -0.868 coefficient and with a 5% level significance. This was found significant while performing the lag-test. No evidence suggest that stock ownership is (7) Age of CEO 0.343 (0.186)* 0.024 (0.015)
Observations 201 192
positively related to bank performance, which is in contrast to hypothesis 4. Other empirical research found a positive relationship between ownership of stock and firm performance (e.g. Jensen and Murphy, 1990). Stock options were found significant in all models with a positive relationship. The exception is made for the robustness check (lag-test), which found no statistical significance for any relationship. The mixed evidence suggests that banks that pay more stock options, perform better. This supports hypothesis 5, which is in line with other previous research (Hall and Liebman, 1998).
Conclusions
The objective of this research is to examine the relationship between CEO pay and the performance of banks in the United States during 2006-2014.
Data on executive compensation has been collected and analyzed from a sample of 28 American listed banks during the period 2006-2014. Total compensation has been separated into salary, cash-bonus, stocks and stock options. The variables size, leverage and age are included in the model in order to control for the potential influence on CEO pay. The results suggest that total compensation does not have a relationship with bank performance, which does not support the optimal contracting approach. There appears to be some consensus in the existing literature as Jensen and Murphy (1990) did not find evidence supporting the optimal contracting approach. They found a statistical significance that pointed to a small relationship between compensation pay and firm performance, but it was not large enough to support the implications of the agency models of optimal contracting.
There are statistically significant results that show a positive relationship between stock options and the performance of firms. Also, there has been found slight evidence that size has a positive relationship on the performance of banks. Finally, some evidence suggests leverage has a negative effect on the performance of banks.
There are fundamental limitations in this research which may distort the results. First of all, this study only looked at the compensation pay of CEOs, instead of top executives. Banks are controlled by a larger group of executives. Another limitation is the relative small size of the sample and the limited time period. In other empirical research on this subject, samples include more companies larger time periods (e.g. Hall and Liebman, 1998; Hubbard and Palia, 1998). As a result, the quality of the research is lower. Furthermore, previous research show there are more control variables besides size, leverage and age of the CEO, that could
influence the performance of firms. These have not been taken into account and could influence the results. Furthermore, the financial crisis had a large impact on banks in 2008-2009. As the time-period covered 2006-2014, the results can become chaotic and inconsistent. In addition, banks performed worse during these years and higher compensation pay cannot solve bad performance during the crisis. Follow-up studies should contain a larger sample, with a longer time period and more banks, in order to get more significant results and more consistency. Also, they could add control variables such as tenure and the size of the board. Finally, follow-up studies could test on more executives, such as CFO’s and COO’s.
References
Agrawal, A., & Knoeber, C. R. (1996). Firm Performance and Mechanisms to Control Agency Problems between Managers and Shareholders. Journal of Financial and
Quantitative Analysis, 31(3), 377-397.
Baltagi, B. (2008). Econometric Analysis of Panel Data (Vol. 1). John Wiley & Sons
Bebchuck, A. L., Fried, J. M. & Walker, D., I. (2002). Managerial Power and Rent Extraction in the Design of Executive Compensation. University of Chicago Law Review. 69:3, 751-846.
Bebchuck, A. L. & Fried, J. M. (2003) Executive Compensation as an Agency Problem.
Journal of Economic Perspectives, 17(3), 71-92.
Chen, C. R., Steiner, T. L., & Whyte, A. M. (2006). Does stock option-based executive compensation induce risk-taking? An analysis of the banking industry. Journal of
Banking & Finance, 30(3), 915-945.
Conyon, M. J., & Murphy, K. J. (2000). The prince and the pauper? CEO pay in the United States and United Kingdom. Economic Journal, F640-F671.
Core, J. E., Holthausen, R. W., & Larcker, D. F. (1999). Corporate governance, chief executive officer compensation, and firm performance. Journal of Financial
Economics, 51(3), 371-406.
DeYoung, R., Peng, E. Y. & Yan, M. (2013). Executive compensation and business policy choices at US commercial banks. Journal of Financial and Quantitative Analysis, 48(1), 165-196.
Jensen, M. C. & Ruback, R. S. (1983). The market for corporate control: The scientific evidence. Journal of Financial Economics, 11(1), 5-50.
Garen, J. E. (1994). Executive compensation and principal-agent theory. Journal of Political
Economy, 102(2), 1175-1199.
Hall, B. J. & Liebman, J., B. (1998). Are CEOs Really Paid Like Bureaucrats? The Quarterly
Journal of Economics, 118(3), 653-691.
Holthausen, R.W., Larcker, D.F., & Sloan R. G. (1995). Annual bonus schemes and the manipulation of earnings. Journal of Accounting and Economics, 19(1), 29-74. 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 (2), 405-431.
Hubbard, R.G., & D. Palia. (1995). Executive Pay and Performance: Evidence from the U.S. Banking Industry. Journal of Financial Economics, 39 (1), 105-130.
Jensen, M., C., & Murphy, K., J. (1990). Performance Pay and Top Management Incentives.
Journal of Political Economy. 98 (2), 225-263.
Jensen, M., C., & Meckling, W. (1976). Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure. Journal of Financial Economics, 3 (5), 305-360.
Mehran, H. (1995). Executive compensation structure, ownership, and firm performance. Journal of Financial Economics, 38 (2), 163-184.
Ortiz-Molina, H. (2007) Executive compensation and capital structure: The Effects of Convertible Debt and Straight Debt on CEO Pay. Journal of Accounting and Economics,
43 (1), 69-93.
Smith, C. W., & Watts, R. L. (1992) The Investment Opportunity Set and Corporate Financing, Dividend, and Compensation Policies. Journal of Financial Economics, 32 (3), 263-292.
Tosi, H. L., Werner, S., Katz, J. P., & Gomez-Mejia, L. R. (2000). How much does performance matter? A meta-analysis of CEO pay studies. Journal of Management, 26 (2), 301-339.
Yermack, D. (1995). Do Corporations Award CEO Stock Options Effectively?. Journal of
Appendix
Table 5: List of banks
Bank Name JP Morgan Chase & Co
City National Corp Comerica Inc Commerce Bancshares Inc
Citigroup Inc Cullen/Frost Bankers Inc
Fifth Third Bancorps First Horizon National Corp
Bank of America Corp Keycorp State Street Corp Suntrust Banks Inc UMB Financial Corp Valley National Bancorp
Synovus Financial Corp TCF Financial Corp Webster Financial Corp
N B T Bancorp Inc* Sterling Financial Corp*
CVB Financial Corp Astoria Financial Corp Wintrust Financial Corp
Flagstar Bancorp Inc Umpqua Holdings Corp
BBCN Bancorp Inc Goldman Sachs Group Inc
Privatebancorp Inc Hudson City Bancorp Inc
Pacwest Bancorp Texas Capital Bancshares Inc
