The Relation between Firm Performance and CEO compensation
in the Financial Sector; Pre, during and post the credit crisis of
2007-2008
Elise Rosenkotter
Student number: 10267018 University of Amsterdam Bsc: Economics and Business
Specialization: Finance and Organization Supervisor: Mr. A. R. S. Woerner
Statement of Originality
This document is written by Student Elise Rosenkotter 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 references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of the work, not for the contents.
Abstract
This study aims to examine the relationship between firm performance and CEO compensation in the financial sector. The recent credit crisis of 2007-2008 has resulted in stronger supervision of the financial sector. Therefore, this study analyzes whether the pay-performance relationship has changed, looking at three different periods; pre-, during and post-crisis. It involves data regarding the financial sector of the S&P 1500. To identify if different measures of performance have different relations with firm performance, both total compensation and cash compensation are used as measures of compensation. Two different performance measures; return on equity and return on assets are used to make sure the results are reliable. The results show different patterns for both performance measures and for the different structures of compensation. The periods pre- and during-crisis using total compensation conclude for both periods and performance measures no significant relationship. The post-crisis period has a positive pay-performance relation, which is only significant when using ROA. The cash compensation regressions show also mixed results. The pay-performance relationship is significantly positive when using ROE as the performance measure and non-existent when using ROA. The periods during- and post-crisis generate both no significant results. The conclusion is that the relation between firm performance and total compensation has become stronger post-crisis. Furthermore, the pay-performance relation with cash compensation has become negative however this is not significant.
Table of Content 1. Introduction 2. Literature Review
2.1 Agency Theory
2.2 Executive compensation structure
2.3 The relationship between firm performance and executive compensation 2.4 Crisis
3. Hypothesis
4. Research methodology
4.1 Sample and data collection 4.2 Variables and regression equation
4.2.1 Dependent variables 4.2.2 Independent variables 4.2.3 Control variables 4.3 Regression equation 5. Results 6. Discussion 6.1 Regression analysis
6.2 Limitations and future research 7. Conclusion
References Appendixes
1. Introduction
There exists no consensus in literature of whether the relationship between firm performance and CEO compensation is positive, negative or does not exist at all (Core et al., 1999; Brick et al., 2006; Jensen & Murphy, 1990; Hall & Liebman, 1998; Attaway, 2000). Over the last years, research has focused on examining this relationship in different industry sectors and different countries. However, nothing has been written yet about the financial sector. Compensation in the financial sector is on average higher than compensation in other sectors (Turner, 2009). These high remuneration levels have been a topic of concern in the last years. The compensation is said to be too high and not based on the firm performance (Vemala et al., 2014).
This lack of alignment is named as one of the causes of the credit crisis. The financial crisis, that started in the United States of America in 2007, spread quickly through the entire world (Eichengreen et al., 2012). The problems in the financial sector were the main cause of the crisis. The rising housing prices gave bankers the idea to provide sub-mortgages. This would later be seen as the beginning of the biggest financial crisis since the Great Depression in 1930 (Melvin & Taylor, 2009). Post-crisis the Dodd-Frank act was introduced which proposed new regulations for the financial sector, to make sure that a credit crisis this big would not happen again (Murdock, 2011).
This study contributes to the literature by examining the influence of the credit crisis on the pay-performance relation. Accordingly, the expectation of this study was that firm performance would be weak positive related to CEO compensation. While previous research was primarily focused on different countries or industry sectors, this study contributes by being the first to examine this relation in the financial sector, in particular. It aims to acquire insight in the following question: ‘Has there been a change in the relation between firm performance and the CEO remuneration pre, during and post the crisis in the financial sector of the USA in the past 14 years?’
This study examines this relationship using data of the financial sector of the Standards and Poor’s 1500. The relation of firm performance with total compensation and cash compensation is analyzed using multiple regression equations. Two performance measures will be used; return on equity and return on assets, to increase the reliability of the results.
The expectation of this study was that firm performance would be weakly-, positively related to CEO compensation. Additionally, it was expected that the credit crisis would have a positive effect on the pay-performance relation and that the relation will be stronger
post-crisis. This is because the stronger post-crisis regulation is expected to have a positive impact on the relation of remuneration and firm performance.
The main results obtained from this study are that there is a stronger positive relationship between firm performance and total compensation post-crisis. Although this result is only significant for one of the two performance measures used. The cash compensation shows a negative pay-performance relation post-crisis, but these are also not significant. The regression results and correlation matrixes indicate that there is a stronger relationship between firm performance and total compensation than there is with cash compensation.
In the next section a review and discussion of the literature regarding the relationships between agency theory, executive compensation structure, crisis and pay-performance relationship is provided. In the third section, the hypotheses are stated. In the fourth section, the methodology of the research is reported. In the fifth section, the results are reported. Finally, in the sixth and seventh section, the results are discussed and the implications and limitations are described.
2. Literature review
In this section the literature is discussed. Furthermore, the different aspects of the relationship between firm performance and CEO pay will be presented. The empirical research explores the relationship between the alignment of the incentives of the CEO and the shareholders. The chapter concludes with a section about the link to this thesis.
2.1 Agency Theory
Managerial tasks are not always executed by the company’s shareholders. A management team is hired to make the day-to-day decisions of how the firm will act in the continuous changing market environment. To make sure these decisions will be made in the interest of the shareholders, they have to align the incentives of the management with their own incentives. To align these incentives and design the perfect contract has been proven to be a difficult job (Hall and Liebman, 1998; Joyce, 2001).
The problem of aligning the incentives of the management and the shareholders has become known as the agency theory (Ross, 1973; Fama & Jensen, 1983; Attaway, 2000). The study of Jensen and Meckling (1976) discussed this principal-agent problem, which arises when there is a separation of ownership and control. The agent has to make decisions on behalf of the principal. Moreover, the owners of the firm do not have complete information
on what the CEO is doing and whether he is making the right business decisions. According to Jensen and Murphy (1990b) CEOs will first pursue their private objectives if the costs are not higher than the rewards. It is difficult to design a contract for the CEO, which makes sure that the CEO will work to increase shareholders wealth instead of pursuing their private objectives (Jensen & Murphy, 1990b).
The agency theory is the underlying theory of the relationship between CEO compensation and firm performance. Since this study examines the relationship between firm performance and CEO compensation in the financial sector, the agency theory is used to understand the problem that arises with the separation of ownership and control.
2.2 Executive compensation structure
Many of the studies concerned with agency theory focus on ‘how to reward CEO’s’. Is it most profitable to give them long-term incentives, short-term incentives, cash bonuses or stock options? In order to align the CEO and shareholders incentive, CEO’s receive a part of the compensation in stock. When CEOs receive their compensation partly in stock this is to align the incentives of the shareholders with the CEOs and to give them a long-term incentive to act in the interest of the shareholders. However the stock price sensitivity is different across the amount of share-ownership. Incentives of CEOs are increased when the stock-ownership before was low, but if it was already high before, the increase will not have a significant effect. This is because the CEOs will start selling the stock to spread for instance their risks (Ofek & Yirmack, 2000; Tzioumis, 2008).
Cash bonuses are paid-out to employees to give them a short-term incentive. However the threat of executives manipulating earnings to make sure they meet the requirements to obtain the bonus seems present (Healy, 1985). So creating a contract that improves the alignment of incentives and takes into account both short and long-term incentives is an important task (Khoroshilov & Narayanan, 2008).
In this study, both the relationship of cash compensation and total compensation with firm performance is examined. This was to evince if the cash compensation was influenced differently than the total compensation by the firm performance. As mentioned before, Hall and Liebman (1998) suggested that stock-options, which are included in the total compensation but not in the cash compensation, have a big influence on the relationship. Also the crisis might have a different impact on the relationship of firm performance and cash compensation than firm performance has on total compensation.
2.3 The relationship between firm performance and executive compensation
In this section, the literature which focuses on the relationship between firm performance and CEO compensation is discussed. This literature is inconclusive of whether this relationship is positive or negative or if there is no relation at all.
An early econometric study of Ciscel and Carroll (1980) shows that there is no clear relation between firm performance and executive compensation. They examined the relationship between an increase in net income (other than an increase in sales) and the executive compensation and found no significant results.
According to Loomis (1982) and Drucker (1984) there appears to be a lack of correlation between the compensation and firm performance (Murphy, 1985). Murphy (1985) published a study which examines the previously found lack of positive pay-performance relationship. According to Murphy, the previous studies exclude some important factors, which cause a bias in the results. Firm size was set to be the only relevant determinant when calculating the executive pay (Ciscel & Carroll, 1980). In addition CEO compensation was mostly calculated by using only cash compensation (salary plus bonus). Murphy (1985) concludes that the cash compensation is not the most important determinant of the relation between pay and performance. Especially the stock options lead to a positive pay-performance relationship. Furthermore, the cross-sectional regressions used will cause an omitted variable bias, since e.g. past performance is an important factor in establishing the new compensation. So the cross-sectional regression should be converted into a time-series regression. Murphy’s (1985) results showed a strong positive and significant relation between firm performance and executive compensation.
How strong and significant this relationship is remains unclear. According to Murphy and Jensen (1990b) for every $1000 increase in shareholder wealth, the CEO compensation goes up by $3.25. They argue that CEOs are not paid according to their performance and find a weak relationship between firm performance and executive pay. However, Hall and Liebman (1998) found a strong relationship between firm performance and CEO pay. They argue that the reason for a strong connection between pay and firm performance is almost entirely the result of the relation between stock-options and firm performance. When CEO’s possess stock options, the changes in the stock prices will be the result of firm performance. This will strengthen the positive relation between firm performance and total compensation. The difference in conclusion of both studies could originate from the different definitions used for compensation. Jensen and Murphy (1990b) analyzed a regression, which used cash compensation (salary and bonus) as a dependent variable. Hall and Liebman (1998)
emphasize the importance of using CEO’s stock options to evince the relationship of remuneration with firm performance. Jensen and Murphy (1990a) used in their sample mostly big companies, which could lead to a bias since these react differently to an increase in firm performance than small firms. The firm performance-pay sensitivity becomes smaller when the firm size increased and this might be a reason why the pay-performance relation was low in the study of Jensen and Murphy (Schaefer, 1998).
Some studies conclude that excessive compensation of directors is negatively related to firm performance (Core et al., 1999; Brick et al., 2006;). Other studies find a weak or positive link between firm performance and executive compensation (Jensen & Murphy, 1990a; Hall & Liebman, 1998; Attaway, 2000).
There also have been studies that show the relationship between firm performance and executive compensation in and between different countries. Zhou (2000) shows that there is a weak link between pay and performance in Canada, although this link is positive. Conyon and Murphy (2000) looked at the differences between the UK and the USA and also found a positive relationship, but the significance of the USA was much higher than in the UK. Also a positive pay-performance relation has been found in a study between Japan and the USA (Kaplan, 1994). On the contrary a study of Duffhues and Kabir (2007) in the Netherlands concluded that there was a negative pay-performance relation.
The unclear pay-performance relationship could be due to the fact that the firms studied in each paper are not homogenous. Looking at different industries or even different countries might lead to different conclusions. The pay-performance relation might be influenced differently by the credit crisis when looking at different industry sectors. Some of the previous studies, which used different sectors, might therefore have mixed results. For this study only one sector is examined, the financial sector, to make sure that the difference in industry does not influence the results.
2.4 Crisis
The recent financial credit crisis has been the worst one since the Great Depression of 1930. It started in the financial sector in 2007 and erupted in 2008 (Melvin & Taylor, 2009; Cleassens & Kodres, 2014). The crisis commenced in the USA and spread fast worldwide. The exact cause is unknown, however, there are a few causes mentioned by various academic papers. Blinder (2009) argues that one of the main causes of the crisis was: ‘the poor incentives of the executives of the banks’ (Bebchuck & Spamann, 2009). He argues that the executives’ remuneration was not properly linked to long-term performance and that this lack
of alignment was the cause of the crisis. However, Fahlenbrach and Stulz (2011) results showed that poor alignment of the incentives of the CEO and the firm performance cannot be the main cause of the financial crisis.
Another cause that is mentioned as a primary cause of the credit crisis is the mortgage market with the subprime mortgages. The increasing housing prices caused banks to be not as strict as usual with providing mortgages. People received mortgages without having the usual background check whether they were viable enough to pay it back. In 2007 this bubble burst and housing prices started to decline fast. Several banks needed government support or were taken over by other banks to prevent them from going bankrupt. Nevertheless on September 15, 2008 Lehman Brothers collapsed and the government refused to save it. When this happened, the crisis could not be retained and it spread across all markets and worldwide (Eichengreen et al., 2012).
Analyzing the financial crisis showed that weak corporate governance originated the crisis as well (Kirkpatrick, 2009). Post-financial crisis there have been a lot of rectifications in the financial regulation to ensure a crisis this big and costly will not happen again. One of the changes was the adoption of Basel Accord III, which required the capital for banks to be set at a higher level. In these new capital requirements a countercyclical capital buffer and a capital conservation buffer were instated. Furthermore the government revised the too-big-to-fail list of banks, which lists the banks that will be saved during times of crisis. The banks that are on this list need to keep a higher amount of capital and are under intense supervision, to make sure they do not take on too much risk (Turner, 2009; Claessens & Kodres, 2014).
Since the credit crisis of 2007 and 2008 started in the financial sector, this is an interesting sector to explore. The high amount of remuneration and specifically the high bonuses in the financial sector have been a critical point since the credit crisis (Turner, 2009; Murdock, 2011; Vemala et al., 2014). After the crisis occurred the Dodd-Frank Wall Street Reform and Consumer Protection Act was established in 2010 with principles to improve the risk management of the financial firms and prevent a crisis with such a large impact to happen again. The main changes in the regulation of the remuneration in the financial sector were the addition of the claw-backs and the say-on-pay (Murdock, 2011). Claw-backs make sure that firms can recover excess pay received by executives when there is for instance a misconduct (Fried & Shilon, 2001). The say-on-pay gives shareholders the opportunity to give a non-binding vote on the amount of remuneration (Cotter et al., 2012). The Dodd-Frank Act requires firms to provide information on how the firm performance is related to the CEO’s compensation (Murdock, 2011). The changes in the financial regulation with the
establishment of the Dodd-Frank act and the required disclosure of information on the relation between firm performance and CEO compensation may have a positive effect on the pay-performance relation. In this study, the effect of this crisis has been examined to see if the relationship between firm performance and CEO compensation has improved.
3. Hypotheses
Based on the literature discussed in the previous section it can be concluded that there is still an indecisive conclusion regarding the CEO pay and performance relationship. As mentioned before post the financial crisis the regulation of the financial sector has become stricter. The Dodd-Frank Act improves the regulation of the remuneration in relation to firm performance. This change in regulation could show that the firm performance CEO compensation relationship has become stronger post-crisis.
Based on this reasoning, two hypotheses are defined: Hypothesis 1
Hypothesis 1a: There is no significant relationship between firm performance and total CEO compensation in the financial sector pre-crisis and during crisis.
Hypothesis 1b: There is a significant positive relationship between firm performance and total compensation post credit crisis.
There will also be a regression, with cash compensation as the dependent variable, which consists of salary and bonus.
The hypothesis here is the same as with the total compensation: Hypothesis 2
Hypothesis 2a: There is no relationship between firm performance and cash compensation in the financial sector in the periods pre-crisis and during crisis.
Hypothesis 2b: There is a significant positive relationship between firm performance and cash compensation post credit crisis.
Overall the main expectation is that the relationship between firm performance and total CEO compensation in the financial sector will be significantly positive post-crisis and that pre and during the crisis this relationship will not be significantly different from 0. The relationship will be stronger due to stricter regulation.
4. Research methodology 4.1 Sample and data collection
For this research data was gathered from the two different datasets about firms in the financial sector. These datasets were obtained from the Wharton Research Data Services and are updated six times a year. From Compustat Executive Compensation in which data of the S&P 1500 is presented, annual data was retrieved about the structure of the CEO compensation and info on the CEO. From the literature it is acquainted that in the S&P 1500 each market segment can be found by looking at SIC codes. The numbers 6000 up to and including 6999 represent all the firms in the financial sector that are present in the database (Allen et al., 2002).
From the Institutional Brokers’ Estimate System (IBES) database the annual return on equity (ROE), return on assets (ROA) and Sales (revenue) could be retrieved. The IBES database contains forecasts and actual data on approximately 69,000 companies from different industries and from different countries. ROE and ROA are used as performance measures and Sales is used as an indicator for the size of the firm.
From the two databases the period from January 2000 until December 2014 was recovered. This is because three periods will be used in the regression, pre-crisis (2000 till 2006); during the crisis (2007 till 2008) and post-crisis (2009 till 2014). In this study the period 2007 and 2008 is used as the period of the crisis. In 2007 the crisis started with the declining housing prices and in 2008 it erupted to the rest of the world with the fall of Lehman Brothers (Baba and Packer, 2009; Melvin & Taylor, 2009; Beltratti & Stulz, 2012; Eichengreen, 2012; Erkens, et al., 2012). These events later turned into the worst credit crisis since the Great Depression 80 years ago.
Since the needed data was not in one database, the datasets had to be merged into two separate datasets, one for ROE and one for ROA. After merging the two databases and the observations which had blanks for the important variables were removed. The ROE regression had a total of 3303 observations, and the ROA regression had 2247 observations. 4.2 Variables and regression equation
In this section the regression variables and the regression equation are examined. First the dependent variables are discussed, then the independent variables. Finally the regression equation is discussed.
4.2.1 Dependent variables: CEO compensation
In this study two different measures of compensation will be used: total compensation and cash compensation.
The first regressions are based on the total compensation which is calculated using the following formula: salary + bonus + other annual + restricted stock grants + LTIP (long term incentive plan) payouts + all other + value of option grants (Yang et al, 2014). Other annual includes income that could not be classified as salary or bonus. This could be for instance: tax reimbursements, perquisites and other personal benefits. All other can be interpreted as all other compensation and can be life insurance premiums or discounted share purchases.
Secondly, cash compensation is used as the dependent variable in the regressions. It is calculated by adding the Salary and Bonus payments. For this regression the same characteristics and variables as in the first regression will be used.
Of both measures of compensation the natural logarithm will be calculated, which is done to adapt the distribution to a normal distribution (Duffhues & Kabir, 2007).
This study uses two different dependent variables to check whether the results are similar or quite different as e.g. Hall and Liebman (1998) suggest in their paper. Also it is interesting to see whether cash compensation was differently affected by the crisis than total compensation. 4.2.2 Independent variables: Firm performance measures
The firm performance in the regression will be measured using two different measures: return on equity (ROE) and return on assets (ROA). Different performance measures will be used to make sure that the results are accountable and not influenced by the use of a certain formula. Return on equity
The return on equity is calculated by dividing the net income by shareholders equity. It has been used as a performance measure in various other academic papers about the subject of pay and performance (Zhou, 2000; Tosi et al., 2000; Attaway, 2000; Brick et al., 2006; Fahlenbrach & Stulz, 2011).
Return on assets
The return on assets is calculated by dividing net income by total assets. It is a performance measure that is often used in financial accounting (Antle & Smith, 1986). ROA has also been widely used as a firm performance measure in other papers (Core et al., 1999; Zhou, 2000;
Tosi et al., 2000; Kato & Kubo, 2004; Brick et al., 2006; Duffhues & Kabir, 2008; Yang et al., 2014; Fahlenbrach & Stulz, 2011).
4.2.3 Control variables
In the regression two variables will be used as control variables: Firm size and Executives’ age. Control variables are entered in the regression to account for their effect on the dependent variable. So the control variables are held constant throughout the regressions. Firm size
In previous literature the importance of using firm size as a variable when measuring the relationship between firm performance and executive compensation is highlighted (Conyon & Murphy, 2000; Zhou, 2000). In 1985 Murphy wrote in his paper that firm size has been proven by econometric studies to have an important effect on executive compensation. Tosi et al. (2000) wrote that firm size contributes to more than 40% of the variance of CEO compensation while firm performance only explains 5% (Baker et al., 1988; Duffhues & Kabir, 2007). Firm size will be measured using the annual sales value of the firms, of which the natural logarithm (ln) will be calculated (Murphy, 1985; Mehran, 1995; Conyon & Murphy, 2000; Zhou, 2000; Brick et al., 2006; Bizjak et al., 2008; Tzioumis, 2008). Since the financial sector does not have similar sales as non-financial firms do, this is taken into consideration. For banks sales consist of the net interest income plus non-interest income. For insurance companies sales is calculated by adding net technical income and net financial income.
CEO age
CEO age is used as a control variable, because the older the CEO is the closer he is to his retirement so he will need a bigger incentive. This is because the CEO will not be working very long for the company and will not profit from the benefits of a long-term incentive. So the costs for the alignment of the goals will be higher (Gibbons & Murphy, 1992). Another study also found a positive relationship between CEO age and compensation (McKnight et al., 2000). In other papers about the relationship between CEO pay and firm performance CEO age has also been used in the regressions as a control variable (Conyon & Murphy, 2000; Attaway, 2000; Tzioumis, 2008).
4.3 Regression equation
First the data was split into three periods; pre-crisis (2000-2006); during the crisis (2007-2008) and post-crisis (2009-2014). The observations that missed important variables were then deleted from the dataset. As mentioned before, this results in a total number of 3303 observations for the ROE regressions and of 2247 observations for the ROA regressions. Furthermore, the dataset was winsorized which limits the impact that outliers have on the regression. The extreme values are not omitted from the dataset, but replaced by less extreme values. In this dataset 1% of the highest and lowest values were adjusted to the values closest to these observations.
For the multiple regression analysis the following equation will be used: Model 1
Ln(Total compensation) = α + β1 ∗ Perf measure!!! + β2 ∗ ln(size)!!!+ β3 ∗ age of CEO The performance measure is ROE in the first three regressions and ROA in the last three. An Ordinary Least Squares (OLS) regression is performed using the option cluster to get the cluster-robust standard errors. The data is clustered according to the ticker symbol which is a company specific code. Since the datasets have information on multiple years of the same companies, the regressions will be panel data regressions. Furthermore, the performance measures and ln(size) will make use of the period t-1. This is intended, to regress the performance of last year with the compensation of this year. Compensation changes annually so the performance of last year will be seen in the remuneration of this year (Tosi et al., 2000; Brick et al., 2006). Because of the use of t-1 the period 2000 is not used in the regression; therefore, the period pre-crisis is 2001-2006.
After these regressions have been performed, there will be another six regressions. These will have ln(Total Cash Compensation) as a dependent variable.
The multiple regression equation that is used is: Model 2
Ln(Total cash compensation) = α + β1 ∗ Perf measure!!! + β2 ∗ ln(size)!!!+ β3 ∗ age of CEO
The three periods are calculated separately, this results in six different regression outputs. All the multiple regressions are performed in Stata and are using the cluster robust standard errors. The results are analyzed by looking at correlations between the variables and the significance levels of the t-values. Furthermore, Chow break tests are performed to see if
the coefficients of the parameters in the multiple linear regressions are the same across all time periods.
5. Results
In this section, the most important results will be discussed. First the descriptive statistics will be presented. Then the correlation matrixes will be presented. Finally, there will be an analysis of the results obtained from the regressions. The results are obtained using the dataset with winsorized data. In Appendix A a more extensive table of descriptive statistics is presented.
Table 1: The most important descriptive statistics for the ROE and for the ROA datasets.
Variables Pre_crisis No. of obs. Pre_crisis Mean Dur_crisis No. of obs. Dur_crisis Mean Post_crisis No. of obs. Post_crisis Mean Totalcomp 251 5396.7230 642 4937.5080 1764 5662.4660 Ln total comp 251 8.0786 642 7.9753 1764 8.2637 Cashcomp 895 1757.5540 643 1014.5440 1765 1151.5410 Ln cashcomp 895 7.1200 643 6.6729 1765 6.8220 ROE t-1 798 12.6331 550 6.1111 1538 9.3448 Sales t-1 836 4983.3320 593 4103.8760 1420 5237.9860 Ln(sales) t-1 836 7.0613 593 6.8190 1420 7.1098 Executives’ age 895 55.9263 643 54.1205 1765 55.9193 Variables Pre_crisis No. of obs. Pre_crisis Mean Dur_crisis No. of obs. Dur_crisis Mean Post_crisis No. of obs. Post_crisis Mean Totalcomp 213 5305.0310 430 4871.2720 1045 5781.9740 Ln totalcomp 213 8.0294 430 7.9414 1045 8.2492 Cashcomp 770 1597.3970 431 998.7354 1046 1162.8690 Ln 770 7.0446 431 6.6534 1046 6.8202
cashcomp ROA t-1 770 2.9517 431 2.1011 1046 2.7958 Sales t-1 770 4477.7900 431 4564.1790 1046 5933.5320 Ln(sales) t-1 770 6.9920 431 6.8019 1046 7.1488 Executives’ age 747 55.7564 431 54.1079 1046 55.7352
In table 1 the ROE and ROA summary statistics of the different periods are shown. The ROE is greatly affected by the crisis, a positive mean of 12.63% for the period pre-crisis can be obtained from the table, then during the crisis the mean decreases to 6.11% and post-crisis it increases to 9.34%. However, a similar pattern cannot be observed in the mean of the ROA which stays more or less the same pre-, during and post-crisis. The mean does become lower in the period of the crisis, however it is a small change of 0.75%.
In the total compensation and cash compensation the same pattern, of the mean going down in the period during the crisis and going up again post-crisis, can be seen but to a smaller extent. The mean of both compensation measures, becomes lower during the crisis and rises again post-crisis. This can also be seen in the means of the ln of compensation. Furthermore, the sales were lower in the period during the crisis than in the periods pre- and post-crisis.
The declines of the averages of all the different variables could be the effect of the credit crisis. During the crisis the reserves of financial institutions deteriorated fast and a lot of companies went bankrupt or where taken over by other companies (Beltratti & Stulz, 2012). This is why the ROE and ROA were lower during the crisis, the net income was low and sometimes even negative. The ROE shows a stronger decline in the period of the crisis than the ROA does. This could be the result of the equity deterioration. Firms had no capital left to pay their executives and started firing them or stopped paying-out bonuses.
Table 2: Shows the correlation matrixes for ROE and ROA Variable 1. 2. 3. 4. 5. 1. ln Tot comp 1 2. ln Cash comp 0.5778*** 1 3. ROE t-1 0.1537*** 0.1394*** 1 4. ln Sales t-1 0.4936*** 0.4956*** 0.1482** 1 5. Executive’s age 0.6760*** 0.1332*** 0.0258 0.0591*** 1 Variable 1. 2. 3. 4. 5. 1. ln Tot comp 1 2. ln Cash comp 0.5976*** 1 3. ROA t-1 0.1589*** 0.0417** 1 4. ln Sales t-1 0.7076*** 0.5077*** 0.1074*** 1 5. Executive’s age 0.0709*** 0.1441*** -0.0194 0.0438** 1 * p<.10, ** p<.05, *** p<.01
The correlation matrixes are presented in table 2, with the corresponding significance levels. These matrixes show the results for the whole period, in Appendix B the correlation matrixes of the different periods; pre-, during- and post-crisis are shown. In the ROE matrix all variables have a significance of at least p<0.05, except for the relation of Executive’s age with ROE which has does not have a significant relation.
In the matrix of ROA more variables have no significant relation. Moreover, cash compensation has no significant relation with the firm performance ROA. The total compensation does have a positive relation (p<0.01).
Table 3: the results obtained from the first six regressions. The ln of total Compensation is used as a dependent variable.
Variable Regression 1: Tot Comp Pre_crisis Regression 2: Tot Comp Dur_crisis Regression 3: Tot Comp Post_crisis Regression 4: Tot Comp Pre_crisis Regression 5: Tot Comp Dur_crisis Regression 6: Tot Comp Post_crisis ROE -0.0015 (-0.26) 0.0049 (1.56) 0.0041 (1.57) ROA -0.0093 (-0.57) 0.0104 (0.81) 0.0268*** (4.01) Ln(Size) 0.4647*** (16.74) 0.4088*** (16.22) 0.3805*** (18.04) 0.0259*** (18.61) 0.4221*** (14.85) 0.3990*** (17.00)
The regression analysis presented in table 3, shows mixed results. In these regressions the winsorized data is used. Appendix C shows the difference in results when outliers are not accounted for. Analyzing the relationship between performance measure ROE and the total compensation shows no change in the different periods. Pre-crisis the firm performance has a negative effect on compensation, but this result is not significant.
The performance measure ROA shows a different change in the relationship between firm performance and total compensation. Pre- and during the crisis the relationship was not significant, so not statistically different from 0. Post-crisis there is a positive relationship, which is significant (p<0.01). However the return on assets performance measure does not show the same relationship and is insignificant at all periods. The control variable for size is significant (p<0.01) for all the periods and has a positive effect. The CEO age variable is mostly not consistent.
Executive’s age 0.0004 (0.06) 0.0001 (0.02) 0.0080* (1.69) 0.0043 (0.57) 0.0025 (0.37) 0.0073 (1.54) Constant 4.8261*** (12.06) 5.1433*** (13.47) 5.0485*** (17.87) 4.4856*** (9.45) 4.9108*** (11.96) 4.9175*** (17.35) No. of obs. 233 541 1371 211 430 1045 R2 0.5400 0.4483 0.4632 0.5394 0.4628 0.5211
The OLS t-statistics are reported in the parentheses. The stars indicate the significance level; *(10%), **(5%), ***(1%).
Table 4 shows the results obtained from the regressions using ln Cash Compensation as a dependent variable. Variable Regression 1: Cash Comp Pre_crisis Regression 2: Cash Comp Dur_crisis Regression 3: Cash Comp Post_crisis Regression 4: Cash Comp Pre_crisis Regression 5: Cash Comp Dur_crisis Regression 6: Cash Comp Post_crisis ROE 0.0096*** (2.61) 0.0013 (0.79) -0.0016 (-0.81) ROA 0.0057 (0.54) 0.0011 (0.17) -0.0112 (-1.44) Ln(Size) 0.2846*** (14.41) 0.1692*** (9.15) 0.1673*** (8.56) 0.2853*** (14.68) 0.1751*** (8.47) 0.1758*** (8.88) Executive’s age 0.0140*** (2.67) 0.0053 (1.08) 0.0104*** (3.14) 0.0135*** (2.67) 0.0053 (1.11) 0.0132*** (3.75) Constant 4.1801*** (12.49) 5.2374*** (18.47) 5.0665*** (22.07) 4.2836*** (13.25) 5.1747*** (17.94) 4.8609*** (20.57)
The regression output using ln Cash Compensation as the dependent variable presents different results. The ROE coefficient is significant (p<0.01) and positive for the period pre- crisis and changes to not significant for the periods during and post-crisis. The ROA is not significant for all periods.
The firm size and executive’s age are significant at a 1% level just like in the first regressions. The CEO’s age is significant for both ROE and ROA in the periods pre- and post-crisis and not significant for the period during the crisis.
The two dependent variables total compensation and cash compensation give different results. In the first regressions ROE has no statistical significance for the pay-performance relation and ROA is significant (p<0.01) for the period post-crisis.
In the second regressions, there is significance (p<0.01) for the pay-performance relation of ROE and cash compensation in the period pre-crisis, the ROA has no significant coefficients. The firm size is significant (p<0.01) in both regressions for all periods. The main difference between the regressions is the significance of the ROA, in the first regressions in the period post-crisis and in the second regressions ROE is significant in the period pre-crisis.
Table 5: Results of the Chow Break test Pre-During ROE During-Post ROE Pre-During ROA During-Post ROA F-test 2.13 1.7 0.74 0.29 P-value 0.1454 0.1933 0.3902 0.5876
In table 5 the results of the Chow break test are shown. The first two are calculated using the ROE data file and the other two are from the ROA file. The Chow break test indicates whether there is a structural break in the results. A structural break is present when there is an abrupt change in the data at a point in time. If there is structural break in the data this could mean that there is a significant change of the pay-performance relation in the periods. Four tests were performed on the periods pre-during and during-post for both ROE and ROA. The results were all not significant (p>0.10). This concludes that the statistical break is not
No. of obs. 791 542 1372 747 431 1046
R2 0.4021 0.2046 0.2122 0.3657 0.2228 0.2517
The OLS t-statistics are reported in the parentheses. The stars indicate the significance level; *(10%), **(5%), ***(1%).
significantly different from 0 and that there is no abrupt change in the data between the three periods.
6. Discussion
6.1 Analysis of the results
Analyzing the literature on the relationship of firm performance and CEO compensation, shows that there are mixed conclusions. This study adds two elements which have not, or only by a few studies, been discussed in the previous literature. The effect of the 2007-2008 credit crisis on the pay-performance relationship and in addition, the impact on the CEO compensation and firm performance relationship in the financial sector.
This study added the credit crisis to find out whether the firm performance and CEO compensation relation has changed in the periods pre-, during- and post-crisis. Furthermore, this study used data from the financial sector of the S&P 1500 for the three different periods. Both ln total compensation and ln cash compensation were used as the dependent variable, to analyze whether there is a difference in conclusion between both.
From the descriptive statistics tables the period of the crisis can be clearly distinguished. During the crisis the averages were lower for all variables; total compensation, cash compensation, sales and for the firm performance measures, than in the periods pre- and post-crisis.
The correlation matrixes show the correlations between the variables. Both firm performance measures show significant (p<0.01) positive correlations between the performance measures and both measures of compensation. The correlation matrixes for the different periods are presented in Appendix B. For every period and for both performance measures the total compensation has a positive significant (p<0.01) correlation with firm performance. For the cash compensation a different pattern can be observed, the periods pre- and during the crisis show a significant (p<0.01) positive relation between firm performance and cash compensation. The period pre-crisis shows a higher correlation than the period during the crisis and the period post-crisis shows a not significant relation. However, Jensen and Murphy (1990a) indicated no significant correlation between firm performance and CEO cash compensation. This difference could be explained by looking at the different time period that was used in Murphy’s study. In the 1990s equity-based compensation spread very quickly throughout the United States (Ofek & Yermack, 2000). Since Murphy’s (1990a) study covered a dataset of the years 1970 to 1996, the correlation found is mostly based on the period before the extensive growth in equity-based compensation. After the fast growth of
equity-based compensation, the equity-based compensation became more important in the total compensation. Also Hall and Liebman (1998) conclude that the pay-performance relation is mostly obtained by the relation of stock-based compensation and firm performance and not by the relation with cash compensation. Furthermore, the correlation between compensation and the control variable firm size is significant (p<0.01) for both periods and performance measures. Zhou (2000) concludes a strong positive correlation of firm size and total CEO compensation. Executives’ age is not significantly correlated to ROE and also not to ROA, this could be due to the fact that in periods of economic distress the CEO age does not influence performance.
As mentioned before, the regressions show mixed results. Two different equations were used to conclude whether the relation is significant or not. The first hypothesis uses total compensation as the dependent variable and the second hypothesis uses cash compensation as the dependent variable.
The first hypothesis 1a is accepted for both the ROE and ROA performance measures. Pre- and during-crisis the total compensation performance relationship is not significantly different from 0. In the period post-crisis the relationship is significant (p<0.01) and positive for the performance measure ROA, however it is not significant for ROE. So, the hypothesis 1b is supported using ROA, this could elaborate that the new regulations set post-credit crisis by the Dodd-Frank act had a positive impact on the pay-performance relation. However, when looking at the ROE the hypothesis is rejected because of an insignificant positive result. So both performance measures conclude a positive relationship between firm performance and CEO compensation post-crisis, but since the ROE coefficient is not significant it cannot be reliable accepted.
In the second hypotheses the relation of firm performance with ln cash compensation is analyzed. In the first part of the hypothesis 2a the relation is analyzed pre- and during the crisis. The regressions have mixed results, the ROE regression rejects the hypothesis and the ROA regression accepts it. The ROE shows a significant (p<0.01) positive result for the period pre-crisis. Furthermore, for the period post-crisis both performance measures reject het hypothesis and have a negative but not significant result. In the financial crisis a lot of firms were taken over or went bankrupt. At the end of 2008, most banks had experienced a large decline in equity (Beltratti & Stulz, 2012). The deterioration of capital impacted the remuneration, CEO’s were fired and cash bonuses could not be realized. This might be an explanation of the negative, although not significant, pay-performance relation post-crisis.
Firm size has in all regressions a positive significant (p<0.01) relation with the different compensation measures. These results are confirmed by the literature (Murphy, 1985; Mehran, 1995; Conyon & Murphy, 2000; Zhou, 2000; Brick et al., 2006; Bizjak et al., 2008; Tzioumis, 2008). The relation of CEO age with compensation shows mixed results. In the first regressions with ln total compensation as the dependent variable, the relation is mainly not significant. In the second regression analysis the CEO age is significant for the periods pre- and post-crisis and not significant for the period during the crisis. This could be the result of firm takeovers and bankruptcies, so that during economic distress CEO age is not a significant variable.
Comparing the results of total compensation and cash compensation cannot lead to a reliable conclusion. The results are weakly aligned with the literature and the hypotheses, firm performance and total compensation have a stronger positive relationship post-crisis, while cash compensation has a negative relationship. However, these results are not all significantly different from 0. It can be concluded that the total compensation has a stronger effect on the pay-performance relationship after the credit crisis and that this relationship is stronger than the pay-performance relation using cash compensation. This could be because the pay-performance relation is more dependent on equity-based compensation (Hall & Liebman, 1998).
6.2 Limitations and future research
This study has several limitations, which should be taken into account when interpreting the findings of this study.
First, in this study only two different performance measures were used in the regression analysis. Both these performance measure are seen as accounting based performance measures (Antle & Smith, 1986). In other studies other performance measures e.g. Tobins’ Q or stock returns were applied, these other measures might give a different conclusion (Mehran, 1995; Duffhues and Kabir, 2007; Tzioumis, 2008).
Second, when merging the two databases observations that had missing data items for important variables (like ROE, ROA, Executives’ age or sales value) were removed from the datasets and large outliers were adjusted by winsorizing the data.
When the observations that had missing data items were removed the total number of observations that was left was not very high in certain periods. Therefore generalizing the results across the whole of United States of America is difficult. In particular the period
pre-crisis had a lot of missing observations for the variable total compensation, since these observations were removed from the dataset this could have lead to a bias in the results.
Since the crisis was a global one and did not only influence the USA, the regressions could in further research be analyzed in other continents or countries. Adding the other continents/countries to the dataset will give a higher amount of observations for the dataset and that will make the results more reliable for the whole financial sector.
7. Conclusion
This study examined the impact of the financial crisis of 2007-2008 on the relationship between CEO compensation and firm performance in the financial sector. Two different measures of compensation were used; total compensation and cash compensation. Also two different performance measures were used; return on equity (ROE) and return on assets (ROA).
The first hypothesis stated that there was no relation between firm performance and total compensation pre- and during-crisis and a significant positive relation post-crisis. The results of this study concluded mixed results for this hypothesis. The performance measure ROE was not significant for all periods, so rejected the hypothesis of the period post-crisis. The ROA performance measure accepted both sub-hypotheses. Both the pay-performance relationships are positive post-crisis, but do not result a significant result for both performance measures. Furthermore, the correlations between total compensation and the performance measures are significant for all three periods.
The second hypothesis used cash compensation as the dependent variable, to analyze the relation between firm performance and CEO compensation. This regression also resulted in mixed conclusions. Pre-crisis the ROE showed a significant positive pay-performance relationship. However, during and post-crisis this relationship was not significant. The ROA performance measure concluded no significant results for all periods. The correlation tables show a significant positive relation between firm performance and cash compensation for the periods pre- and during the crisis, for the period post-crisis this correlation show no significant result. The main thing that was remarkable was that both performance measures showed a negative pay-performance relation and a not significant correlation between performance and cash compensation for the post-crisis period, although the regression results were both not significant.
Since the results of this study are not completely decisive, it is difficult to give a reliable answer to the question of whether the pay-performance relationship has changed in
the past 14 years. The results show that the relationship between total compensation has become stronger positively related to firm performance, although this relation is not significant for the ROE performance measure. Also the cash compensation relationship with firm performance has changed due to the crisis, but this relationship has become negative post-crisis, although not significant. The conclusion indicates that the Dodd-Frank act along with the crisis improved the pay-performance relationship when looking at total compensation. And that there is a stronger relationship between firm performance and total compensation than there is with firm performance and cash compensation. This is the result of the strong relation between equity-based compensation and firm performance.
References
Ackermann, J. (2008). The subprime crisis and its consequences. Journal of Financial Stability, 4(4), 329-337.
Allen, L., Bali, T. G., Tang, Y. (2002). Does Systemic Risk in the Financial Sector Predict Future Economic Downturns? Baruch College Working Paper.
Antle, R. and Smith A. (1986). An Empirical Investigation of the Relative Performance Evaluation of Corporate Executives. Journal of Accounting Research, 24(1), 1-39.
Attaway, M. C. (2000). A Study of the Relationship Between Company Performance and CEO Compensation. American Business Review, 18(1), 77-86.
Baba, N. and Packer, F. (2009). From turmoil to crisis: Dislocations in the FX swap market before and after the failure of Lehman brothers. Journal of International Money and Finance, 28(8), 1350-1374.
Baker, G. P., Jensen, M. C. and Murphy K. J. (1988). Compensation and Incentives: Practice vs. Theory. The Journal of Finance, 43(3), 593-616.
Bebchuk, L. A. and Fried, J. M. (2003). Executive Compensation as an Agency Problem. Journal of Economic Perspectives, 17, 71–92.
Bebchuk, L. A. and Spamann, H. (2009). Regulating Bankers’ Pay. Georgetown Law Review, 98, 247-287.
Beltratti, A. and Stulz, R. M. (2012). The credit crisis around the globe: Why did some banks perform better? Journal of Financial Economics, 105(1), 1-17.
Bizjak, J. M., Lemmon, M. L. and Naveen, L. (2008). Does the use of peer groups contribute to higher pay and less efficient compensation? Journal of Financial Economics, 90(2), 152-168.
Brick, I. E., Palmon, O. and Wald, J. K. (2006). CEO compensation, director compensation, and firm performance: Evidence of cronyism? Journal of Corporate Finance, 12(3), 403-423. Ciscel, D. H. and Carroll, T.M. (1980). The Determinants of Executive Salaries: An Econometric Survey. The Review of Economics and Statistics, 62(1), 7-13.
Claessens, S. and Kodres L. E. (2014). The Regulatory Response to the Global Financial Crisis: Some Uncomfortable Questions. Working paper No. 14/46, International Monetary Fund.
Conyon, M. J. and Murphy K. J. (2000). The Prince and the Pauper? CEO pay in the United States and United Kingdom. The Economic Journal, 110(467), F640-F671.
Cotter, J. F., Palmiter, A. R. and Randall, T. S. (2012). Dodd-Frank's Say on Pay: Will It Lead to a Greater Role for Shareholders in Corporate Governance? CORNELL L. REV, 1213. Core, J. E., Holthausen, R. W. and Larcker, D. F. (1999). Journal of Financial Economics, 51(3), 371-406.
Duffhues, P. and Kabir, R. (2007). Is the pay-performance relationship always positive? Evidence from the Netherlands. Journal of Multinational Financial Management, 18(1), 45-60.
Eichengreen, B., Mody, A., Nedeljkovic, M. and Sarno, L. (2012). How the Subprime Crisis went global: Evidence from bank credit default swap spreads. Journal of International Money and Finance, 31(5), 1299-1318.
Erkens, D. H., Hung, M. and Matos, P. (2012). Corporate governance in the 2007-2008 financial crisis: Evidence from financial institutions worldwide. Journal of Corporate Finance, 18(2), 389-411.
Fahlenbrach, R. and Stulz, R. M. (2011). Bank CEO incentives and the credit crisis. Journal of Financial Economics, 99(1), 11-26.
Fama, E. F. and Jensen, M. C. (1983). Separation of Ownership and Control. The Journal of Law & Economics, 26(2), 301-325.
Fried, J. and Shilon, N. (2001). Excessive-pay clawbacks. Harvard Law School working paper.
Gibbons, R. and Murphy K. J. (1992). Does Executive Compensation Affect Investment? Journal of Applied Corporate Finance, 5, 99-109.
Hall, B. J. and Liebman, J. B. (1998). Are CEOs Really Paid Like Bureaucrats? The Quarterly Journal of Economics, 113, 653-691.
Haubrich, J. G. (1994). Risk Aversion, Performance Pay, and the Principal-Agent Problem. Journal of Political Economy, 102(2), 258-276.
Healy, P. M. (1985). The Effect of Bonus Schemes on Accounting Decisions. Journal of Accounting and Economics, 7(1-3), 85-107.
Hölmstrom, B. (1979). Moral Hazard and Observability. The Bell Journal of Economics.
10(1), 74-91.
Jensen, M. C. and Murphy K. J. (1990a). CEO Incentives-It’s Not How Much You Pay, But How. Harvard Business Review, 3, 138-153.
Jensen, M. C. and Murphy, K. J. (1990b). Performance pay and top-management incentives. Journal of Political Economy, 98(2), 225-264.
Jensen, M. C. and Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4), 305-360.
Joyce, W. B. (2001). Return and Reward: Bank Performance and CEO Compensation. American Business Review, 19(2), 93-99.
Kaplan, S. (1994). Top Executive Rewards and Firm Performance: A Comparison of Japan and the U.S. Journal of Political Economy, 102, 510-546.
Kato, T. and Kubo, K. (2006). CEO compensation and firm performance in Japan: Evidence from new panel data on individual CEO pay. Journal of the Japanese and International Economies, 20(1), 1-19.
Khoroshilov, Y. and Narayanan, M. P. 2008. The role of profit-based and stock-based components in incentive compensation. Journal of Financial Intermediation, 17(3), 357-378. Kirkpatrick, G. (2009). The corporate governance lessons from the financial crisis. OECD Journal: Financial Market Trends, 1, 61–87.
McKnight, P. J., Tomkins, C., Weir, C. and Hobsen, D. (2000). CEO Age and Top Executive Pay: A UK Empirical Study. Journal of Management and Governance, 4(3), 173-187.
Mehran, H. (1995). Executive compensation structure, ownership, and firm performance. Journal of Financial Economics, 38(2), 163-184.
Melvin, M. and Taylor, M. P. (2009). The global financial crisis: Causes, threats and opportunities. Introduction and overview. Journal of International Money and Finance, 28(8), 1243-1245.
Murdock, C., W. (2011). The Dodd-Frank Wall Street Reform and Consumer Protection Act: What Caused the Financial Crisis and Will Dodd-Frank Prevent Future Crises? Southern Methodist University Law Review, 64, 1243–1325.
Murphy, K.J. (1985). Corporate Performance and Managerial Remuneration: An Empirical Analysis. Journal of Accounting and Economics, 7, 11-42.
Ofek, E. and Yermack, D. (2000). Taking Stock: Equity-Based Compensation and the Evolution of Managerial Ownership. The Journal of Finance, 55(3), 1367-1384).
Ross, S. A. (1973). The Economic Theory of Agency: The Principal’s Problem. The American Review, 63(2), 134-139.
Sanders, A. (2008). The subprime crisis and its role in the financial crisis. Journal of Housing Economics, 17(4), 254-261.
Schaefer, S. (1998). The Dependence of Pay-Performance Sensitivity on the Size of the Firm. The Review of Economics and Statistics, 80(3), 436-443.
Tosi, H. L., Werner, S., Katz, J. P. and Gomez-Mejia, L. R. (2000), How Much Does Performance Matter? A Meta-Analysis of CEO Pay Studies. Journal of Management, 26(2), 301-339.
Turner, A. (2009). A Regulatory Response to the Global Banking Crisis. Financial Services Authority, March.
Tzioumis, K. (2008). Why do firms adopt CEO stock options? Evidence from the United States. Journal of Economic Behavior & Organization, 68(1), 100-111.
Vemala, P., Nguyen, L., Nguyen, D. and Kommansani, A. (2014). CEO Compensation: Does Financial Crisis Matter? International Business Researchi, 7(4), 125-131.
Yang, F., Dolar, B. and Mo, L. (2014). CEO Compensation and Firm Performance: Did the 2007-2008 Financial Crisis Matter?, Journal of Accounting and Finance, 14(1), 137-146. Zhou, X. (2000). CEO pay, Firm Size and Corporate Performance: Evidence from Canada. The Canadian Journal of Economics, 33(1), 213-251.
Appendixes Appendix A
This Appendix shows the descriptive statistics of the variables included in the regressions. The first three tables contain the three periods of the ROE dataset and the last three contain the ROA periods.
The descriptive statistics of the variables in the regression for the pre-crisis period of the ROE data
Variables No. of obs.
Mean Sd. Min Max P25 Median P75
Totalcomp 251 5396.7230 6013.6210 397.8080 26961.3000 5662.4660 1366.5740 6547.6110
comp Cashcomp 895 1757.5540 1787.4240 254.7120 8625 700.9180 1087.6460 2014.4320 Ln cashcomp 895 7.1200 0.7949 5.5401 9.0624 6.5524 6.9918 7.6081 ROE t-1 798 12.6331 9.5230 -53.5000 51.8100 8.7000 12.8300 16.6500 Sales t-1 836 4983.3320 12404.8400 64.1040 83780 337.2090 915.9960 3264.6310 Ln(sales) t-1 836 7.0613 1.6114 4.1605 11.3360 5.8207 6.8200 8.0909 Executives’ age 895 55.9263 6.6618 40 73 52 56 60
The descriptive statistics of the variables in the regression for the period during the crisis of the ROE data
Variables No. of obs.
Mean Sd. Min Max P25 Median P75
Totalcomp 642 4937.5080 5706.9550 397.8080 26961.3000 1348.3230 2746.9960 5995.3760 Ln total comp 642 7.9753 1.0280 5.9860 10.2022 7.2066 7.9182 8.6987 Cashcomp 643 1014.5440 1071.9160 254.7120 8625 532 750 1000 Ln cashcomp 643 6.6729 0.6227 5.5401 9.0624 6.2766 6.6201 6.9078 ROE t-1 550 6.1112 16.5054 -53.5000 51.8100 2.6000 7.8950 13.4200 Sales t-1 593 4103.8760 11041.4000 64.1040 83780 271.3900 680.1420 2742.5780 Ln(sales) t-1 593 6.8190 1.6102 4.1610 11.3360 5.6040 6.5223 7.9167 Executives’ age 643 54.12053 6.8812 40 73 49 54 58
The descriptive statistics of the variables in the regression for the post-crisis period of the ROE data
Variables No. of obs.
Totalcomp 1764 5662.4660 5106.8190 397.8080 26961.3000 2068.9180 3883.1730 7689.3480 Ln total comp 1764 8.2640 0.8989 5.9860 10.2022 7.6348 8.2644 8.9476 Cashcomp 1765 1151.5410 1128.2170 254.7120 8625 645.9620 850 1100 Ln cashcomp 1765 6.8220 0.5909 5.5401 9.0624 6.4707 6.7452 7.0031 ROE t-1 1379 9.3448 11.0333 -53.5000 51.8100 5 8.8400 12.7000 Sales t-1 1420 5237.9860 13228.2600 64.1040 83780 368.8025 908.8820 3620.2440 Ln(sales) t-1 1420 7.1098 1.5929 4.1605 11.3360 5.9103 6.8122 8.1943 Executives’ age 1765 55.9193 6.5175 40 73 51.5000 56 60
The descriptive statistics of the variables in the regression for the pre-crisis period of the ROA data
Variables No. of obs.
Mean Sd. Min Max P25 Median P75
Totalcomp 213 5305.0310 6111.7620 366.3350 28887.5300 1284.5170 2949.2990 6613.0950 Ln total comp 213 8.0294 1.0553 5.9035 10.2712 7.1581 7.9893 8.7968 Cashcomp 770 1597.3970 1566.5120 225 8000 673.2150 1024.0080 1875 Ln cashcomp 770 7.0446 0.7782 5.4161 8.9872 6.5121 6.9315 7.5364 ROA t-1 770 2.9517 3.8612 -5.5600 22.0100 1.0500 1.5500 3.7800 Sales t-1 770 4477.7900 11281.5100 62.9710 85116 316.3400 878.1520 3060.9000 Ln(sales) t-1 770 6.9920 1.5853 4.1427 11.3518 5.7568 6.7778 8.0265 Executives’ age 747 55.7564 6.3678 40 72 52 56 60
The descriptive statistics of the variables in the regression for the period during the crisis of the ROA data
Variables No. of obs.
Mean Sd. Min Max P25 Median P75
Totalcomp 430 4871.2720 5812.2680 366.3350 28887.5300 1289.9040 2652.4570 6222.7620 Ln total comp 430 7.9414 1.0448 5.9035 10.2712 7.1623 7.8832 8.7360 Cashcomp 431 998.7354 1058.4300 225 8000 521.6670 750 990.2500 Ln cashcomp 431 6.6534 0.6272 5.4161 8.9872 6.2570 6.6201 6.8980 ROA t-1 431 2.1011 4.1250 -5.5600 22.0100 0.2900 1.0600 3.3300 Sales t-1 431 4564.1790 12531.9000 62.9710 85116 250.9890 608.0940 2763.1120 Ln(sales) t-1 431 6.8019 1.6690 4.1427 11.3518 5.5254 6.4103 7.9241 Executives’ age 431 54.1079 6.5803 40 72 50 54 58
The descriptive statistics of the variables in the regression for the post-crisis period of the ROA data
Variables No. of obs.
Mean Sd. Min Max P25 Median P75
Totalcomp 1046 1156.7880 1114.5590 225 8000 629 859.7865 1100 Ln total comp 1046 6.8202 0.6058 5.4161 8.9872 6.4441 6.7567 7.0031 Cashcomp 1046 2.7958 4.0573 -5.5600 22.0100 0.8000 1.2600 3.7100 Ln cashcomp 1046 5933.5320 14849.6200 62.9710 85116 362.7370 925.4975 3958.9990 ROA t-1 1046 7.1488 1.6498 4.1427 11.3518 5.8937 6.8303 8.2837 Sales t-1 1046 55.7352 6.4826 40 72 51 56 60 Ln(sales) t-1 1045 5781.9740 5372.4950 366.3350 28887.5300 1914.3520 3857.0860 8222.8300 Executives’ age 1045 8.2492 0.9467 5.9035 10.2712 7.5571 8.2577 9.0147
Appendix B
This Appendix shows the correlation matrixes for the three periods for both performance measures.
The correlation matrixes for the pre-crisis period of ROE
Variable 1. 2. 3. 4. 5. 1. ln Tot comp 1 2. ln Cash comp 0.6167*** 1 3. ROE t-1 0.1413** 0.2365*** 1 4. ln Sales t-1 0.7038*** 0.6136*** 0.2152** 1 5. Executive’s age 0.0231 0.1583*** 0.0352 0.0676* 1
The correlation matrixes for the during-crisis period of ROE
Variable 1. 2. 3. 4. 5. 1. ln Tot comp 1 2. ln Cash comp 0.4798*** 1 3. ROE t-1 0.1376*** 0.0986** 1 4. ln Sales t-1 0.5909*** 0.4446*** 0.1232*** 1 5. Executive’s age 0.0307 0.0246 -0.0185 0.0161 1
The correlation matrixes for the post-crisis period of ROE
Variable 1. 2. 3. 4. 5. 1. ln Tot comp 1 2. ln Cash comp 0.5567*** 1 3. ROE t-1 0.1416*** 0.0317 1 4. ln Sales t-1 0.6759*** 0.4502*** 0.1318*** 1 5. Executive’s age 0.0979*** 0.1384*** 0.0125 0.0588** 1
The correlation matrixes for the pre-crisis period of ROA
Variable 1. 2. 3. 4. 5. 1. ln Tot comp 1 2. ln Cash comp 0.6156*** 1 3. ROA t-1 0.0719*** 0.0783** 1 4. ln Sales t-1 0.7341* 0.5982*** 0.0941*** 1 5. Executive’s age 0.0312 0.1319*** -0.0285 0.0361 1
The correlation matrixes for the during-crisis period of ROA Variable 1. 2. 3. 4. 5. 1. ln Tot comp 1 2. ln Cash comp 0.6046*** 1 3. ROA t-1 0.0947** 0.0430 1 4. ln Sales t-1 0.6789*** 0.4687*** 0.0791 1 5. Executive’s age 0.0411 0.0730*** -0.0233 0.0383 1
The correlation matrixes for the post-crisis period of ROA
Variable 1. 2. 3. 4. 5. 1. ln Tot comp 1 2. ln Cash comp 0.5788*** 1 3. ROA t-1 0.1974*** -0.0214 1 4. ln Sales t-1 0.7114*** 0.4750*** 0.1206*** 1 5. Executive’s age 0.0726** 0.1610*** -0.0274 0.0376 1 Appendix C
This Appendix shows the results of the regression using the data before the outliers were adjusted.
The results obtained from the first six regressions. The ln of total Compensation is used as a dependent variable. Variable Regression 1: Tot Comp pre_crisis Regression 2: Tot Comp dur_crisis Regression 3: Tot Comp post_crisis Regression 4: Tot Comp pre_crisis Regression 5: Tot Comp dur_crisis Regression 6: Tot Comp post_crisis ROE 0.0085 (1.42) 0.0015 (0.69) 0.0070** (2.24) ROA -0.0192 (-1.25) 0.0017 (0.15) 0.0024 (1.49) Ln(Size) 0.4326*** (11.64) 0.3733*** (14.76) 0.3687*** (15.36) 0.4743*** (14.49) 0.4519*** (15.09) 0.4079*** (16.35) Executive’s age 0.0191** (2.18) 0.0086 (1.23) 0.0103* (1.91) 0.0035 (0.46) 0.0007 (0.11) 0.0068 (1.4) Constant 4.0265*** (6.77) 4.8887*** (12.1) 4.9441*** (14.71) 4.6103*** (8.41) 4.8325*** (11.8) 4.9425*** (16.67) No. of obs. 230 539 1369 211 262 1043 R2 0.4501 0.3294 0.3844 0.5212 0.4697 0.5024
The OLS t-statistics are reported in the parentheses. The stars indicate the significance level; *(10%), **(5%), ***(1%).
The results obtained from the regressions using ln Cash Compensation as a dependent variable. Variable Regression 1: Cash Comp pre_crisis Regression 2: Cash Comp dur_crisis Regression 3: Cash Comp post_crisis Regression 4: Cash Comp pre_crisis Regression 5: Cash Comp dur_crisis Regression 6: Cash Comp post_crisis ROE 0.0033*** (5.35) -0.0003 (-0.55) -0.0012 (-0.86) ROA -0.0002 (-0.02) -0.00005 (-0.01) -0.0003 (-0.42) Ln(Size) 0.3041*** (14.24) 0.1779*** (9.56) 0.1653*** (8.42) 0.2916*** (12.97) 0.2244*** (7.66) 0.1711*** (8.7) Executive’s age 0.0156*** (2.75) 0.0037 (0.75) 0.0098*** (3.02) 0.0128*** (2.59) 0.0053 (1.09) 0.0134*** (3.95) Constant 4.0370*** (10.82) 5.2688*** (18.52) 5.0976*** (22.37) 4.3125*** (12.52) 4.8826*** (14.96) 4.8526*** (20.8) No. of obs. 791 540 1370 742 260 1042 R2 0.3941 0.2005 0.173 0.39593 0.2917 0.2417
The OLS t-statistics are reported in the parentheses. The stars indicate the significance level; *(10%), **(5%), ***(1%).
