The relationship between CEO compensation and firm
performance
Evidence for 2003-2009 from the Netherlands
Erik de Jong
Master Thesis
University of Groningen
Faculty of Economics and Business
MSc Business Administration, Finance
Studentnumber:
s1555863
Supervisor:
Dr. A. Schertler
The relationship between CEO compensation and firm
performance
Evidence for 2003-2009 from the Netherlands
Abstract
This thesis investigates whether the agency theory holds that CEO compensation should be aligned with firm performance. I have collected a dataset which contains four parts of the compensation package granted to CEOs of Dutch listed companies from 2003 to 2009 in order to analyze if CEO compensation is determined by firm performance, if the incentives of CEOs are aligned with those of the shareholders’ and whether there exists an optimal ratio of variable to fixed compensation with respect to three different firm performance measures. The analysis controls for two proxies of firm size, firm age, leverage, time effects and industry effects. My findings indicate that firm performance is a positive determinant of CEO compensation, although few results show a negative relationship. Furthermore, results do not show persuasive evidence that CEOs incentives are aligned with shareholders but do show that an optimal ratio between variable and fixed compensation exists. The results are at large in line with prior research.
JEL classification: G30, J31, J33, M52
Table of contents
1. Introduction...4
2. Literature Review...7
2.1 Theoretical background and hypotheses ...7
2.1.1 CEO compensation as a function of firm performance ...8
2.1.2 Firm performance as a function of CEO compensation ...10
2.2 Empirical findings...12
3. Data ...13
3.1 Data collection ...13
3.2 CEO compensation variables ...14
3.3 Firm performance variables...17
3.4 Control variables ...18
3.5 Correlation ...20
4. Methodology ...21
4.1 CEO compensation as a function of firm performance ...21
4.2 Firm performance as a function of CEO compensation...22
4.3 Estimation Strategy...23
5. Results ...24
5.1 CEO compensation as a function of firm performance ...24
5.1.1 Robustness check ...28
5.1.2 Lagged Relationship...28
5.1.3 Robustness check ...29
5.2 (Future) Firm performance as a function of CEO compensation ...30
5.2.1 The optimal contract ...30
6. Conclusion ...34
6.1 CEO compensation as a function of firm performance ...34
6.2 Firm performance as a function of CEO compensation...35
6.3 Limitations and further research...36
1. Introduction
In this study I will analyze the relationship between the compensation of the Chief Executive Officer (CEO) and firm performance for Dutch listed firms. First I will examine whether CEO compensation is determined by firm performance (i.e. ex-post compensation), second I will analyze whether future firm performance is determined by CEO compensation (i.e. ex-ante compensation) and finally I will examine whether there exists an optimal ratio between variable compensation and fixed compensation.
The total compensation granted to the CEO has risen substantially over the past years, which is primarily driven by an increase in equity based compensation, such as stocks and options (Murpy, 1999; Bebchuk and Fried, 2004). The relationship between CEO compensation and firm performance is of great public concern and received a lot of attention in the media, which especially holds when a link between compensation and firm performance is absent. When this occurs, it is doubtful that the assets of public companies are being managed efficiently, which eventually will have a negative impact on shareholders’ value (Hall and Liebman, 1998). For example, CEOs of Dutch multinationals like Ahold and Royal Dutch Shell were given generous compensation at the time these firms were performing poorly. Anders Moberg, former CEO of Ahold, received a total remuneration of over 26 million euro in only four years. Besides, in a report published in December 2006, the Monitoring Commission on Corporate Governance in the Netherlands has expressed its concern about the lack of transparency on executive compensation of Dutch listed firms (Duffhues and Kabir, 2007).
any areas of non-compliance in light of their own particular circumstances (Arcot et al, 2010). This implies that either a company complies with the code and aims to improve their corporate governance or companies rather follow the regulatory requirements without making a serious commitment to corporate governance: they depart from best practice and provide an explanation which is uninformative.
What exactly is corporate governance? The dominant view in economics is that corporate governance deals with the way suppliers of finance assure themselves of getting a return on their investment (Shleifer and Vishny, 1997). A prominent theory within corporate governance research is the principal-agent theory. A classic example of this theory is the conflict of interest between shareholders and the CEO. If shareholders had complete information regarding the CEO’s activities and the firm’s investment opportunities, they could design a contract specifying and enforcing the actions to be taken in each state of the world. However, the actions taken by a CEO and the firm’s investment opportunities are not perfectly observable by shareholders due to asymmetric information. In these situations the agency theory predicts that compensation policy will be designed to give the CEO incentives to select and implement only those actions that increase shareholder wealth (Jensen and Murphy, 1990), since CEOs are intended to pursue their own interests instead of the interests of their shareholders. In other words, how can the interests of (more or less) uninformed shareholders be aligned with that of a powerful and sometimes opportunistic CEO (Fama and Jensen 1983; Randøy and Nielsen, 2002)? Within this perspective, a potential weak alignment between CEO compensation and firm performance may be due to a lack of correctly designed incentives (Tirole, 2006). Thus, CEO compensation and the design of the compensation package are important governance mechanisms used to align the interests of CEOs and shareholders and to mitigate the agency problem.
They failed to find a significant relationship between either accounting performance or stock return and CEO compensation. To my knowledge, the study by Duffhues and Kabir (2007) was the only one who examined the pay-performance relationship for the Netherlands and they find a significantly negative relationship between CEO compensation and firm performance.
The study of Duffhues and Kabir (2007) differs from this study with respect to their sample period, which contains Dutch listed companies from 1998 until 2001. This sample period was before the introduction of the Code Tabaksblat. Due to the introduction of the Code Tabaksblat I expect to find an increase in the pay-performance relationship. This study differs from previous studies on several other aspects. First, in this study firm performance is being measured by both accounting-based and market-accounting-based variables, like return on assets (ROA), stock return and the Tobin’s Q ratio. Second, the few non-Anglo-American empirical studies use either relatively small or only cross-sectional samples. This study uses both cross-cross-sectional and times-series data (panel data) for all Dutch listed firms. Third, extensive research has been done by using only total CEO compensation. However, few is known about the relationship between the different components of the compensation package with respect to firm performance such as base salary, bonuses, stocks and options. Authors like Randøy and Nielsen (2002), Duffhues and Kabir (2007) and Attaway (2000) intended to investigate this relationship, but due to a lack of data on stocks, options and bonuses they could not make any general recommendations with regard to their potential usefulness. To gain insight into the different components of the compensation package is of substantial importance, because a well-designed compensation package can contribute to an improvement of the CEOs incentives to act in line with the interests of the shareholders’.
The objective of this study is to extent the empirical literature and to fill the gap for non-Anglo-American countries, by using unique panel data on individual CEO compensation of companies listed on the Dutch Stock Exchange from 2003 until 2009. In so doing, the paper contributes to one of the most important recent public-policy debates and corporate governance reforms in the Netherlands.
2. Literature Review
2.1 Theoretical background and hypotheses
CEO compensation research began in the 1980s together with the emergence and general acceptance of the agency theory founded by Jensen and Meckling (1976). Much earlier, Berle and Means (1932) were the first to identify the consequences of a separation of ownership and control. They suggested that the separation of ownership and control may introduce a principal-agent problem due to asymmetric information between executives and shareholders. In particular, the CEO (agent) has more or better information regarding the company than the shareholder (principal). Jensen and Meckling (1976) were the first to formalize the agency problem. They argue that the agency problem causes managers to pursue their own interests instead of the interests of the shareholders. For instance risk-neutral shareholders are assumed to be interested in a return on their investment, whereas the risk-averse CEO may also value other benefits such as growing a large company, or using company assets to satisfy her private needs. After all, with a separation of ownership and control, the marginal benefit to the CEO of her work effort does not reflect the marginal contribution of such work effort to corporate performance. Consequently, a CEO may decide to shirk her duties by delivering effort which is considered sub-optimal by the shareholders. Therefore shareholders have to make sure the CEOs will be deterred from behaving opportunistically and ineffectively, which may result in a substantial decrease in shareholder value (Ferrani and Molloney, 2010). For instance, CEOs can put insufficient effort into their work, engage in extravagant investments, take benefits through self-dealing and use entrenchment strategies that hurt shareholders in order to keep their own position (Tirole, 2006).
compensation on (future) firm performance and the behavior (incentives) of the CEO (i.e., ex ante compensation). The former determines the pay-performance relationship based upon actual or realized performance, while the latter captures the incentives of the CEO whether he or she will act in the interests of the shareholders (van der Laan, 2010) This study focuses both on ex post compensation and ex ante compensation, seeking to determine whether CEO compensation is either determined by realized performance and that CEO compensation determines (future) firm performance.
2.1.1 CEO compensation as a function of firm performance
Despite extensive research about the relationship whether CEO compensation depends on firm performance (ex-post compensation), evidence is still not persuasive. Bebchuk and Fried (2004) state for example that executives’ large compensation packages have been much less sensitive to firm performance than has been commonly recognized. If there is no meaningful link between CEO compensation and company performance, it is doubtful that the assets of listed companies are being managed efficiently which eventually will hurt shareholders’ value (Hall and Liebman, 1998). Therefore, in the first hypotheses I will test whether the prediction of the agency theory holds that total CEO compensation depends on firm performance.
Hypothesis 1: Total compensation depends positively on firm performance
different variable forms of compensation on firm performance. The short-term and long-term variable components are meant to induce CEOs to internalize the shareholders’ interest (Tirole, 2006).
Salary measures the component of compensation that is fixed at the beginning of the year (Core et al., 1999). Most studies which examine the pay-for-performance relationship do not expect any sensitivity in the compensation due to the fixed pay. Only indirect effects of performance on fixed pay may emerge, as evidenced by the amount of variable compensation increases the risk a CEO faces. This translates in a risk premium that may become apparent as an increase in fixed compensation (van der Laan et al, 2010). Therefore, I do not expect to find a direct relationship between fixed compensation and firm performance.
Hypothesis 2: Fixed compensation does not depend positively or negatively on firm performance
In general, short-term variable compensation (bonus) is distinguished from long-term variable compensation (stocks and options). The bonuses are usually determined at the end of year t , are based on prior performance criteria which are set at the beginning of year t like current profit (accounting data), and are used to motivate the CEOs which create a strong incentive for them to prefer the short-term over the long-term. CEOs who solely focus on short-term accounting profits may therefore avoid actions that reduce current profitability but may increase future profitability, such as cutting R&D expenditures (Dechow and Sloan, 1991). Accounting-based firm performance measures (e.g. ROA) are essentially backward looking. In contrast, market-based firm performance measures like stock return also reflect future performance. Since the effect of accounting-based and market-based firm performance measures is expected to be different, I expect the bonus to be particularly affected by accounting-based measures of firm performance.
Hypothesis 3: Short-term variable compensation (bonus) depends positively on firm performance
supported the benefits of equity-based compensation (Core et al, 2003). Also a widely cited theoretical paper by Jensen and Murphy (1990) states that most of the pay-for-performance sensitivity comes from equity-based compensation. Hall and Liebman (1998) criticized the Jensen and Murphy approach, but through a slightly adapted methodology they also argue that virtually all of the pay to performance sensitivity is attributable to changes in the value of CEO stock and stock options holdings. Therefore, in the third hypotheses I will test whether long-term variable compensation (stocks and options) depends positively on firm performance.
Hypothesis 4: Long-term variable compensation (stocks and options) depends positively on firm performance
2.1.2 Firm performance as a function of CEO compensation
In the second part of this thesis I will emphasize on the relationship whether (future) firm performance is determined by CEO compensation (i.e. ex-ante compensation).Jensen and Meckling (1976) argue that the agency problem causes managers to pursue their own interests instead of the interests of the shareholders’. A possible way to mitigate this agency problem is to align CEO compensation to future performance in order to make them behave according to the principals’ objectives. As stated above, according to Tirole (2006), Bebchuck and Fried (2004), long-term variable components of the compensation package are meant to align these interests between the CEOs and the shareholders.
performance is poor. Another drawback of stocks and options is that companies have given executives broad freedom to unwind equity incentives, a practice that has been beneficial to executives but unfavorable for shareholders since such unwinding either weakens executives’ incentives or forces the firm to give executives new equity incentives to restore the CEO’s incentives to the pre-undwinding level (Bebchuk and Fried, 2003). Nevertheless, this will not be taken into account in the remaining part of this thesis. Furthermore, stock options which are out-of-the-money may encourage CEOs to take significant risks in order to increase the value of their options, since the exercise price is above the market price of the options. Thus, the CEO has to do particularly well in order to gain reward compared to compensation by at-the-money options (Bebchuk and Fried, 2003). Finally, the amount of stocks and options increases the risk a CEO faces, since it has been argued that stock options and share plans have different behavioral consequences as there is no downside risk attached to stock option plans (van der Laan et al, 2010). For example, when the options of a CEO are out-of-the-money the CEO will take more risk, since they do not care much about the losses (their options are worthless anyway); all the CEO is interested in are the potential gains.
Since eminent studies like Hall and Liebman (1998) and Jensen and Murphy (1990) argue that most of the pay-for-performance sensitivity comes from equity-based compensation, I hypothesize there exists positive relationship between long-term variable compensation (stocks and options) and future firm performance.
Hypothesis 5: Future firm performance depends positively or negatively on long-term variable compensation (stocks and options)
compensation should be structured such that he or she is motivated to maximize shareholder interests. Widely cited surveys of this work include Murphy (1999) and Core et al (1999).
Bebchuk and Fried (2003) also address a different link between the agency problem and CEO compensation. Under this approach, which they label the ‘managerial power approach’, executive compensation is viewed not only as a potential remedy for the agency problem but also as part of the agency problem itself. They argue that compensation is not related to performance since CEOs are likely to have an important influence on the design of compensation arrangements. Moreover, CEOs’ influence over their own pay might impose substantial costs on shareholders – beyond the excess pay executives receive – by diluting and distorting managers’ incentives and thereby hurting corporate performance. I follow the arguments of Bebchuk and Fried (2003) on the managerial power approach that CEOs have influence over their own compensation but this will not be further examined within the rest of this thesis. Meanwhile, in the sixth hypothesis I will examine whether there exists an optimal ratio between fixed and variable compensation with respect to several firm performance measures in order to mitigate the agency problem according to the ‘optimal contracting approach’.
Hypotheses 6: Firm performance is a parabolic (first increasing, then decreasing) function of the ratio of variable compensation to fixed compensation
2.2 Empirical findings
Several studies even report a negative pay-performance relationship or examine to relationship at all. For example, Miller (1995) and Firth et al (1995) fail to identify a positive pay-performance relationship and the only Dutch study performed by Duffhues and Kabir (2007) gives no systematic evidence that CEO compensation of Dutch firms is positively related to firm performance as well. Of particular interest are the results from a large number of regressions that even show a statistically significant negative pay-performance relationship. Moreover Randøy and Nielsen (2002) fail to find a significant relationship between both company ROA and stock performance and CEO compensation in either Norway or Sweden. Bebchuk and Fried (2004) state that empirical studies have failed to find any significant relationship between non-equity compensation like salary and pension payments and company performance during the 1990s.
According to Hall and Liebman (1998) the sensitivity of top executives pay to shareholder returns has increased tenfold between the 1980s and the early 1990s. Contrary, Bebchuk and Fried (2004) argue tat CEOs compensation packages have been much less sensitive to their performance than has been commonly recognized and in a theoretical paper Jensen and Murphy (1990) also argue the pay-performance relationship has fallen by an order of magnitude in the last five decades. They argue that companies could have generated the same increase in incentives at a much lower cost to their shareholders, or they could have used the amount spent to obtain more powerful incentives.
3. Data
3.1 Data collection
The dataset for this study consists of all companies listed on the Amsterdam Stock Exchange with at least three year records between 2003 and 2009. The reason for choosing this particular time interval is due to the lack of available data on CEO compensation prior to 2003, since before Dutch firms were not legally obliged to disclose CEO remuneration in their annual reports1. The Dutch stock market is characterized by relatively few listings and a large variety in firm size and industries; I therefore aimed to include all listed firms in the sample. The initial sample of 141 firms has been reduced after removing banking and insurance companies and firms without any available data on CEO compensation. Several companies were excluded because their primary listing was outside the Netherlands and thus CEOs of these companies would probably not conform to Dutch pay practices, since there was no material business activity in The Netherlands. Eventually the dataset
contains 95 firms listed on the Amsterdam Stock Exchange. However, the results should be interpreted with a sample bias in mind, since in total 56 companies have been removed and there is no data available for non-listed companies. The primary sources for the dataset are Thomson Reuters’ DataStream and Worldscope. Data for most of the dependent variables and control variables are derived from these databases. Furthermore a website of the Dutch Investor Association (Vereniging van Effectenbezitters)2 has been used to obtain data on the variables concerning CEO compensation. This website does not only contain the total amount of compensation a CEO receives during a fiscal year, but also data on the different components of the compensation package.3 However, as mentioned before in section 2, I make a distinction between fixed compensation (salary), short-term variable compensation (bonus) and long-term variable compensation (stocks and options). Finally, I also examined annual reports in order to complete the dataset and whenever annual reports were not found, I contacted the company to request for the reports.
The variables used to specify the model of the empirical analyses include several measures of CEO compensation, three firm performance measures and a set of control variables, which will be defined below. I refer to table A of Appendix A for the definition and source(s) of all independent, dependent and control variables.
3.2 CEO compensation variables
I follow the approach used by Randøy and Nielsen (2002), Kato and Kubo (2003) and Duffhues and Kabir (2007) to take the natural logarithm of the compensation variables, in order to reduce the heteroscedasticity. Thus, the total compensation consists of the log of total annual payment to the CEO, the fixed compensation equals the log of the of the salary payments, the short-term variable compensation equals log of the bonus payments and finally the long-term variable compensation is defined as the natural logarithm of stocks and options. Severance pay and pension payments have been removed from the dataset since the compensation and benefits a CEO receives at the time he or she leaves the company are not of particular interest for this study, as the agency problem assumes the CEO is employed at the company. It occasionally occurred that two or more CEOs received compensation during a particular year. To properly account for appointments and leaves
2 www.bestuursvoorzitter.nl
3 The compensation package has been divided into the following components; salary, bonus, stocks, options, pension
for CEOs during a year, the number of months the CEO worked in the position has been recorded, and the salary is annualized accordingly. The Black and Scholes options pricing formula is used to derive the value of the options and the stocks and options are valuated at the date of issuance. For additional assumptions regarding the stocks and options pricing I refer the descriptive statistics and to table A of Appendix A.
Table 1
Descriptive statistics of CEO compensation
Year No. of obs. CEO compensation (in € 1000)
Mean Median Std. Dev. Min Max Panel A: Fixed CEO Compensation (Salary)
2003 84 407 300 345 75 1848 2004 86 404 316 312 63 1500 2005 89 434 330 366 49 1928 2006 92 443 351 329 83 1625 2007 93 452 384 329 94 1775 2008 93 457 381 314 104 1925 2009 92 454 398 266 120 1500 Pooled 629 437 355 323 49 1928 % Change 2003-2009 11% 28%
Panel B: Short-term variable CEO Compensation (Bonus)
2003 65 201 94 322 9 1792 2004 65 320 129 566 4 3677 2005 71 292 154 373 6 1938 2006 81 334 167 428 7 2040 2007 81 458 207 709 14 3600 2008 76 374 150 596 10 3750 2009 72 353 193 493 10 3029 Pooled 511 338 150 521 4 3750 % Change 2003-2009 72% 43%
Panel C: Long-term variable CEO Compensation (Stocks & Options)
2003 39 428 66 972 1 4467 2004 42 635 117 1450 2 8165 2005 43 594 223 796 8 3720 2006 40 829 404 1035 8 4291 2007 46 860 388 1217 7 5620 2008 48 732 418 1026 10 4986 2009 45 621 254 903 10 4724 Pooled 303 676 274 1073 1 8165 % Change 2003-2009 37% 135%
Panel D: Total CEO Compensation (Salary, Bonus, Stocks & Options)
2003 84 762 384 1157 75 7967 2004 86 956 467 1599 63 10985 2005 89 954 482 1228 49 7182 2006 92 1098 569 1389 103 7627 2007 93 1277 594 1744 94 8996 2008 94 1128 542 1551 100 10661 2009 92 1034 593 1222 145 8088 Pooled 630 1035 518 1432 49 10985 % Change 2003-2009 30.63% 43.43%
3.3 Firm performance variables
The variables which proxy firm performance are measured in a number of ways for which both accounting-based and market-based performance measures are considered. The choice of performance measure is not conceptually unimportant. Empirically, there is some guidance as to which measures are used frequently. I follow the approach by Core et al (1999) and use return on assets (ROA), stock return and the Tobin’s Q as firm performance measures.
ROA has been widely used in corporate governance studies (Brick et al, 2006). The ROA as a firm performance measure is accounting-based and is defined as net income divided by average total assets. Devers (2007) argues that accounting-based measures of firm performance are backward-looking. The second performance measure equals stock return which is market-based and is defined as the natural logarithm of the total return index of current year divided by the natural logarithm of the total return index of last year. The third and final measure for firm performance is Tobins’ Q, which is a hybrid of an accounting-based and market-based measure. The numerator of Tobin’s Q, being the market value of assets, partly reflects the value investors assign to a firm’s assets. Yet the denominator, the book value of the company’s assets, does not include investments the firm has made in assets. The company’s future revenue stream is treated as if it can be generated from investments made only in tangible capital (Demetz and Villalonga, 2001). Summarized, the Tobin’s Q can be defined as the ratio of the sum of market value of assets to the book value of total assets and indicates a firm’s growth opportunity and is based on investors' evaluation of its future profitability, so it is in fact forward-looking (Demsetz and Villalonga, 2001). A high Tobin's Q indicates success in the sense that a firm has deployed investments and the firm is valued higher by the market value than its book value.
According to Murphy (1999) performance measurement is quite imperfect, since accounting profits are for most companies the primary determinant of executive compensation. It is important to note two major drawbacks of accounting measures as a performance variable. First, accounting profits focus on the short-term. As mentioned before in section 2.1.1, executives who focus only on accounting profits may thereby avoid actions that reduce current profitability but may increase future profitability, such as cutting R&D expenditures (Dechow and Sloan, 1991). Second, accounting profits could be manipulated, for example by shifting earnings across periods or by accelerating or delaying revenues and costs (Healy, 1985).
management and also align the interests between executives and shareholders effectively (Murphy, 1999).
Descriptive statistics of firm performance variables are presented in Table 2. Panel A provides statistics about firm performance measures and Panel B on firm characteristics. These statistics are based on 634 observations and more. ROA of an average firm equals 2.7 % with a median value of 4.8 %. The market-based measure, stock return equals – 0.025 % (median = 0.074). This higher value of both medians reflects the fact that the sample consists of a few companies with a relatively low ROA and stock return. The hybrid performance measure Tobin’s Q is positive over the sample period with a mean (median) value of 1.257 (1.009).
Mean Median St. dev. Min Max No. of obs.
Return on Assets 0.027 0.048 0.140 -0.758 0.516 654
Stock Return -0.025 0.074 0.513 -2.664 1.366 634
Tobin's Q 1.257 1.009 0.875 0.200 6.711 637
Panel B: firm characteristics
Size (mln €) 4.850 0.484 19.859 0.001 201 661
Employees (number) 20047 2359 5617 1 583830 650
Age (years) 60 36 64 5 358 665
Leverage (%) 0.24 0.23 0.17 0.00 0.97 657
Table 2
Panel A: firm performance measures
Descriptive statistics of firm performance measures and firm characteristics
3.4 Control variables
Following related research (e.g. Duffhues and Kabir, 2007; Core et al, 1999; Brick et al, 2006) several control variables are included to account for changes in other firm characteristics and industry-specific factors that are potentially related to the pay-performance relationship. I use the following control variables; firm size, firm age, leverage, industry effects and time effects.
(total assets = € 4.850 million and employees = 20.047). Furthermore, the age of an average firm in the sample is 60 years and the average leverage of sample companies is 24%.
As Tosi et al (2000) demonstrate, the biggest part of compensation is explained by firm size. Also Murphy (1999) and Baker and Hall (2004) argue that pay-performance sensitivities vary with respect to the size of the company. It is not surprising that larger firms, for example, may employ better-qualified and better-paid managers (Oxelheim and Randoy, 2006). In addition, Core et al (2003) and Kato and Kubo (2003) argue that larger firms with greater growth opportunities and more complex operations will demand higher-quality managers with higher equilibrium wages. Since firm size is an important determinant of firm performance it is essential to include this control variable, measured by the total assets of each firm at year t. As a second proxy of firm size the natural logarithm the number of employees is used.
As regards the age of the firm, I follow Brown and Medoff (2003) and measure the age by the number of years from the company’s foundation until 2009. Firm age is occasionally included as a control variable in this type of research (Randoy and Nielsen, 2002); however I follow Brown and Medoff (2003) with their expectations that older firms pay a higher level of compensation.
Leverage reflects the capital structure of a company and is defined as the book value of debt to the book value of total assets. An explanation according to the agency theory states that a high amount of debt is associated with an increasing monitoring of CEOs by the suppliers of the debt and thereby reducing the payment of excess compensation. Furthermore, Jensen (1989) argues that due to the disciplining function of debt, companies with a high debt ratio (and thus less free cash flow) are less able to pay a high level of CEO compensation. In contrast, higher leverage increases firm risk which in turn necessitates the payment of a higher amount of compensation (Duffhues and Kabir, 2007). I follow the arguments of Jensen (1989) and assume that debt has a disciplining function and thus reduces the level of compensation of the CEO.
These eight industries comprise 98.9% of the original sample. For an overview of which company belongs to a particular industry I refer to table B of Appendix B. Finally, to capture time effects I have created dummy variables which receive a one for a company belonging to a specific year and a zero otherwise.
Table C of Appendix C represents the information on the distribution of sample firms across different industries according to the Industry Classification Benchmark (ICB). I observe that largest part (32.6%) of the sample come from the industrials sector (204 firm-year observations). The technology sector represents 15.8% (102 firm-year observations); the consumer goods and consumer services sectors represent each 13.7% (89 and 91 firm-year observations respectively).
3.5 Correlation
The correlation matrix in table 3 suggests that total assets is substantially positive correlated with total compensation and salary. Also the other proxy of firm size, the number of employees, shows a high correlation with four compensation variables (1-4). This is no surprise since firm size is assumed to explain the biggest part of CEO compensation (Murphy, 1999; Tosi et al, 2000). Furthermore the results of the correlation matrix do not indicate any significant correlations between the firm performance variables and control variables. The largest observed correlation between these variables is 0.45 between age and the amount of employees and this is sufficiently small (< 0.50) that the possible multicollinearity can reasonably be ignored.
Correlation Matrix 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. Compensation Variables 1. Total Compensation 1 2. Salary 0.93 1 3. Bonus 0.88 0.79 1
4. Stocks and Options 0.92 0.78 0.75 1
5. (Stocks & Options/Salary) 0.71 0.47 0.49 0.78 1 6. (Stocks & Options/Salary)^2 0.53 0.33 0.34 0.55 0.90 1
Firm Performance 7. ROA 0.26 0.17 0.32 0.30 0.21 0.13 1 8. Stock Return 0.05 -0.04 0.13 0.11 0.06 0.03 0.26 1 9. Tobin's Q 0.14 0.05 0.13 0.16 0.21 0.18 0.22 0.17 1 Control Variables 10. Total Assets 0.50 0.51 0.44 0.41 0.31 0.22 0.07 -0.02 -0.16 1 11. Employees 0.74 0.80 0.67 0.62 0.32 0.19 0.22 0.01 -0.20 0.42 1 12. Age 0.28 0.30 0.25 0.23 0.17 0.17 0.01 0.06 -0.30 0.17 0.38 1 13. Leverage 0.09 0.15 0.07 0.00 -0.03 0.00 -0.42 -0.20 -0.30 -0.09 0.15 0.16 1 Table 3
Note: The table shows the correlations between the dependent, independent and the control variables. Correlations above 0.50 are marked in italics. The natural logarithm has been taken of all compensation variables and the number of employees.
4. Methodology
4.1 CEO compensation as a function of firm performance
I start from the following specification in order to test whether the various measures of firm performance have a significant impact on CEO compensation:
∑
CEO compensation it = α + β1 ROA it + β2 Stock Return it + β3 Tobin’s Qit + β4 Total Assets it + β5 Employees it + β6 Age i + β7 Leverage it + δ Time dummies t + λ Industry dummies i + ε it , (1)avoid the dummy variable trap, one time dummy and one industry dummy will be removed from the equation (Brooks, 2008). The first four hypotheses are captured by equation 1. I also use lagged performance measures to account for the hypothesis that CEO compensation paid in year t is usually determined by previous year’s firm performance (Duffhues and Kabir, 2007). According to Brick et al (2006), when the performance measures in the regressions are lagged with one year also a potential endogeneity problem will be reduced. I am aware of a potential endogeneity problem when measuring the contemporaneous relationships, since the independent variables are possibly correlated with the error term, which can lead to biased results. However, this way of testing is often done in the field of corporate governance literature (e.g. Randøy and Nielsen, 2002; Kato and Kubo, 2003; Duffhues and Kabir, 2007)
4.2 Firm performance as a function of CEO compensation
The second part of this paper examines the relationship between ex-ante compensation and future firm performance in order to test whether the incentives of the CEO are aligned with those of the shareholders. I estimate the following equation (2) to determine the relationship between long-term variable compensation (stocks and options) and future firm performance.
∑
Firm Performance it = α +∑
= 20 k
β1,k Stocks & Options i,t-k + β2 Total Assets it + β3 Employees it +
β4 Age i + β5 Leverage it + β6 Stocks & Options*Employees it + δ Time dummies t + λ Industry
dummies i + ε it, (2)
where long-term variable compensation is the amount of stocks and options a CEO of firm t received for year t – k. Thus the coefficient β1,k estimates the contribution of stocks and options in year t – k to subsequent firm performance measures. The coefficient β6 captures an interaction effect. In this way I can also examine whether the relationship between stocks & options and firm performance differs according to the size of a company.
it e Performanc Firm
∑
= α+β1 it on Compensati Fixed on Compensati Variable + β2 2 it on Compensati Fixed on Compensati Variable + it Variables Controlβ + δ Timedummiest+ λ Industrydummiesi+ εit, (3)
where the control variables are total assets, employees, age and leverage. According to the sixth hypothesis I expect the sign of the coefficient of the ratio β1 to be positive and significant and the coefficient of the quadratic term β2 to be positive and significant. Thus β2 captures the second-order payoff to the same compensation ratio. Note that I further label the coefficient of the linear term β1 as the ‘first-order’ effect and the coefficient on the non-linear squared term β2 as the ‘second-order’ effect.
4.3 Estimation Strategy
I will first apply a pooled Ordinary Least Squares (OLS) method in order to test the hypotheses (1)-(4) corresponding to equation (1). Besides measuring the contemporaneous relationship between CEO compensation and firm performance, the independent variables which proxy for firm performance will lagged one year in order to account for the hypothesis that CEO compensation paid in year t is usually determined by previous year’s firm performance (Duffhues and Kabir, 2007). Also a potential endogeneity problem will be reduced when the performance measures in the regressions are lagged one year (Brick et al, 2006).
5. Results
5.1 CEO compensation as a function of firm performance
Table 4 presents the pooled OLS estimates of equation (1) to gain insight into the first four hypotheses, using total compensation, fixed compensation and both short-term and long-term variable compensation for the full sample of firms. The regression specifications are similar except that I replace one firm performance measure with another for each hypothesis. The number of firm-year observation varies between 609 and 298. For the sake of brevity I do not report regression estimates for the dummy variables that capture time effects and industry effects.
The results of table 4 indicate that total compensation and short-term variable compensation (bonus) are significantly positive related to ROA. It follows that a 1 percentage point increase in ROA (say 0.03 to 0.04) will lead to a 0.35% and a 1.91% increase in total compensation and short-term variable (bonus) respectively. Since the average CEO earns a total compensation of € 1.04 million and a bonus of € 0.34 million, improving ROA by 1 percentage point on average results in an increase in total compensation of about € 3600 and an increase in the bonus of about € 6500. These relationships were expected to be positive and significant. Moreover, fixed salary is negatively related to ROA (p < 0.10), which is in contrast with the hypothesis of no relationship between fixed salary and firm performance. It follows that a 1 percentage point increase in ROA (say 0.03 to 0.04) will lead to a 0.134 % decrease in fixed salary. Since the average CEO earns a fixed salary of € 0.44 million, improving the stock return by 1 percentage point on average results in a decrease in fixed salary of about € 600 which is quite remarkable. Thus, there is evidence that there are direct effects of firm performance on fixed salary.
Furthermore, all compensation measures are positively related to the Tobin’s Q (p < 0.01) and for example a 1 percentage point increase in the Tobin’s Q results on average in an increase in long-term variable compensation of about € 2680. The number of observations for stocks and options is limited (300 firm-year observations), therefore a sample bias has to be kept in mind when interpreting the results.
associated with a percentage increase equal to the height of the coefficient. For example, a 1% increase in the number of employees is associated with on average an increase of 0.316% of total compensation, which equals on average € 3270. These findings are in line with findings of Tosi et al (2000) and Murphy (1999), arguing that larger firms pay higher CEO compensation. Subsequently, firm age is small but significantly positive related to total compensation (p < 0.10) and to bonus (p < 0.01), indicating that older firms pay higher CEO compensation (Brown and Medoff, 2003). Finally, leverage is positively related to bonus payments. This is in line with the arguments of Duffhues and Kabir (2007) indicating that an increase in leverage entails high risk and thus CEOs should be rewarded for that risk, but contradicts the arguments of a disciplining function of debt (Jensen, 1989) or an increased monitoring of CEOs by the suppliers of debt and thereby reducing the payment of excess compensation.
The results of model (1) – (4) show that all estimations reach statistical significance at the 1% level. The high value of explanatory power ( 2
Table 4
Effects of firm performance on four CEO compensation variables
Total Compensation Fixed Short-term variable Long-term variable Salary Bonus Stocks and options
(1) (2) (3) (4)
Variable Coefficient Coefficient Coefficient Coefficient (t-statistic) (t-statistic) (t-statistic) (t-statistic)
ROA 0.352 -0.134 1.913 0.557 (2.09)** (-1.79)* (6.85)*** (0.88) Stock Return -0.064 -0.072 0.024 0.076 (-0.85) (-1.44) (0.16) (0.39) Tobin's Q 0.233 0.089 0.298 0.396 (9.00)*** (8.59)*** (6.40)*** (4.23)*** Control Variables Total Assets 0.006 0.004 0.008 0.002 (10.07)*** (9.87)*** (7.64)*** (1.63)* Employees 0.316 0.222 0.291 0.509 (17.57)*** (22.69)*** (12.53)*** (16.35)*** Age 0.000 -0.000 0.002 0.002 (1.75)* (-0.04) (2.86)*** (4.19)*** Leverage 0.074 -0.023 0.942 -0.338 (0.80) (-0.41) (5.04)*** (-1.24) Constant 11.190 11.370 9.656 7.942 (59.50)*** (169.31)*** (33.62)*** (22.576)*** Adj R-squared 0.74 0.74 0.57 0.64 F-statistic 83.17*** 84.18*** 32.10*** 25.82*** N 609 608 497 298
Table 5
Effects of firm performance on four CEO compensation variables for both a contemporaneous relationship and a lagged relationship
Total Fixed Short-term variable Long-term variable Total Fixed Short-term variable Long-term variable
Salary Bonus Stocks & Options Salary Bonus Stocks & Options
(1) (2) (3) (4) (5) (6) (7) (8)
Variable Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
(t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic)
ROA 0.181 -0.020 1.914 0.918 -0.185 0.040 0.717 -1.915 (0.79) (-0.33) (3.29)*** (2.12)** (-1.04) (0.27) (1.27) (-2.34)** Stock Return 0.040 0.000 0.141 -0.030 0.004 -0.047 -0.085 0.040 (0.87) (0.00) (1.18) (-0.11) (0.10) (-2.60)* (-1.01) (0.16) Tobin's Q 0.068 -0.034 0.130 0.115 -0.000 -0.009 0.061 0.009 (3.00)** (-2.31)** (2.00)** (0.79) (-0.00) (-0.79) (1.02) (0.11) Control Variables Total Assets 0.005 0.004 0.026 -0.004 0.003 0.003 0.011 -0.005 (2.01)** (2.56)** (2.25)** (-1.18) (1.02) (1.67)* (2.84)*** (-1.00) Employees 0.168 0.080 -0.036 -0.111 0.106 0.081 -0.030 -0.151 (3.77)*** (3.63)*** (-0.35) (-0.41) (2.58)** (2.94)*** (-0.30) (-0.62) Age Leverage -0.052 -0.040 0.787 0.600 0.033 0.170 -0.042 0.261 (-0.57) (-0.44) (2.18)* (1.47) (0.25) (3.97)*** (-0.17) (0.56) Constant 11.690 12.091 11.270 12.224 12.325 12.005 11.806 13.420 (33.83)*** (83.08)*** (12.26)*** (5.44)*** (37.47)*** (60.01)*** (14.53)*** (6.63)*** Adj. R-squared 0.93 0.93 0.82 0.86 0.94 0.94 0.84 0.88 F-statistic 67.10*** 66.66*** 19.69*** 18.47*** 63.77*** 62.69*** 18.47*** 19.01*** N 609 608 497 298 519 518 427 254
Contemporaneous relationship Lagged relationship
When comparing the pooled regression estimates with the cross-sectional fixed effects estimates the positive relationship between CEO compensation and firm performance still exists. When examining the contemporaneous relationship in table 5, ROA has a significant positive impact on bonus (p < 0.01) and on stocks & options (p < 0.05). These coefficients are relatively high, which implies that 1 percentage point increase in ROA is associated with an increase in bonus payment of 1.91%, which on average equals about € 6470 and an increase in stocks & options of 0.918% which on average equals about € 6200 for the whole sample. Though, stock return does not show significant estimates regarding a positive impact on the four CEO compensation measures for model (1) – (4), which is in contrast with the hypotheses 1, 3 and 4 that respectively total compensation, short-term variable compensation (bonus) and long-term variable compensation (stocks and options) depend positively on firm performance. The results are also in line with hypothesis 2 that fixed salary does not depend on firm performance.
The coefficients of the relationship between Tobin’s Q and CEO compensation show significant results for total compensation, fixed salary and short-term variable compensation (bonus). However, compared to the pooled estimates it seems that Tobin’s Q is negatively related to fixed salary which is quite remarkable. The results of model (1) – (8) show that all estimated coefficients reach statistical significance at the 1% level.
5.1.1 Robustness check
I also performed a robustness test in order to check whether the results are not driven by something else. For this test I have created three sub-samples according to the size (total assets) of the company, which are labeled ‘low’, ‘medium’ and ‘high’. In general the sub-sample regressions show similar results, however in the sub-sample ‘high’ there seems to be a positive relationship between ROA and total compensation and for both the sub-sample ‘medium’ and ‘high’ there seems to be a positive relationship between stock return and salary which were not shown in the results for the whole sample. This indicates that the results are partly driven by the size of the company, which is in line with arguments of Baker and Hall (2004) that pay-performance sensitivities vary with respect to the size of the company. For further details I refer to table E of Appendix F.
5.1.2 Lagged Relationship
in contrast with hypothesis 3 which assumes a positive relationship. Furthermore results show that fixed salary depends negatively on previous years’ stock return (p < 0.10), with implies that a 1 percentage point decrease in the previous year’s stock return results in a decrease in fixed salary of about € 200. As formulated in hypothesis 2, I do not expect to find any firm performance measure to be a significant determinant of fixed salary compensation; however this result contradicts the hypothesis. Furthermore there are some results indicating a small but significant positive effect of firm size on CEO compensation. Also the control variable leverage has a significant positive impact on bonus payments and on salary for the lagged relationship, which is in favor of the arguments of Duffhues and Kabir (2007) that higher leverage increases firm risk which in turn necessitates the payment of a higher amount of compensation.
A possible explanation for the different signs with respect to each firm performance variable is the way in which these are measured. As stated in section 3, ROA is an accounting-based measure reflecting firm performance over the prior fiscal year, whereas stock return in a market-based performance measure reflecting both the current performance and the expectation of future performance. Tobin’s Q which is a hybrid of an accounting-based and market-based measure indicating a firm’s growth opportunity and is based on investors' evaluation of its future profitability. The high value of explanatory power ( 2
R ) and F- statistic are also indicative of the reliability of the regression models. However, the R2 is artificially high in all regressions for cross-sectional fixed effects estimations, because this model estimates a constant for every company that is included in the regression. The model iterates until the regression has the best fit, therefore the R2should be interpreted with caution.
5.1.3 Robustness check
5.2 (Future) Firm performance as a function of CEO compensation
The results for hypothesis 5 whether future firm performance depends positively or negatively on long-term variable compensation (stocks and options) are expressed in table 6. The results of this table do not fully support the hypothesis. I only find significant relationships for model (3), (6) and (8). The coefficients of the variable stocks & options should be interpreted as the relationship between stocks & options and the firm performance measures conditional on zero employees, since an interaction term of stocks & options * employees is included. Thus the results indicate that stocks and options is a small but significant negative determinant for the ROA and stock return of year t + 2 given zero employees. An increase in 1% of the compensation in stocks and options is associated with a decrease in ROA of on average only 0.012% for year t + 2 and a decrease in stock return of only 0.057% for year t + 2. These results support the arguments of Jensen and Meckling (1976) that the agency problem causes managers to pursue their own interests instead of the interests of the shareholders, but indicate that (long-term variable) CEO compensation and future firm performance are not aligned in order to mitigate this agency problem. However, there is also evidence that CEO compensation and future firm performance are aligned as stocks & options seem to be a positive determinant of the Tobin’s Q for year t +1. Due to the inclusion of lags, the sample size has been reduced which will contribute to a sample bias.
Besides, the coefficients for leverage are negative for fixed salary and long-term variable compensation which do not support the core prediction of the agency cost hypothesis of Jensen and Mecling (1976) that a higher leverage is associated with improved firm performance. The coefficients of the interaction term stocks & options*employees, show that the relationship between stocks & options and firm performance slightly differs positively according to the number of employees only for model (2).
5.2.1 The optimal contract
between the ratio and firm performance and when the ratio is high there is a negative relationship associated with firm performance. In this way I can examine whether an optimal ratio exists.
Table 6
Effects of long-term variable compensation (stocks & options) on future firm performance
No lag 1 year lag 2 year lag No lag 1 year lag 2 year lag No lag 1 year lag 2 year lag
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Variable Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
(t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic)
Stocks & Options 0.032 -0.007 -0.012 -0.193 -0.010 -0.057 -0.03 0.072 0.091
(1.16) (-0.97) (-1.83)* (-1.45) (-0.37) (-2.36)* (-0.08) (1.97)* (1.18)
Stocks & Options * Employees -0.002 0.001 0.003 0.023 -0.001 0.006 0.010 0.010 0.004
(-0.94) (2.06)** (1.56) (-1.54) (-0.27) (1.61) (0.27) (1.58) (0.42) Total Assets -0.001 -0.001 -0.002 -0.006 0.006 0.016 -0.006 -0.004 0.001 (-3.13)*** (-0.97) (-2.45)** (-0.99)* (1.41) (2.71)* (-1.62) (-0.91) (0.17) Employees 0.062 0.024 -0.028 -0.443 -0.243 -0.045 -0.509 -0.561 -0.089 (2.16)** (1.08) (-0.60) (-1.94)** (-2.25)** (-0.60) (-0.82) (-1.87)* (-0.25) Leverage -0.188 -0.074 -0.209 -0.729 -1.253 -1.452 -0.853 -0.472 -1.936 (-1.35) (-0.90) (-2.15)** (-2.40)*** (-5.35)*** (-3.28)*** (-1.74)* (-0.54) (-2.03)** Constant -0.604 -0.216 0.208 3.567 2.882 0.663 5.119 4.678 1.266 (-2.04)** (-0.99) (0.83) (1.72)** (2.58)** (1.53) (0.89) (2.08)** (0.46) Adj. R-squared 0.67 0.73 0.69 0.63 0.66 0.73 0.78 0.79 0.87 F-statistic 9.44*** 10.41*** 8.05*** 8.21*** 7.85*** 9.49*** 15.16*** 14.00*** 21.99*** N 302 224 173 300 223 173 300 223 173
ROA Stock Return Tobin's Q
Table 7
Parabolic function to determine an optimal ratio between variable compensation an fixed compensation with respect to firm performance
ROA Stock Return Tobin's Q ROA Stock Return Tobin's Q
(1) (2) (3) (4) (5) (6)
Variable Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient (t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic)
(Stocks&Options/Salary) 0.028 0.055 0.509 0.020 0.038 0.228 (1.97)** (2.11)*** (6.46)*** (3.59)*** (0.63) (1.35) (Stocks&Options/Salary)^2 -0.006 -0.013 -0.05 -0.005 -0.010 -0.043 (-5.28)*** (-3.50)*** (-6.14)*** (-2.46)*** (-1.09) (-1.20) Total Assets -0.001 -0.001 -0.005 -0.000 -0.001 -0.006 (-4.55)*** (-1.69)* (-4.44)*** (-4.55)*** (-0.97) (-1.67)* Employees 0.013 -0.004 -0.048 0.029 -0.119 -0.350 (4.74)*** (-0.36) (-1.88)* (1.41) (-1.03) (-1.52) Age -0.000 -0.000 -0.004 (-10.19)*** (-1.07) (-21.79)*** Leverage -0.357 -0.555 -1.029 -0.190 -0.757 -0.861 (-4.80)*** (-3.00)*** (-1.74)* (-1.42) (-2.40)** (-1.70)* Constant 0.156 0.202 2.771 -0.201 0.912 4.470 (2.88)** (1.10) (6.22)*** (-1.19) (0.88) (2.26)** Adj. R-squared 0.38 0.61 0.38 0.67 0.63 0.78 F-statistic 10.16*** 24.49*** 10.20*** 9.38*** 8.02*** 15.38*** N 302 300 300 302 300 300
Pooled Cross-sectional fixed effects
6. Conclusion
In this study I investigate the relationship between CEO compensation and firm performance, by using unique panel data on individual CEO compensation of companies listed on the Dutch Stock Exchange from 2003 until 2009. The analysis is based on total compensation, fixed salary, short-term variable compensation (bonus) and long-short-term variable compensation (stocks & options). The first part of this thesis has been contributed to the examination whether CEO compensation is determined by firm performance (i.e. ex-post compensation), second the impact of CEO compensation on future firm performance (i.e. ex-ante compensation) and whether an optimal ratio between variable and fixed compensations exists have been examined.
A theoretical foundation has been provided by the agency theory (Dalton et al., 2008) and predicts that CEO compensation policies will depend on changes in shareholder wealth (Jensen and Murphy, 1990). Relevant for the second part of this thesis are the arguments of Jensen and Meckling (1976) that the agency problem causes managers to pursue their own interests instead of the interests of the shareholders’, therefore a possible way to mitigate this agency problem is to align compensation to future performance in order to make the CEO behave according to the principals’ objectives. A design of the compensation package can also contribute to the alignment between CEOs and shareholders.
6.1 CEO compensation as a function of firm performance
I find systematic evidence that ROA and the Tobin’s Q are significant determinants of total compensation which confirms the hypothesis that total compensation is determined by firm performance and these results are in line with a Japanese study by Kato and Kubo (2003), but in contrast with a Dutch study by Duffhues and Kabir (2007) who found a negative pay-performance relationship. A possible explanation could be that despite examining the same companies, due to the Code Tabaksblat and other governance reforms, pay is better aligned with performance than within the observation window of the study of Duffhues and Kabir (2007), this should however be examined in further research.
of a negative relationship between ROA and fixed salary and a positive relationship between the Tobin’s Q and fixed salary for the pooled effects estimations and a negative affect for the fixed effects estimations which is in contrast with existing literature.
Additionally I find evidence for the third hypothesis, since results show that ROA seems to be a positive determinant of the short-term variable (bonus) compensation, which is in line with findings of for instance Core et al (2003), Jensen and Murphy (1999) and Hall and Liebman (1998). This proves the prediction that bonuses are particularly affected by accounting-based firm performance measures like ROA, as these are essentially backward-looking. The results confirm arguments of Tirole (2006) that short-term variable compensation (bonus) is meant to induce CEOs to internalize the shareholders’ interests and with Hall and Liebman (1998) that most of the pay-performance relationship is attributable to equity-based compensation.
The assumption that CEO compensation paid in year t is determined by previous year’s firm performance (Duffhues and Kabir, 2007) only holds in this study for the relationship between ROA and long-term variable compensation (stocks & options) and between stock return and fixed salary. Contrary to my hypothesis, the results indicate a negative relationship.
In general, the results indicate that the agency theory proves to be a good theoretical foundation for the study of the relationship between CEO compensation and firm performance and that CEO compensation policies should depend on firm performance. The results are also robust for removed observations between the contemporaneous relationship and the lagged relationship for all firm performance measures except Tobin’s Q. As a second robustness check a sub-sample regression has been performed, which indicates that the results are partly driven by size (total assets), which is in line with the arguments of Murphy (1999) and Baker and Hall (2004) that pay-performance sensitivities vary with respect to the size of the company.
6.2 Firm performance as a function of CEO compensation
positive determinant of the Tobin’s Q for the next year. As Jensen and Meckling (1976) argue, a possible way to mitigate this agency problem is to align compensation to future performance in order to make the CEO behave according to the principals’ objectives. This study only supports this theory for Tobin’s Q as a firm performance measure. Thus, there is room for compensation committees of Dutch firms to improve their future Tobin’s Q ratio when granting long-term variable compensation (stocks and options) in order for the CEO to be incentivized according to the interests’ of the shareholders.
The hypothesized parabolic relationship between the ratio of variable compensation to fixed compensation and firm performance proves to be existent. Therefore, the results are in line with the optimal contracting approach (Bebchuk and Fried, 2003) Thus, when holding total CEO compensation fixed the design of the compensation package has an impact on firm performance and hence can mitigate the agency problem. However, the ratio explains only a relative small proportion of the firm performance. This optimal contract does not imply a ‘perfect’ contract, it only implies that the firm designs the best contract it can in order to avoid opportunism and malfeasance by the CEO given the contracting constraints it faces and thus the CEO is motivated to maximize shareholder interests (Conyon, 2006). Thus, firms can improve their performance when compensation committees hold the total compensation constant, but take the ratio between variable and fixed compensation into account. Besides, these recommendations hold for the whole sample of Dutch listed firms, not particularly for a single company. Further research can aim to make firm-specific recommendations.
6.3 Limitations and further research
A limitation of this study concerns the sample size, especially when stocks and options and sub-sample regressions are taken into account. Unfortunately it was not possible to increase the sub-sample size due to a lack of data on CEO compensation before 2003. Because before Dutch listed firms were not obliged to disclose remuneration into their annual reports. Furthermore, 56 firms have been removed from the sample. Consequently, the association between CEO compensation and firm performance will be biased due to a truncated sample.
This study did not control for a potential endogeneity problem. When endogeneity is present there is a correlation between an independent variable and the error term, which is assumed to be present in many corporate governance studies. Further research can aim to solve this potential endogeneity problem to include instrumental variables (IV) that have to be exogenous, and then perform to two-stage-least squares (2SLS) analyses.
Appendix A
Definition of variables
Variable Definition Data source(s)
Dependent Variables
Return on Assets (ROA) The ratio of net income divided by total assets in year t Worldscope: Net Income (WC01751), Total Assets (WC02999) Stock Return Ratio of the natural logarithm of the price at the end of year t to the end Datastream: Return Index (RI)
of year t-1 adjusted for dividends and splits
Tobin's Q Ratio of market capitalization plus the book value of debt divided by the Worldscope: Market Capitalization (WC08001); book value of total assets in year t Total Assets (WC02999), Total Debt (WC03255)
Independent Variables
Total Compensation Natural logarithm of the total annual compensation paid to the CEO in year t Bestuursvoorzitter.nl Fixed Salary Natural logarithm of the fixed salary paid to the CEO in year t Bestuursvoorzitter.nl Short-term Variable Compensation Natural logarithm of the bonus paid to the CEO in year t Bestuursvoorzitter.nl (Bonus)
Long-term Variable Compensation Natural logarithm of the stocks & options paid to the CEO in year t Bestuursvoorzitter.nl (Stocks & Options)
Control Variables
Total Assets (Size) Proxied by the total assets at time t in billions Worldscope: Total Assets (WC02999) Employees Natural logarithm of the number of employees working for the firm at time t Datastream: Employees (DWEN) Age The number of years from the foundation of the company Annual Reports
Leverage Exchange Ratio of book value of debt to total assets Worldscope: Total Assets, (WC02999), Total Debt (WC03255) Industry effect A dummy variable, which receives a one for a company belonging to Euronext.nl (ICB Classification)
a specific industry and a zero otherwise
Time effect A dummy variable, which receives a one for a company belonging to a specific year and a zero otherwise
Table A
Appendix B
Sample firms and industry
Firm Industry Firm Industry
Aalberts Industrials Kardan Financials
Accell Consumer Goods Kendrion Industrials
Acomo Consumer Goods KPN Telecommunications
Ajax Consumer Services Macintosh Ret Consumer Services Ahold Consumer Services Mediq Consumer Services Akzo Nobel Basic Materials Nedap Industrials Alenheri Consumer Goods Nedsense (Blue Fox) Technology
AMG Industrials Neways Industrials
AMT Healthcare Nieuwe Steen Invst Financials AND INT Consumer Services Nutreco Consumer Goods
Arcadis Industrials Oce Technology
Arcelor Mittal Basic Materials Octoplus Healthcare
ASMI Technology Oranjewoud Industrials
ASML Technology Ordina Technology
Ballast Nedam Industrials Pharming Healthcare
BAM Industrials Philips Consumer Goods
Batenburg Industrials Porceleyne Fles Consumer Goods BE Semiconductor Technology Qurius Technology Beter Bed Consumer Services Randstad Industrials Bever Holding Financials Royal Dutch Shell Oil and Gas Boskalis Industrials Reed Elsevier Consumer Services Brill Consumer Services Royal Reesink Industrials
Brunel Industrials Roodmicrotec Technology
Corio Financials Roto Smeets Industrials
Crown v Gelder Basic Materials SBM Offshore Oil and Gas Crucell Healthcare Simac Techniek Technology
CSM Consumer Goods Sligro Consumer Services
CTAC Technology Spyker Consumer Goods
Dico Internat Consumer Goods Stern Consumer Services DocData Consumer Goods Telegraaf Consumer Services
DPA Industrials Ten Cate Industrials
Draka Holdings Industrials TIE Technology
DSM Basic Materials TKH Industrials
Eurocommercial Financials TNT Industrials
Exact Technology Tom Tom Technology
Fornix Healthcare Unilever Consumer Goods
Fugro Oil and Gas Unit 4 Technology
Gamma Industrials USG People Industrials
Grontmij Industrials Value8 Industrials
Groothandelsgebouwen Financials VastNed Offices Financials Heijmans Industrials VastNed Retail Financials
Heineken Consumer Goods Vopak Industrials
Hes Industrials Wavin Industrials
HITT Industrials Wegener Consumer Services
Holland Colours Basic Materials Wereldhave Financials ICT Automatisering Technology Wessanen Consumer Goods Imtech Industrials Wolters Kluwer Consumer Services Innoconcepts Industrials
Appendix C
Table C
Industry distribution and CEO compensation (in € 1.000)
Industry Oil & gas Basic Industrials Consumer Healthcare Consumer Telecom Financials Technology
Materials Goods Services
ICB Code 0001 1000 2000 3000 4000 5000 6000 8000 9000 No. of firms 3 5 31 13 5 13 1 9 15 % of total 3.16% 5.26% 32.63% 13.68% 5.26% 13.68% 1.05% 9.47% 15.79% Total Comp Mean 3651 1580 741 1096 596 1658 3253 430 574 Median 2109 1366 553 529 428 571 2996 408 400 Max 10661 5020 3541 5853 1828 10985 5626 1751 3031 Min 673 94 120 63 269 171 2003 49 90 No of obs. 21 35 204 89 26 91 7 55 102 Salary Mean 862 618 380 461 292 557 983 338 325 Median 542 612 365 303 281 377 1005 336 283 Max 1925 1828 918 1928 455 1848 1006 667 735 Min 367 94 102 63 153 132 853 49 87 No of obs. 21 35 204 88 26 91 7 55 102 Bonus Mean 889 617 380 461 292 557 983 338 325 Median 361 379 365 303 281 377 1005 336 283 Max 3750 3677 918 1928 455 1848 1006 667 735 Min 136 9000 102 63 153 132 853 49 87 No of obs. 21 29 204 88 26 91 7 55 102
Stocks & Options
Mean 1900 829 339 582 224 1401 1163 783 285
Median 1055 823 196 444 109 291 1143 783 181
Max 4986 1610 2130 2081 1022 8165 2015 1086 1989
Min 106 174 1727 25 8406 34 529 481 1292
Appendix D
Figure A
Plot of the ratio of long term variable compensation (stocks & options) to salary and ROA
-0.05 -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04 0 1 2 3 4 5 6
Stocks & Options / Salary
R
O
A
Figure B
Plot of the ratio of long term variable compensation (stocks & options) to salary and stock return
-0.1 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0 1 2 3 4 5 6
Stocks & Options / Salary
Figure C
Plot of the ratio of long term variable compensation (stocks & options) to salary and the Tobin’s Q
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 1 2 3 4 5 6
Stocks & Options / Salary
