Identifying causality between CEO bonus compensation and firm performance

(1)

Identifying causality between CEO bonus

compensation and firm performance

Galjaard, Koen

10430067

June 27, 2016

(2)

2

Statement of Originality

This document is written by Koen Galjaard, 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 completion of the work, not for the contents.

(3)

3

Abstract

Incentive contracts are designed to provide bonus compensation to the CEO when he acts in the firms interest, by augmenting firm performance. Therefore bonus compensation should influence the firm performance. The goal of this thesis is to identify causality in the relation between firm performance and bonus compensation for the CEO. Granger causality tests are used to test for indications on causality. Firm performance is measured by return on assets and return on equity, bonus compensation is measured by summing all available incentive bonus awards. In this field of study, many studies have focused on measuring the magnitude of the incentive bonus effect, while causality has often been overlooked. This study is performed on 503 North-American firms with at least $10 billion total assets. The results show no statistical evidence that bonus compensation causes firm performance, while an indication is found of firm performance causing bonus compensation.

(4)

4

1. Introduction

The rapid rise in chief executive officer compensation over the past three decades has intensified the debate about CEO compensation (Frydman, 2010). Reports on exorbitant bonuses for CEOs linked to poor performances continue to appear in newspapers. For example, the Wall Street Journal reported on 30 January 2015 that IBM’s CEO Gini Rometty was to receive a $3,6 million bonus, alongside $13,3 million on long-term incentive stock, after leading the company in fifteen consecutive quarters of declining profit and share-price. Reports like this have left the public baffled and have created an outcry over CEO bonus compensation.

Some researchers, like Fahlenbrach (2009), are of the opinion that the level of CEO compensation is a result of demand and supply in the managerial market. Where they assume demand for a powerful and competent manager is high and the supply of managerial talent is scarce. Other researchers find that incentive contracts of the CEO are influenced by managerial power and are therefore supplying excessive and ineffective compensation (Brick et al, 2006).

The main goal of incentive bonus contracts for the CEO, besides attracting the right person for the job, is to motivate the CEO to create firm value (Eisenhardt, 1989). Hence the effectiveness of incentive bonuses can be measured by identifying the effect of bonus rewards for the CEO on firm performance (Jensen & Murphy, 1990).

The goal of this thesis is to give an indication of causality in the relation between incentive bonus rewards for the chief executive officer and firm performance. The main hypothesis used to test this relation is: bonus compensation for the CEO has a positive significant effect on firm performance. If no positive significant effect can be found, the use of large incentive bonus awards can be questioned. To form an answer, 503 North American companies with total assets of more than ten billion dollar have been analyzed between the years 1996 and 2014. This sample has been picked randomly, only to be filtered on firm size and the fact that the firms are North American.

The hypothesis is tested by Granger-causality tests, an explanation on this particular test is provided in the method section. Granger-causality is not the same as causality, as will also be explained in the method section, but it does provide an indication of causality and the direction in which causality runs. Hence, it can identify whether bonus compensation truly has a positive effect on firm performance.

There is a discussion among researchers if bonus incentives for the CEO achieve their goal of properly motivating the CEO to create firm value. Mehran (1995) found a positive relation between firm performance and CEO bonus compensation. Thus confirming that the bonus incentive for the

(6)

6 CEO achieved its goal of creating firm value. Bebchuk (2009) on the other hand finds no significant relation.

This study differs from previous research, because it does not focus on measuring the magnitude of the relation between bonus compensation and firm performance, but attempts to identify a causal relation. This approach is inspired by the work of Bertrand and Mullainathan (2001), who investigated whether CEOs are being rewarded for luck.

Furthermore, previous work, like Murphy (1999) and Mehran (1995), does not focus on lagged observations for the dependent variable, while this study does. When not using lagged observations, one encounters the classic chicken-and-egg causality problem. As it is impossible to tell whether the dependent variable causes the independent variable or vice versa; which came first, the chicken or the egg? The bonus or the firm performance?

The results show statistical evidence that there is no indication that bonus compensation for the CEO causes firm performance. Since no significant indication of causality can be found, the existence of large incentive bonus awards must be questioned.

This thesis also tries to identify whether causality might be running in the opposite direction. Thus if lagged firm performance might be causing the CEO’s bonus compensation. The tests used in this work provide an indication of causality, but do no test true causality. Therefore causality cannot be established conclusively. The results do indicate that causality might be present, as statistical evidence is found that indicates a causal relation of firm performance on CEO bonus compensation.

(7)

7

2. Literature

In this section an overview of related concepts and several studies on the same subject is provided. First some theoretical background is presented, then several studies on the same subject are presented. After presenting the studies, the relation of those studies to this thesis is discussed.

2.1 Theoretical Background

The main theory that supports the relation between firm performance and CEO bonus, is the agency theory (Berk and DeMarzo, 2007). Even though the agency theory is not without controversy, it is widely used among organizations to form compensation contracts for executives (Eisenhardt, 1989). The principal-agent theory divides two parties, the agent and the principal. It assumes that both the agent and the principal are utility maximizing. A problem then arises, when the principal and the agent have different interests (Jensen and Meckling, 1976). This problem is known as the agency problem.

In big listed firms, it is common that shareholders have delegated the day-to-day authority to a CEO. In this case the CEO is the agent, whereas the shareholders act as the principle. The interaction between these two parties is an example of the agency problem.

To deal with the agency problem in modern management, incentive contracts are used to align executives interest with the interest of shareholders. The agency problem is the main rationale for shareholders to pursue incentive an incentive contract for their chief executive. This thesis will investigate if incentive bonus compensation contribute to firm performance. One could say that, if there is no significant relation, there is no ground for large bonus incentives.

The solution to the agency problem can, according to economic literature, be found in the optimal contracting theory. The optimal contracting theory assumes that the board of directors, acting on behalf of the shareholders, designs an incentive contract for the CEO. This contract provides incentive for the CEO to maximize shareholders value. These incentives should be measurable, so that compensation depends on the performance of the CEO (Eisenhardt, 1989).

The goal of the optimal contracting theory is to align shareholders’ and CEO’s interest. The optimal contracting theory assumes that executives are utility maximizing and are therefore willing to go the extra mile, when their effort will lead to a higher reward. In a perfect contract, the interest of shareholders and executives will be perfectly aligned.

(8)

8

2.2 Overview of existing literature

The study that stands closest to this thesis is ‘Are CEO’s rewarded for luck?’ of Bertrand and Mullainathan from 2001. Bertrand and Mullainathan investigate whether CEOs are rewarded for luck, rather than skill. They qualify luck as ‘changes in firm performance that are beyond the CEO’s control’. Tying pay to luck will not create more incentive for the CEO to increase firm value and is therefore undesired (Holmstrom, 1979). Bertrand and Mullainathan find that the incentive payment of a CEO is evenly sensitive for a lucky dollar as for a general dollar.

The conclusion of these results is that a substantial part of the CEO’s incentive compensation is caused by luck. This detracts from the view that incentive payments are a stimulus for the CEO to set his interest in maximizing firm value, because it is not possible to tell whether a CEO is creating firm value or is just lucky (Bertrand and Mullainathan, 2001). Consequently, it shows that incentive contracts do not achieve their goal of creating purely incentive payments, because they also pay for luck.

The first to deliver an acknowledged study on the relation of firm performance and bonus compensation were Jensen and Murphy in 1990. They measured the change in CEO income per change of 1000 dollar in shareholders income. The results showed that CEO income would change by $3,25 per $1000 change in shareholder wealth. Jensen and Murphy (1990) argue that restrictions, caused by political forces have reduced pay-performance sensitivity. They use this argument to explain the result that pay to performance ration has declined over time for executives. Another interesting conclusion of Jensen and Murphy (1990) is that CEOs in large firms have less

performance based incentives than CEOs in smaller firms, particularly; they have less stock of their company.

Mehran (1995) investigates compensation structure among manufacturing firms. He comes to two general conclusions. The first is that form rather than the level of compensation is important in incentive contracts. Mehran finds that stocks and options in particular motivate the CEO to increase firm performance. His work also finds that compensation incentives have a positive and significant relation with firm performance. He therefore claims that the evidence found in his paper supports advocates of incentive compensation (Mehran, 1995).

Garen (1994) sets up two models to examine the level and structure of executive compensation. He finds that the incentive payments in incentive contracts are inconsistent with performance payment, meaning executives don’t necessarily get paid when they perform well and vice versa. This adds to his conclusion that the statistical significance of relative performance pay is low. The main reason for the low statistical significance is that it is unclear which determinants

(9)

9 should be put in incentive contracts. There is no clear determination of incentive payment that creates incentive and measures the amount of effort, made by the executive (Garen, 1994).

Bebchuk and Fried (2003) investigate the process of setting up a contract for the CEO through the agency problem lens. In theory, the principal should be able to set an optimal contract to contract an agent. In their study, Bebchuk and Fried investigate whether this theory is true. They hypostatize that companies with dispersed ownership can’t set an optimal contract, because they lack power in the negotiation with a CEO. The results indeed show that managerial power has influence on the contracting process. This leads to inefficient and non-optimal contracts (Bebchuk & Fried, 2003).

2.3 Relation with existing literature

The results of previous studies regarding the relation between firm performance and executive compensation are ambiguous. They do, however, have a common factor, as no previous work focuses on identifying a causal relation. Some studies touch the subject of causality, like Garen and Bertrand, but their main focus is never causality. In that respect, this study seems unique.

Bertrand and Mullainathan show that a CEO does not fully control firm performance, subsequently he does not control the parameters that determine his bonus compensation. For this reason, one could question the causal relation of firm performance and bonus compensation, which is exactly what this study will do. Therefore this thesis will form an addition to the work and conclusions of Bertrand and Mullainathan.

The results of Mehran’s paper contradict with the results in this thesis, as no evidence for an impact of compensation incentives on firm performance is found in this thesis. Two reasons can be named for the contradiction in results. Firstly, Mehran only investigates firms within one industry, the manufacturing industry. For this reason one could argue the significance of his work, as the findings may be industry specific. In addition, his sample period and size are quite small, his work considers 153 firms over a period of two years.

The second and most important explanation for the contradiction in results is the difference in approach. Incentive contracts provide bonus payments when certain goals, regarding firm value are met. Hence bonus distributed to executives is higher when firm performance is better. Mehran (1995) measures the equity and non-equity incentive compensation effect on firm performance, but does not investigate a causal relation. Whereas this thesis attempts to identify the existence of a

(10)

10 causal relation between incentive compensation and firm performance. Using the words of Bertrand Mullainathan: ‘the firms and executives in Mehran’s work could just have been lucky’.

The same reason applies to the difference in results between this study and that of Jensen and Murphy (1990). They also did not attempt to find causality, but focused on measuring the magnitude of the relation between CEO compensation and firm performance. Because of this, they find that CEO compensation increases when firm performances increases and vice versa, but they cannot identify which is caused by the other.

Garen takes the economic approach by setting up models for the CEO’s utility and income. This study takes a different approach, but the conclusions of our work don’t differ that much. Both works conclude that the statistical significance of bonus compensation on firm performance is low, or even nonexistent. The conclusion of Garen, that it is unclear which determinants should be put in incentive bonus contracts, is one of the advices of future research in this study. Because this could be part of the reason that no causal effect is found of bonus compensation on firm performance.

This work and the work of Bebchuk and Fried complement each other. This study shows that incentive contracts are not working as intended and are thus non-optimal. Whereas the study of Bebchuk and Fried names managerial power as an explanation for non-optimal incentive contracts. That provides an explanation on the results in this study; namely that bonus compensation does not cause firm performance, because the bonus contracts are not optimal.

(11)

11

3. Data

3.1 Construction of variables

The data on CEO compensation are gathered from Execucomp and the annual firm performance data are gathered from Compustat. All data are for 1996-2014. After merging these two datasets, all companies with missing values for company results or CEO compensation have been removed. The latter was often reason for removal. This thesis focusses on companies with a large amount of assets. Companies with less than ten billion assets have been removed. As a result, the data includes 503 North American companies.

The downloaded data are manipulated in order to be able to carry out the tests. Table 1 shows how data are manipulated. Firm performance will be measured in return on equity and return on assets, these will be the dependent variables. Because these variables are relative, a relative version of the main explanatory variable is required as well. Therefore the variable relative total bonus is created, by dividing total bonus by total compensation.

Total bonus is computed as follows; pre-2006 the total bonus is equal to the bonus fee, while after 2006 the total bonus is the sum of the bonus fee and the value of shares and options awarded. Total bonus is computed this way, because financial reports on CEO bonus have changed over time. Pre-2006 there is no data available on bonus rewards other than the bonus fee ($) paid to CEOs. Whereas from 2006 and on, data is available on the value of shares and options awarded.

(12)

12

3.2 Descriptive Statistics

Table 1: List of variables

Variables Description

Total Assets Total Assets

Net Income Net Income

Shareholder Equity Shareholder Equity

Salary CEO salary

Bonus Bonus fee awarded to the CEO Stock Awards ($) Value of options awarded to the CEO Option Awards ($) Value of stocks awarded to the CEO Total Compensation Total compensation for the CEO

Mutated Variables

Total Bonus Bonus fee, stock awards, option awards combined Return on Equity Net income divided by shareholder equity

Return on Assets Net income divided by total assets

Relative Total Bonus Total bonus divided by total compensation Relative Stock Stock awards divided by total compensation Relative Options Option awards divided by total compensation Relative Bonus Bonus fee divided by total compensation Relative Salary Salary divided by total compensation Log Total Assets Logarithm of total assets

Log Salary Logarithm of salary

Log total Compensation Logarithm of total compensation

(13)

13

Table 2: Descriptive statistics

Variables

Original Variables Obs Mean Std. Dev. Min Max Total Assets 7,935 53228.42 186822.3 90.12 3270108 Net Income 7,933 1160.408 3995.408 -99289 83963 Shareholder Equity 7,935 9381.684 19040.65 -86154 243471 Salary 7,942 986.0794 520.6812 0 8100 Bonus 7,942 951.2674 2422.875 0 76951 Stock Awards ($) 4,256 3894.442 5217.707 -7230.189 131939.8 Option Awards ($) 4,256 2165.907 4558.121 0 90693.4 Total Compensation 7,923 9784.925 15439.83 0 655448 Total Bonus 7,893 4206.633 6546.001 0 145059.6 Computed Variables Return on Equity 7,933 0.1668313 3.08013 -113.4568 141.7419 Return on Assets 7,933 0.0374522 0.1037096 -4.583101 0.9017305 Relative Total Bonus 7,856 0.4347459 0.59511 0 31.95739 Relative Stock 4,242 0.3807816 0.4297334 -0.486027 12.53919 Relative Options 4,242 0.2030378 0.5120973 0 23.74856 Relative Bonus 7,905 0.1204619 0.1712174 0 0.9999995

Relative Salary 7,905 0.1998696 0.1839565 0 1

Log Total Assets 7,935 9.724708 1.307211 4.501142 15.00033

Log Salary 7,942 6.684633 1.035739 0 8.999743

Log Total Compensation 7,923 8.674961 1.216607 0 13.39308 Shareholder Equity Ratio 7,935 0.3171 0.2099365 -1.39967 0.965588

(14)

14

4. Method

4.1 Hypotheses

The main hypothesis is;

 Bonus compensation does not Granger-cause firm performance. The second-order hypothesis, to determine the direction of causality, is;

 Firm performance does not Granger-cause bonus compensation.

To test these hypotheses we will need quantification for firm performance and bonus compensation. The firm performance will be measured in return on assets and return on equity, the bonus

compensation will be measured in relative total bonus. An explanation on the computation of these variables is found in the data section.

To answer the main hypothesis, several more specific sub-hypotheses have been set up

 Relative total bonus does not Granger-cause return on equity.

 Return on equity does not Granger-cause relative total bonus.

 Relative total bonus does not Granger-cause return on assets.

 Return on assets does not Granger-cause relative total bonus.

4.2 Models

To test the relation between bonus compensation for CEOs and return on equity, granger causality tests will be used. The Granger causality test is a statistical concept, which determines whether one time series is able to help predict another time series. A variable X is Granger-causing variable Y, if past values of X combined with past values of Y are significantly better in predicting Y than past values of Y alone. As this means that past values of X contain information, which is not present in past values of Y, that help predict Y (Granger, 2001).

Granger causality tests use lagged versions of the dependent variable as independent variables, combined with lagged versions of a dependent variable. This test provides accurate approximations in which direction the causality runs, but one should keep in mind that Granger causality does not imply pure causality.

(15)

15 To test whether bonus compensation Granger-causes firm performance, regressions were estimated with different levels of lag. The regressions are as follows.

Regression 1

𝑅𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝐸𝑞𝑢𝑖𝑡𝑦𝑡 = µ + ∑[𝛼𝑘Return on Equity𝑡−𝑘+ β𝑘Relative Total Bonus𝑡−𝑘] 4

𝑘=1

+ 𝜀

Regression 2

𝑅𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑇𝑜𝑡𝑎𝑙 𝐵𝑜𝑛𝑢𝑠𝑡 = µ + ∑[𝛼𝑘Relative Total Bonus𝑡−𝑘+ β𝑘Return on Equity𝑡−𝑘] + 𝜀 4

𝑘=1 Regression 3

𝑅𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝐴𝑠𝑠𝑒𝑡𝑠𝑡 = µ + ∑[𝛼𝑘Return on Assets𝑡−𝑘+ β𝑘Relative Total Bonus𝑡−𝑘] 4

𝑘=1

+ 𝜀

Regression 4

𝑅𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑇𝑜𝑡𝑎𝑙 𝐵𝑜𝑛𝑢𝑠𝑡 = µ + ∑[𝛼𝑘Relative Total Bonus𝑡−𝑘+ β𝑘Return on Assets𝑡−𝑘] + 𝜀 4

𝑘=1

These regressions are set up to test if the variable related to the β coefficient, provides information that helps predict the dependent variable. Consequently, the main coefficient of interest is β. When β turns out to be significant, the variable related to β is said to Granger-cause the dependent variable.

The tests will be performed on different levels of lag, namely one year up to a maximum of four years. Lag is denoted as ‘k’, the constant is denoted as ‘µ’, the error term is denoted as ‘ε’.

(16)

16

5. Results

This section describes the results of the regressions and evaluates the hypotheses, given the outcome of the tests. The numerical results are represented in the tables at the bottom of this section.

There is no evidence that relative total bonus Granger-causes return on equity, as can be seen in table 3. The coefficient related to relative total bonus in regression 1 is β. The results show that β is not significant for any level of lag. Therefore no indication is found that relative total bonus might cause return on equity.

The results on causality running in the opposite direction are also insignificant. Return on equity provides no significant information that helps predict relative total bonus. Hence, return on equity does not Granger-cause relative total bonus and therefore no indication is found that return on equity causes relative total bonus.

Although the total regression is significant, the results show that the influence of relative total bonus on return on assets is not significant. These conclusions hold for every level of lag and can be derived from table 5, as the F-statistic is significant at a 1% level, but the coefficient of relative total bonus is not significant at a 10% level.

The results in table 6 show that return on assets has a significant effect on relative total bonus, for some levels of lag. The value of the coefficient associated with return on assets is significant on a 5% level for lag values of one and two years. Thus there is statistical evidence that return on assets Granger-causes relative total bonus.

In table 7 the coefficients of regression 4 and their significance have been listed. The main coefficient of interest is return on assets, as we try to identify the effect of return on assets on relative total bonus. The coefficients of return on assets for one and two years ago are positive and significant, hence we can conclude that past return on assets, for up to two years ago, have a significant positive effect on relative total bonus.

Considering the four sub-hypotheses, not enough evidence has been found to reject the first three hypotheses. However, statistical evidence is found to reject the fourth sub-hypothesis;

concluding that return on assets does Granger-cause relative total bonus.

The tests on the four sub-hypotheses allow for an answer on the main hypothesis and the second-order hypothesis. All tests regarding the influence of bonus compensation on firm

performance gave no significant results. Consequently, the main hypothesis still stands; bonus compensation does not Granger-cause firm performance. The second-order hypothesis can be rejected, as evidence is found that firm performance does Granger-cause bonus compensation.

(17)

17

Table 3: results of tests with regression 1

k Degrees of freedom F-Statistic Prob. > F P-value β P-value α R-squared

1 (2 , 7300) 0.32 0.7282 0.943 0.428 0.0001

2 (2 , 6793) 1.16 0.3129 0.751 0.135 0.0003

3 (2 , 6306) 0.66 0.66 0.320 0.569 0.0002

4 (2 , 5818) 21.08 0 0.846 0.000 0.0072

Table 4: results of tests with Regression 2

k Degrees of freedom F-statistic Prob. > F P-value β P-value α R-squared

1 (2 , 7244) 217.94 0.000 0.2 0.000 0.0568

2 (2 , 6740) 231.39 0.000 0.977 0.000 0.0642

3 (2 , 6255) 95.62 0.000 0.357 0.000 0.0297

4 (2 , 5771) 123.19 0.000 0.184 0.000 0.041

k Degrees of freedom F-statistic Prob > F P-value β P-value α R-squared

1 (2 , 7300) 1113.57 0.000 0.857 0.000 0.2238

2 (2 , 6793) 186.11 0.000 0.316 0.000 0.0519

3 (2 , 6306) 118.63 0.000 0.52 0.000 0.0363

4 (2 , 5821) 90.56 0.000 0.702 0.000 0.0302

k Degrees of freedom F-statistic Prob > F P-value β P-value α R-squared

1 (2 , 7244) 220.09 0.000 0.017 0.000 0.0573

2 (2 , 6740) 233.73 0.000 0.036 0.000 0.0649

3 (2 , 6252) 95.42 0.000 0.496 0.000 0.0296

(18)

18

Table 7: coefficients of Regression 4

Coefficient and p-value

k Return on Assets P-value RoA Relative Total Bonus P-value RTB Constant P-value Constant

1 0.1283942 0.017 0.1860822 0.000 0.352263 0.000

2 0.1098976 0.036 0.2291991 0.000 0.347196 0.000

3 0.0373438 0.496 0.1579705 0.000 0.395016 0.000

4 0.0594296 0.302 0.1856491 0.000 0.402268 0.000

6. Discussion

The implication of these results is that firm performance might in fact be the cause for bonus rewards, instead of bonus rewards being the cause for firm performance. No evidence is found that supports the work of Mehran (1995), who found a significant effect of bonus incentives on firm performance. While this thesis can complement the work of Jensen (1990), by giving an indication of causality for the significant relation he found. Jensen found that for every $1000 increase in

shareholder value, the CEO income increases by $3,25. In light of the results found in this thesis, one could say that the increase in shareholder value is more likely to cause the increase in CEO income than vice versa.

This work could also support the findings of Bebchuk and Fried (2003), who find that incentive contracts are non-optimal. Bebchuk and Fried show that managerial power is a cause for non-optimal incentive contracts. Although this work does not investigate that cause, it can be seen as a form of re-assurance that incentive contracts are in fact non-optimal in the real world.

The results of this study form an addition to the conclusion of Bertrand and Mullainathan (2001), that CEOs are partly rewarded for luck. They show that a CEO does not fully control firm performance. This study notes that firm performance of the past has significant influence on the CEO’s bonus compensation. The implication is that a CEO’s bonus compensation is, at least partly, dependent on factors he cannot control. Let alone when a CEO joins a new firm, his payment will then be influenced by past results, over which he had no control at all.

This study’s results are in line with the results of Garen (1994). In fact, they could serve as an explanation for Garen’s results; who concluded that the statistical significance of performance pay is low. This work demonstrated that bonus compensation has no causal effect on firm performance, hence the statistical significance of performance pay is low.

(19)

19

7. Conclusion

The models used in this thesis can only provide indications of causality and do not prove pure causality. Therefore we must be cautious with conclusions, but several conclusions can be made nevertheless.

No indication of causality is found for the effect of bonus compensation on firm

performance. If not even an indication of causality can be found, then a causal effect is not present. Bonus compensation does not cause firm performance. Consequently, no statistical evidence is found that bonus compensation for the CEO has a positive effect on firm performance. In fact, indications of causality running in the opposite direction have been found. On the basis of these results, we cannot justify the existence of large incentive contracts for CEOs.

The indications of causality that have been found are ambiguous. As tests for return on equity on firm performance gave no indication of causality, while tests for return on assets on firm performance did give an indication of causality. It would be interesting to investigate why the causal indication for return on assets is stronger than the causal indication for return on equity.

This study encourages a revision of the way incentive contracts are set up, as they don’t work as intended. The incentive contracts aim to motivate the CEO to increase firm value, for which they will be awarded bonus compensation. Accordingly, a significant positive effect of bonus

compensation on firm performance is expected to be found. No such relation is found in this work, which undermines the legitimacy of large amounts of bonus compensation.

The indications of firm performance having a causal effect on bonus compensation further detracts the legitimacy of bonus incentives. This thesis made clear that past firm performance, on which the current CEO can have no control, is influencing the CEO’s bonus compensation. Therefore the conclusion can be drawn that CEO’s are in fact partially rewarded for luck, as Bertrand and Mullainathan also concluded in their work.

Interesting subjects for further research have been found. Return on assets gave significant results when tested for a causal effect on firm performance, while return on equity did not. A study on the cause of these different results would form a great addition to this study.

Causality is a delicate issue and so is testing for it. Causality tests have often been disputed and an agreement on the definition of causality is yet to be found. For this reason I like to emphasize that this study provides an indication of causality, further investigation is required to identify if the indication for causality is indeed a trace for true causality.

(20)

20 A problem has been addressed in this study; namely the problem that incentive bonus contracts are not functioning as intended. This problem is yet to be solved, hence we suggest that future studies attempt to find possible solutions for this problem.

(21)

21

8. References

Bebchuk, L. and Fried, J. (2003). ‘Executive Compensation as an Agency Problem.’ Journal of Economic Perspectives, 17(3), pp. 71–92.

Bebchuk, L. and Fried, J. (2004). ‘Pay without performance. The unfulfilled promise of executive compensation.’ Cambridge: Harvard University Press.

Berk, J.B. and DeMarzo, P.M. (2007). ‘Corporate Finance’. Boston: Addison Wesley.

Bertrand, M. and S. Mullainathan (2001). ‘Are CEOS rewarded for luck? The ones without principals are’. The Quarterly Journal of Economics, Augustus, pp. 901-932.

Brick, Ivan E., Oded Palmon, and John K. Wald. 2006. ’CEO compensation, director compensation, and firm performance: Evidence of cronyism?’ Journal of Corporate Finance, 12: 403-423.

Core, J., Holthausen, R. and Larcker, D. (1999). ‘Corporate governance, chief executive officer compensation, and firm performance.’ Journal of Financial Econonomics, 51, pp. 371–406.

Eisenhardt, K.M. (1989). ‘Agency Theory: An Assessment and Review.’ The Academy of Management Review, 14 (1), pp. 57-74.

Fahlenbrach, R. (2009). ‘Shareholder rights, boards, and CEO compensation.’ Review of Finance, 13, pp. 81-113.

Fama, E. (1980). ‘Agency Problems and the Theory of the Firm.’ Journal of Political Economy -Vol. 88, pp. 288-307

Frydman, K. and Jenter, D. (2010). ‘CEO Compensation’. National Bureau of Economic Research Working Paper No. 16585 - Issued in December 2010

Garen, J.E. (1994). ‘Executive compensation and principal-agent theory’. Journal of political economy, volume 102, pp. 1175-1199

(22)

22 Granger, C.W.J. (1969). ‘Investigating causal relations by econometric models and cross-spectral methods.’ Econometrica: Journal of the Econometric Society.

Granger, C.W.J., Ghysels, E., Swanson, N.R. and Watson, M.W. (2001). ‘Essays in econometrics: collected papers of Clive WJ Granger.’ Econometric Society Monographs.

Hall., B., and J. Liebman (1998). “Are CEOs really paid like bureaucrats?”. Quartely Journal of Economics, 102, pp. 653-691

Hölmstrom, B. (1979). ‘Moral hazard and observability’. The Bell journal of economics, vol. 10, pp. 74-91

Jensen, M.C. and Murphy, K.J. (1990). ’Performance pay and top-management incentives.’ Journal of political economy, volume 98, pp.225-264.

Jensen, M.C. and W.H. Meckling (1976). ‘Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure.’ Journal of Financial Economics, October, pp. 305-360.

Mehran, H (1995). ‘Executive-compensation structure, ownership, and firm performance.’ Journal of financial economics, 88, pp. 1-25.

Murphy, K. (1999). ‘Handbook of Labor Economics’ -Volume 3, Part B, Pages 2485–2563, Chapter 38 Executive compensation .

The Wall Street Journal, January 30, 2015 ‘IBM CEO Rometty to receive $3.6 million bonus’. - http://www.wsj.com/articles/ibm-ceo-rometty-to-receive-3-6-million-bonus-1422654063

Identifying causality between CEO bonus compensation and firm performance