• No results found

The effect of executive compensation on firm performance

N/A
N/A
Protected

Academic year: 2021

Share "The effect of executive compensation on firm performance"

Copied!
28
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

The Effect of Executive Compensation on Firm

Performance

Abstract

This paper investigates the influence of executive compensation on firm performance. For executive compensation total compensation and four components namely salary, bonus, option based compensation and stock based compensation were used. Neither total compensation nor any of the four components of total compensation was a significant predictor of firm performance. When equity holdings is added as a predictor still none of the predictors is significant, implying no effect of executive compensation on firm performance. In addition a lag-test showed that total compensation, salary, bonus, option based compensation and stock-based compensation for the timepoint t are no significant predictors of firm performance at timepoint t+1. Conclusively, the results in this paper provide no support of an effect of executive compensation on firm performance, which contradicts the optimal contracting approach. However a note should be added that due to a lower number of observations compared to former research the statistical power of this study might not have been sufficiently high to detect a possibly small effect.

Klaas van Bork 10444017

Semester 2, 2014/2015 29/06/2015

Bachelor thesis Economics and Finance Supervised by Jan Lemmen

(2)

Statement of Originality

This document is written by Student Klaas van Bork 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)

Contents

1 Introduction ... 4 2 Literature review ... 5 2.1 Theoretical background ... 5 2.2 Former research ... 6 2.3 Hypotheses ... 9 3 Methodology ... 10 3.1 Model ... 10 3.2 Data ... 15 3.3 Summary statistics ... 16 4 Results ... 18

5 Conclusion & discussion ... 23

References ... 25

(4)

1 Introduction

Since the financial crisis of 2008 a new wave of criticism has evolved on the decision making process of firms. The risk return trade-off is one of the key parts of discussions and debate lately. Since the 1980s there has been an increase in executive compensation (Yermack, 1995). As the executives are responsible for making most decisions, the executive compensation is directly related to the risk return trade-off. Also according to Yermack (1995) this increase in executive compensation consisted mainly of an increase in option based compensation. These options are granted as an incentive to take on more risk, leading to higher firm performance. For CEO’s and CFO’s of 30 companies listed on the Dow Jones Industrial Average between 2006 and 2014 the average compensation was 11,8 million US$. Of this compensation on average only 1,1 million US$ is salary while 2,5 million US$ is option based compensation. Because of the incentive based background of these compensations and the fact that they cover relatively large amounts of money it is interesting to look into the effects of these amounts of compensation and its components on firm performance.

To get a deeper understanding of this matter it is important to examine the relation between executives and shareholders. There is a difference in interest between these two parties. This difference in interest is defined by Berle and Means (1932) as a typical principal-agent or agency problem (Aggarwal & Samwick, 1998). According to Aggarwal and Samwick tying executive compensation to firm performance is a mechanism to align executives with shareholders (1998). Compensation can thus be seen as a mechanism providing incentives to executives to take on more risk than they normally would. Executive compensation can be divided into four components: salary, bonus, option based compensation and stock based compensation. In this paper the relation between all these four components of compensation and firm performance is examined to provide answers on how and to what extend these four components form incentives to executives in the decisions they make to maximize firm performance. Salary is a predetermined fixed component of compensation received by the executive. Bonus is an incentive based compensation based on the performance of an executive received on a yearly basis. Option-based compensation and stock-based compensation are two interesting types of compensation which both have an incentive based background. According to Agrawal and Mandelker option based compensation and stock based compensation both have a significant influence on risk taken in investment strategies by executives (1987). Deciding on the levels of these components of executive compensation and thus choosing to what extent

(5)

incentives will be given to executives to increase firm performance is up to the board of directors.

This paper deals with the influence of total compensation and the four previously described components on firm performance. The results in this paper do not provide evidence of an influence of executive compensation on firm performance and thus do not support the optimal contracting approach. The outline of this paper is as follows. First, in chapter two, existing literature will be reviewed to give a theoretical background, followed by a summary of earlier findings and the hypotheses that will be tested in this paper. Thereafter, Chapter three provides an explanation of the methodology used to test the hypotheses, complemented with a brief description of the data. Chapter four discusses the results of this study. Finally, chapter five discusses the theoretical interpretation of these results, draws a cautious conclusion and provides some suggestions for follow up studies.

2 Literature review

In this chapter a theoretical background on the subject will be provided and the findings of earlier research and previous literature will be summarized.

2.1 Theoretical background

Previous studies have a widespread view on the influence of executive compensation on firm performance. The reasoning behind this influence starts with the relation between executives and share- or stakeholders of a company. This relation is often seen as an agency problem. Executives are risk averse and self-interested. This is often not aligned with the shareholders’ targets of maximizing shareholders return. To shareholders there is incomplete information about investment opportunities and possible managerial actions top-executives face (Jensen & Murphy, 1990a). Previous literature has brought forward three different approaches on the debate on how executive compensation can be linked to firm performance and the role of the agency problem in this debate.

The first and most dominant of these approaches among financial economists is the optimal contracting approach (Bebchuk & Fried, 2003). In this approach the board of directors design executive compensation as a principal agent contract to give risk averse executives an incentive to maximize shareholders value (Bebchuk, Fried & Walker, 2001). Bebchuk and Fried (2003) suggest that according to the optimal contracting approach executive compensation can be seen as a possible instrument that solves the agency problem. According to Jensen and Murphy (1990b) the board of directors can structure salaries, bonuses, and stock so that it provides

(6)

rewards for superior performance and penalties for poor performance. They state that tying the welfare of top-executives to that of shareholders through compensation policy is a common way of dealing with the agency problem (Jensen & Murphy, 1990a)

Bebchuk and Fried (2003) proposed a second approach on how to understand the link between executive compensation and firm performance: the managerial power approach. In this

approach executive compensation is viewed as a part of the agency problem. Executives and managers are able to influence their own compensation and these compensations rather show characteristics of rent-seeking behavior. This view stems from the fact that the compensation for executives is determined by the board of directors. Often top executives themselves are members of board, indicating that they have a substantial influence on their own compensation. Murphy (2002) writes in response to Bebchuk, Fried and Walker (2001), that even though he also thinks there is more to executive compensation in relation to the agency theory than just optimal contracting, the managerial power approach is too simplistic to explain executive compensation. Also it doesn’t cover the growth in option based compensation since 1990. Murphy (2002) introduced the perceived-cost approach. This approach covers the large increase in option based compensation. According to this view options are valued at their economic value with the Black-Scholes formula. Yet, from the perspective of the board of directors, the cost of granting options is valued far below this economic value. This is due to the fact that these options can be granted on a non-cash charge basis. Thus granting these options has tax and accounting advantages and is viewed as a cheap form of compensation (Murphy, 2002).

2.2 Former research

The most dominant of these approaches among financial economists is the earlier mentioned optimal contracting approach (Bebchuk & Fried, 2003). According to this approach executive compensation is used as a contractual instrument to minimize agency costs and in this way maximize shareholders value. Murphy (1985) finds that executive compensation and firm performance are strongly positively related when measured by shareholder return and growth in firm sales. Hall and Liebman (1997) confirm Murphy’s findings, finding a strong positive relation between executive compensation and firm performance when equity holdings are taken into account. Jensen and Murphy (1990b) suggest that conceptually the board of directors can combine three basic policies to create incentives for executives to maximize firm performance.

(7)

These three policies are:

- Boards can require that CEOs become substantial owners of company stock. - Salaries, bonuses, and stock options can be structured so as to provide big

rewards for superior performance and big penalties for poor performance. - The threat of dismissal for poor performance can be made real.

Nevertheless Jensen and Murphy (1990b) state that their findings on compensation are, in reality, in contradiction with the principals of these policies. Though their earlier findings suggest that the statistical effect of executive compensation on firm performance is small but significant (1990a), they state that these findings are to small to suggest that changes in annual compensation of executives affect firm performance (Jensen & Murphy, 1990b). This is in contrast to Murphy’s earlier findings (1985). Aggarwal and Samwick (1998) state that the influence of executive compensation on firm performance depends on the variance of firm performance. For firms with more volatile stock prices the relation between executive compensation and firm performance is stronger.

When focusing on the equity based part of total compensation Ofek and Yermack (2000) suggest that stock based compensation and option based compensation succeed in providing incentives to executives when these executives are not yet substantial owners of company stock. When ownership of company stock increases the incentive based nature of equity based compensation decreases because of the possibility to sell previously owned shares (Ofek & Yermack, 2000). Conyon and murphy (2000) find that when comparing firms in the United States with firms in the United Kingdom a higher option based compensation leads to higher firm performance. Agrawal and Mandelker (1987) show that holding more stock and options induces executives to take more risk in investments. These component specific views towards the influence of compensation on firm performance are later supported by Mehran (1995). He suggests that it is rather the form of the compensation than the level of compensation that motivates the executive to increase firm performance (Mehran, 1995). Mehran (1995) states that Firm performance measured by Tobins Q and return on assets are positively related to the percentage of stock and options held by executives.

(8)

Table 1 provides a list of the findings on the influence of executive compensation on firm performance, as discussed above.

Table 1: Former findings

Author(s) Region Year Factor of effect Has effect on Effect

Murphy US 1985 Executive compensation Firm performance Positive

Agrawal &

Mandelker US 1987 Equity holdings Induced risk Positive

Jensen & Murphy US 1990a,b Executive compensation Firm performance Small to none Mehran US 1995 Equity holdings Firm performance Positive Hall &

Liebman US 1997 Executive compensation Firm performance Positive (strong relation)

Aggarwal &

Samwick US 1998 Executive compensation Firm performance Positive (effect stronger when variance of firm performance higher) Conyon &

Murphy US/UK 2000 Option based compensation Firm performance Positive

Ofek &

Yermack US 2000 Equity based compensation Firm performance Positive

(9)

2.3 Hypotheses

This research paper focusses on the influence of total compensation and four of its components on firm performance. Based on the results a conclusion can be drawn on whether they support the optimal contracting approach. In this chapter the hypotheses are given with a theoretical background based on the optimal contracting approach.

Optimal contracting theory states that executive compensation is designed to give risk averse executives an incentive to maximize shareholders value (Bebchuk, Fried & Walker, 2001). Correspondingly, it is expected that total compensation has a positive effect on firm performance (measured by total shareholders return):

(H1) Total compensation has a positive effect on firm performance

Salary is a predetermined fixed component of compensation received by the executive. Because this component of compensation has no upside potential and only limited downside risk in the form of firm default it is expected to have no significant influence on firm performance. For salary the following hypothesis was formed:

(H2) Salary has no effect on firm performance

Option based compensation is expected to have a positive effect on firm performance. Options can only be exercised when a certain stock price is reached and otherwise are worthless. This implies that option based compensation leads to an incentive to take on more risk and increase firm performance as they bare no downside risk. Former research by Conyon and Murphy (2000) confirms the incentive based nature of option based compensation. In accordance with the optimal contracting approach the following hypothesis was formed:

(H3) Option based compensation has a positive effect on firm performance

Stock based compensation is also expected to have a positive effect on firm performance. This equity based form of compensation make the executive a shareholder and in this way align his interests with those of other shareholders. For stock based compensation the influence is expected to be smaller than the effect of option based compensation. Considering the fact that the amount of shares needed to make the executive a substantial owner of company stock to align his interest with that of other shareholders is large. Yet stock based compensation does provide an incentive to increase firm performance (Ofek & Yermack, 2000). Even though a smaller effect is expected than the effect of option based compensation the following hypothesis was formed:

(10)

Bonus is a performance based compensation. This implies a causal relationship from firm performance on bonus instead of the other way around. However, because this relationship will lead to shared variance between bonus and firm performance, bonus will be taken into account as a covariate. In this way bonus is allowed to explain some of the variance in firm performance even though no hypothesis will be formed about the effect of bonus on firm performance.

3 Methodology

In this Chapter the research method will be explained. First in section 3.1, the regression model and factors used will be explained. Section 3.2 presents the data and thereafter, section 3.3 provides a brief summary of the statistical analyses that will be performed on the data.

3.1 Model

This research makes use of panel data. To analyze whether executive compensation or any of the components that compensations consists of could predict firm performance, a linear regression was conducted that would account for this specific type of data. A simple Ordinary least squares regression would not suit the data, as this analysis assumes independent and identically distributed observations, which is not the case with panel data. In panel data some observations are dependent because they belong to the same timepoint and some observations are dependent because they belong to the same company. This leads to time specific and company specific variance in the data which is possibly associated with the parameters in the model. When these sources of variance are ignored (i.e., omitted from the model), this variance will be absorbed by the error term and as such lead to a correlation between the parameters in the model and the error term. This problem is known as the problem of endogeneity, which leads to a biased and inconsistent estimator. To solve this problem the data have to be pooled. Assuming parameter homogeneity and due to pooling the data across i and t an unbiased estimator can be formulated for pooled regression (Croissant & Millo, 2008).

𝑦𝑖𝑡 = α + 𝛽 ∗ 𝑥𝑖𝑡 + 𝜀𝑖𝑡

To model heterogeneity among firms an individual error component μi is added to the model.

𝑦𝑖𝑡 = α + 𝛽 ∗ 𝑥𝑖𝑡 + 𝜀𝑖𝑡+ 𝜇𝑖

For consistency this individual error component µi cannot have correlation with either Xit or εit.

Since εit is assumed to be well-behaved and independent of Xit and µi there cannot be correlation

(11)

of inconsistent estimators due to this possible correlation the First-differences model is introduced. The First-differences model focusses on the first differences over time for all variables. This implies that ΔXit =Xit - Xi,t-1 , ΔYit =Yit - Yi,t-1 and Δεit it - εi,t-1.

∆𝑦𝑖𝑡 = β ∗ ∆𝑥𝑖𝑡 + ∆𝜀𝑖𝑡

Note that due to the fact that the differences over time are examined α and µi are dropped from

the model since neither of them varies over time (Croissant & Millo, 2008). Dropping these variables of course implies that any possible correlation between µi and the other regressors

and thus inconsistent estimators is eliminated.

The regression models used in this paper, to examine the effect of executive compensation on firm performance, are the Pooled linear regression model and the First-differences model. First, Both a pooled regression and a first difference regression are performed on the influence of (7)Total compensation on (14)Total shareholders return. In both regression models excess returns were used for performance by deducting (15)Market performance for each period. Natural logarithms are introduced to normalize the distribution of these variables and to get more accessible results. The Pooled linear regression model and the First-differences model are given below.

Pooled linear regression model:

(1) (𝑇𝑆𝑅𝑖𝑡− 𝑀𝑃𝑡) = α + 𝛽7∗ ln (𝑥7𝑖𝑡 Tc) + 𝜀𝑖𝑡

First-differences model*:

(2) ∆(𝑇𝑆𝑅𝑖𝑡− 𝑀𝑃𝑡) = 𝛽7∗ ln (∆𝑥7𝑖𝑡 Tc) + ∆𝜀𝑖𝑡

First-differences model* ΔXit =Xit - Xi,t-1 , ΔTSRit =TSRit - TSRi,t-1 and Δεit it - εi,t-1

Where:

α : intercept

βn : corresponding coefficient TSR : (14) Total Shareholders Return MP : (15) Market Performance X7Tc : (7) Total compensation ε : disturbance term

(12)

After these first two regressions again both a pooled regression and a first difference regression are performed are performed to test the influence on (14)Total shareholders return of some of the components of total compensation namely (1)Salary, (2)Bonus, (3)Value of options granted and (4)Fair value of stock awarded. Again regression on both the Pooled linear regression model and the First-differences model will be performed to examine the effects on (14)Total shareholders return of these components. Again excess returns were used. Natural logarithms are introduced to normalize the distribution of these variables and to get more accessible results. The models used for regression are given below.

Pooled linear regression model: (3) ( 𝑇𝑆𝑅𝑖𝑡− 𝑀𝑃𝑡) = α + 𝛽1∗ ln (𝑥1𝑖𝑡 S) + 𝛽2∗ ln (𝑥2𝑖𝑡 B) + 𝛽3∗ ln (𝑥3𝑖𝑡 Op) + 𝛽4∗ ln (𝑥4𝑖𝑡 St) + 𝜀𝑖𝑡 First-differences model*: (4) ∆(𝑇𝑆𝑅𝑖𝑡− 𝑀𝑃𝑡) = 𝛽1∗ ∆ln (𝑥1𝑖𝑡 S) + 𝛽2∗ ∆ln (𝑥2𝑖𝑡 B) + 𝛽3∗ ∆ln (𝑥3𝑖𝑡 Op) + 𝛽4∗ ∆ln (𝑥4𝑖𝑡 St) + ∆𝜀𝑖𝑡

First-differences model* ΔXit =Xit - Xi,t-1 , ΔTSRit =TSRit - TSRi,t-1 and Δεit it - εi,t-1

Where:

α : intercept

βn : corresponding coefficient TSR : (14) Total Shareholders Return MP : (15) Market Performance X1S : (1) Salary

X2B : (2) Bonus

X3Op : (3) Value of Options granted X4St : (4) Fair value of Stock awarded ε : disturbance term

(13)

In the previous models absolute values for (1)Salary, (2)Bonus, (3)Value of options granted and (4)Fair value of stock awarded were used. According to the optimal contracting approach executive compensation is constructed by the board of directors as an instrument to generate an incentive to the executive to optimize firm performance (Jensen and Murphy, 1990b). Each of the components mentioned above have a different influence on the type and magnitude of the incentive given to executives. To look at the influence of the structure of total compensation on firm performance the four components were taken as a share of (7)Total compensation in percentage. The following two regressions test the influence of (10)Percentage Salary of Total compensation, (11)Percentage Bonus of Total compensation, (12)Percentage Options granted of Total compensation, (13)Percentage Fair value of Stock awarded of Total compensation on (14)Total shareholders return. similarly as described above, excess returns were used.

Pooled linear regression model: (5) (𝑇𝑆𝑅𝑖𝑡− 𝑀𝑃𝑡)

= α + 𝛽10∗ 𝑥10𝑖𝑡 PS + 𝛽11∗ 𝑥11𝑖𝑡 PB + 𝛽12∗ 𝑥12𝑖𝑡 POp + 𝛽13∗ 𝑥13𝑖𝑡 PSt + 𝜀𝑖𝑡 First-differences model*:

(6) ∆(𝑇𝑆𝑅𝑖𝑡− 𝑀𝑃𝑡)

= 𝛽10∗ ∆𝑥10𝑖𝑡 PS + 𝛽11∗ ∆𝑥11𝑖𝑡 PB + 𝛽12∗ ∆𝑥12𝑖𝑡 POp + 𝛽13∗ ∆𝑥13𝑖𝑡 PSt + ∆𝜀𝑖𝑡 First-differences model* ΔXit =Xit - Xi,t-1 , ΔTSRit =TSRit - TSRi,t-1 and Δεit it - εi,t-1

Where:

α : intercept

βn : corresponding coefficient TSR : (14) Total Shareholders Return MP : (15) Market Performance

X10PS : (10) Percentage Salary of Total compensation X11PB : (11) Percentage Bonus of Total compensation

X12POp : (12) Percentage Options granted of Total compensation

X13PSt : (13) Percentage Fair value of Stock awarded of Total compensation ε : disturbance term

(14)

The pooled linear regression models for the effect of (7)Total compensation and the percentage effects of the four components of total compensation ((10), (11), (12), (13)) will both be controlled for (16)Equity holdings among executives. This is only done for the pooled linear regression models and apart from the earlier regressions because of less available data. The model including (16)Equity holdings is as follows.

Pooled linear regression models:

(7) (𝑇𝑆𝑅𝑖𝑡− 𝑀𝑃𝑡) = α + 𝛽7∗ 𝑥7𝑖𝑡 Tc + 𝛽16∗ 𝑥16𝑖𝑡 Eh + 𝜀𝑖𝑡 (8) (𝑇𝑆𝑅𝑖𝑡− 𝑀𝑃𝑡) = α + 𝛽10∗ 𝑥10𝑖𝑡 PS + 𝛽11∗ 𝑥11𝑖𝑡 PB + 𝛽12∗ 𝑥12𝑖𝑡 POp + 𝛽13∗ 𝑥13𝑖𝑡 PSt + 𝛽16 ∗ 𝑥16𝑖𝑡 Eh + 𝜀𝑖𝑡 Where: α : intercept βn : corresponding coefficient TSR : (14) Total Shareholders Return MP : (15) Market Performance X7Tc : (7) Total compensation

X10PS : (10) Percentage Salary of Total compensation X11PB : (11) Percentage Bonus of Total compensation

X12POp : (12) Percentage Options granted of Total compensation

X13PSt : (13) Percentage Fair value of Stock awarded of Total compensation X16Eh: (16) Equity holdings

ε : disturbance term

Executive compensation is expected to have an ongoing effect after the timepoint this compensation is received by the executive. To test whether the effects of total compensation and the four components of total compensation on frim performance are also affecting the frim performance one year later a lag-test will be performed on the pooled linear regression models for model (1), (3) and (5). For the lag-test all explaining variables will be tested on the explained variable for timepoint t+1, indicating the lag will consist of one time period (year). Because the number of observations is not that large (n=198) and since the first difference method also removes one time period, the lag-test was only performed on the pooled linear regression models. Assuming a lagged effect one time period later the lagged tests will also contribute to the robustness of the results.

(15)

All models are tested for CEO’s and CFO’s separately. These two groups are tested separately due to the fact that the decisions made by CEO’s might have a different level of impact on firm performance than that of CFO’s (Yermack, 1995).

3.2 Data

To examine the influence of executive compensation on firm performance, two corresponding measurable factors for executive compensation and firm performance had to be found. For executive compensation this paper focusses on firms in the United States (US). To get a varied sample of the US market the Dow Jones Industrial Average (DJIA30) was used, consisting of 30 firms. Data on the various components of the compensations received by executives of firms listed on the DJIA30 were found in the EXECUCOMP database of WRDS. Data of these companies were retrieved for a time period of 6 years from 2007 to 2014. The components of compensation taken from this database were (1) Salary, (2) Bonus, (3) Value of options granted, (4) Fair value of stock awarded. All values for these components of total compensation are in US$. For option value the grant date fair value was used. The change in value of the option when exercised is thus left out to rule out the possibility of a bidirectional relationship between option value and firm performance, and make sure that this relationship is only unidirectional (from option value to firm performance and not vice versa) (Yermack, 1995). Other data retrieved from EXECUCOMP were (5) Full name executive, (6) Company name, (7) Total compensation, (8) Executives age and (9) Fiscal year. Also variables were added for the value of the earlier mentioned components of Total compensation in percentage. (10) Percentage Salary of Total compensation, (11) Percentage Bonus of Total compensation, (12) Percentage Options granted of Total compensation, (13) Percentage Fair value of Stock awarded of Total compensation. All values for these components of total compensation were retrieved in US$ and divided by Total compensation.

Firm performance is measured by (14) Total shareholders return (TSR). For TSR, data on stock prices were retrieved from Datastream (datatype: Total Return Index). The Total Return Index shows a theoretical growth in value of a share-holding over a specified period, assuming that dividends are re-invested to purchase additional units of stock at the closing price on the ex-dividend date. Here the same time period was used from 2007 to 2014.

(16)

The formula used to calculate accumulated total shareholders return is given below. 𝑅𝐼𝑡 = 𝑅𝐼𝑡−1∗ 𝑃𝐼𝑡 𝑃𝐼𝑡−1∗ (1 + 𝐷𝑌𝑡 𝑅𝐼𝑡−1∗ 1 𝑁) Where:

RIt = Return Index at time t RIt-1 = Return Index at time t-1 PIt = Price Index at time t PIt-1 = Price Index at time t-1 DYt = % Dividend yield at time t

To compare with and control for Market performance, total market returns were deducted from total shareholders return to get excess returns. For (15)Market performance, data on the total market returns (datatype: MSCI Total Return Index) for the S&P500 Financials were retrieved from Datastream. For the control variable (16)Equity holdings data on the percentage shares owned (options inc.) by executives were retrieved from the EXECUCOMP database of WRDS. For equity holdings the data were only available for a shorter time period from 2010 to 2014.

3.3 Summary statistics

For a sample of 60 executives the data described in section 3.2 were taken from the EXECUCOMP database of WRDS. This sample of 60 executives consists of 30 Chief executive officers (CEO’s) and 30 Chief financial officers (CFO’s) from 30 companies listed on the Dow Jones Industrial Average. The Dow Jones Industrial Average is a price-weighted average of stocks of 30 companies listed on the New York Stock Exchange (NYSE) and the Nasdaq. For the data on the executives of these companies a period of eight years was examined from 2007 to 2014. For some companies data were only available until 2013. This indicates an unbalanced panel. For this panel in the eight observed years a total of 450 observations were done on 14 variables. For both CEO’s and CFO’s the number of observations was 225.

For the 60 earlier discussed CEO’s and CFO’s the average yearly total compensation is 11,8 million US$, with a median of 9,7 million US$. Of this total compensation only roughly 1,1 million US$ is salary and the average bonus is 0,86 million US$. Equity based compensation is responsible for the biggest part of total compensation, with 2,5 million US$ and 4,6 million US$ for option and stock based compensation respectively. The remaining 2,74 million US$ consist of pension plans and other smaller types of compensation such as severance payments, debt forgiveness, signing bonuses and life insurance premiums. These types of compensation are not

(17)

taken into account in this paper as no theory on possible relations to firm performance could be formed based on former literature.

For CEO’s the average total compensation is 16,9 million US$ while for CFO’s the average total compensation is 6,8 million US$. This difference is also visible in the four components of total compensation. For salary and bonus CEO’s on average receive almost twice the amount of CFO’s. For equity based compensation the amount CEO’s receive is two to three times higher. On average, CEO’s receive 6,2 million US$ worth of stock and 3,8 million US$ worth of options per year. For CFO’s these amounts are 2,6 million US$ and 1,3 million US$ worth of stock and options respectively. For equity holdings CEO’s exceed CFO’s by nine times the amount. The average amount of equity holdings for CEO’s is 0,27% of total equity of the corresponding company while for CFO’s average equity holdings are 0,03%.

One company was dropped from the dataset resulting in the loss of 16 observations of the initial 450 observations before the models were tested. This was because of an abnormal compensation structure (in this case the executive did not receive any compensation at all due to his substantial ownership of company stock and being a co-founder of the company). In addition, 24 observations were dropped because of missing data.

(18)

4 Results

In this chapter the results of the regressions described in the previous chapter will be discussed. First the regression results in terms of parameters will be reported in tables after which they will be discussed individually as well as their implications based on theoretical background. All tables show results for CEO’s. The conclusions on the results for CEO’s are generalized because the output for CFO’s did not differ in the interpretation of the results. The results on the tests for CFO’s are showed in the appendix.

Table 2: Model (1) and (3a) show the output of the Pooled linear regression model with firm performance as explained variable. Model (2) and (4) show the output for the First differences model with firm performance as explained variable. All explaining variables are in absolute amounts. The first number is the regression coefficient while the second number indicates the p-value. *, **, *** indicate levels of 10%, 5% and 1 % significance respectively.

CEO Model (1) Model (2) Model (3a) Model (4)

Intercept - (α) 0,137 (0,8106) -0,017 (0,8010) 0,439 (0,4939) -0,024 (0,7412) (7)ln(Total compensation) (β7) 0,002 (0,9789) 0,081(0,5017) (1)ln(Salary) (β1) -0,047 (0,6099) -0,047 (0,8365) (2)ln(Bonus) (β2) -0,006 (0,6048) -0,010 (0,6860) (3)ln(Options) (β3) 0,009 (0,4497) 0,002 (0,9311) (4)ln(Stock) (β4) 0,002 (0,8809) 0,014 (0,6173) Observations 198 198 198 198 Adjusted R2 0,000 0,003 0,007 0,003

Based on the output shown in table 2 for Model (1) and (2) no significant results were found on the relation between total compensation and firm performance. This means that neither in the

(19)

pooling regression nor in the first difference regression, the total compensation was a significant predictor of firm performance. This is in contradiction to Murphy’s (1985) findings and the first hypothesis (H1) and seems to support Jensen & Murphy’s (1990b) later findings.

This also indicates that these results do not support the optimal contracting approach. Note that these results were for 30 companies listed on the Dow Jones Industrial Average for the time period from 2006 to 2013. This is a relatively short period and since it also comprises the financial crisis that started in 2008 it is hard to compare to the earlier findings of Murphy (1985) and Jensen and Murphy (1990b). Also because of a small sample size the outcomes might have little statistical power.

Model (2) and (3) in table 2 again show the output of the Pooled linear regression model and the First differences model respectively. In these regressions the influence of absolute measurements of salary, bonus, options and stock were tested on firm performance. For all of these variables both models did not produce significant results on the relation with firm performance. The test results on salary show that absolute salary does not predict firm performance. These findings support the hypothesis (H2) that salary has no effect on firm

performance. Results on option based compensation and stock based compensation show that these two variables are not significant predictors of firm performance. This is in contrast to the findings of Conyon and Murphy (2000) and does not support the hypotheses on option based compensation (H3) and stock based compensation (H4) was found. Together the results found

on total compensation, salary and option based compensation and stock based compensation provide no evidence for the optimal contracting approach.

(20)

Table 3: Model (5a) shows the output of the Pooled linear regression model with firm performance as explained variable. Model (6) shows the output for the First differences model with firm performance as explained variable. The total compensation variable is in absolute amounts. All other explaining variables are percentages of total compensation. Model (7) and (8) show the output for Pooled linear regression models with firm performance as explained variable controlled for equity holdings. The first number is the regression coefficient while the second number indicates the p-value. *, **, *** indicate levels of 10%, 5% and 1 % significance respectively.

CEO Model (5a) Model (6) Model (7)a Model (8)a

Intercept (α) 0,336 (0,1224) -0,013 (0,8608) 1,811 (0,3205) 0,438 (0,2250) (7)ln(Total compensation) (β7) -0,172 (0,3560) (10)Percentage Salary (β10) -0,274 (0,5517) -0,844 (0,3463) -0,315 (0,8238) (11)Percentage Bonus (β11) -0,266 (0,4347) -0,584 (0,4199) -0,417 (0,4981) (12)Percentage Options (β12) -0,150 (0,6651) -0,366 (0,5485) -0,302 (0,6128) (13)Percentage Stock (β13) -0,279 (0,3433) -0,583 (0,2867) -0,481 (0,3162) (16)Equity holdings (β16) -0,133 (0,3556) -0,012 (0,9482) Observations 198 198 122 122 Adjusted R2 0,008 0,010 0,007 0,012

a: smaller dataset used due to less available data

Model (5) and (6) test the effect of the structure of total compensation on firm performance. These tests are performed to test whether the proportions of salary, bonus, options and stock relative to total compensation are significant predictors of firm compensation. Table 3 shows that again no significant results were found implying that the structure of the total compensation does not have an effect on firm performance. The proportion of total compensation that was salary did not significantly predict the firm performance, supporting the

(21)

hypothesis on salary (H2). Also the proportion bonus, proportion options and proportion stock

of total compensation did not significantly predict firm performance. Thus again no evidence was found to support the hypotheses (H3) and (H4). Also the optimal contracting approach is not

supported by these results. For Model (5) and (6) the same sample was used as for previous models implying that again a note must be added on the statistical power of these tests.

The output for model (7) and (8) in table 3 show the results of firm performance regressed on total compensation and regressed on the proportions of the four separate components relative to total compensation controlled for equity holdings by the executives.Output for both models show that adding equity holdings as a predictor in the model does not change the results; none of the predictors in model (7) and (8) is significant. Accordingly, the same conclusions with respect to the optimal contracting approach will be drawn as was done for the output in table 2. Interestingly, it should be noted that the finding that equity holdings by executives is no significant predictor of firm performance is in contradiction with former findings of Mehran (1995). Because of the shorter time period used for the dataset on equity holdings this regression was performed separately form the earlier tested models (1) and (5a). Note that this smaller number of observations in the analysis leads to a decrease of statistical power. This will be discussed further in the discussion section.

(22)

Table 4: Model (1b), (3b) and (5b) show the output for the lagged pooled linear regression models with firm performance as explained variable. The lag consists of one time period. For model (1b) and (3b) the explaining variables are absolute amounts. For model (5b) the explaining variables are percentages of total compensation. The first number is the regression coefficient while the second number indicates the p-value. *, **, *** indicate levels of 10%, 5% and 1 % significance respectively.

CEO Model (1b) Model (3b) CEO Model (5b)

Intercept (α) 0,898 (0,1741) 0,306 (0,6716) Intercept (α) 0,183 (0,4662) (7)ln(Total compensation) (β7) -0,076 (0,2684) (1)ln(Salary) (β1) -0,018 (0,8640) (10)Percentage Salary (β10) 0,190 (0,7327) (2)ln(Bonus) (β2) -0,007 (0,6256) (11)Percentage Bonus (β11) -0,216 (0,5739) (3)ln(Options) (β3) 0,009 (0,5140) (12)Percentage Options (β12) -0,030 (0,9413) (4)ln(Stock) (β4) -0,006 (0,7280) (13)Percentage Stock (β13) -0,081 (0,8124) Observations 171 171 Observations 171 Adjusted R2 0,007 0,006 Adjusted R2 0,004

The output in table 4 shows that also when we perform a lag-test in which firm performance at timepoint t+1 is regressed on total compensation at timepoint t, total compensation is not a significant predictor of firm performance.A similar lag test was performed with salary, option based compensation and stock based compensation as well as the proportions of these components relative to total compensation as predictors. Both test indicate that none of these components at timepoint t is a significant predictor of firm compensation at timepoint t+1. For salary these results are in accordance with expectations (H2). Yet, no support was provided for

the hypotheses on option based compensation (H3) and stock based compensation (H4).

In summary, the results do not support the hypotheses on the influence of executive compensation on firm performance. It can therefore be concluded that these results do not provide support for the optimal contracting approach

(23)

5 Conclusion & discussion

In this chapter a conclusion will be given based on the findings and results of the previous chapter and possible shortcomings and suggestions for future research will be discussed.

In this paper the influence of executive compensation on firm performance was tested. Total compensation as well as four components of total compensation, namely salary, bonus, option based compensation and stock based compensation, were used as measures for executive compensation. The tests were performed on a sample of 30 CEO’s and 30 CFO’s of 30 companies listed on the Dow Jones Industrial Average from 2007 to 2014. No evidence was found to support the hypothesis that total compensation would predict firm performance (H1). Also for

salary, option based compensation and stock based compensation no support was found for the hypothesis that these separate components of compensation would predict firm performance. Also notable is an absence of the effect of equity holdings among executives on firm

performance. All these results are in contrast with the optimal contracting approach on the influence of executive compensation on firm performance. In this way, the results in this study contradict with former research on this topic, that was discussed in the literature section. Because the data used and tests performed in this paper differ from those in earlier literature, one should be careful with drawing hard conclusions on the comparison between results of this study and former research. Nevertheless it may be interesting to discuss some possible

explanations for the conflicting results found in the current study. For example these results might suggest that the optimal contracting approach does not explain the current trends in executive compensation implying that current trends in executive compensation may better be explained by the managerial power approach or the perceived cost approach mentioned in theoretical background section. Another possible explanation is that the years taken into

account into this study overlap with the time of the financial crisis. As a result of this, it might be the case that multiple external variables influenced the firm performance, that were not taken into account in this study. Due to the chaotic period caused by the financial crisis, firm

performance might form a less consistent trend over years, making it harder to explain its variance with the models studied in this paper. It would therefore be interesting to study the effect of the financial crisis on the relation between executive compensation and firm

performance. For example, one could examine how variables influential in the economic crisis function as a moderator in the effect of compensation on firm performance.

Also a viable explanation for the conflicting results found in the current study could be that the incentive based nature of executive compensation has reduced significantly over the past two

(24)

decades due to the significant increase in executives’ wealth in this time period. The increase in executives’ wealth could possibly indicate that the level of executive compensation needed to influence the decision making of an executive by providing an incentive should be substantially higher.

However, in addition to these cautious interpretations of the results it is also possible that the hypotheses suggested by the optimal contracting approach are actually true but that this study was not able to detect this (possibly small (Jensen & Murphy, 1990a)) effect, due to shortcomings of this study. For example, it is possible that the statistical power of this study was not sufficiently high to detect an effect. Statistical power is the probability of rejecting the null hypothesis when the null hypothesis is false. Power depends on the number of observations (the more observations the higher the power) and the effect size (the larger the effect the larger the power). When there is an effect but the effect is not that large, many observations are required to actually detect the effect. This study had a maximum of 198 observations (for the first differences method, lag test and the test controlling for equity holdings even less observations were used in the analyses). which is not that many. For that reason it is suggested that in further research more observations will be included in the analyses.

Another suggestion for further research closely related to this is that this study included only maximally eight years as timepoints. Due to this fact performing lagged regressions was only to limited extent possible. When including more timepoints the possibilities for testing a lagged influence of executive compensation on firm performance increase. Finally it might be interesting to look deeper into the perceived cost approach or the managerial power approach. For example research can be done on the influence of the board of directors on executive compensation as CEO’s and CFO’s themselves are often members of board, indicating that they have a substantial influence on their own compensation.

(25)

References

Aggarwal, R., & Samwick, A. A. (1998). The other side of the tradeoff: The impact of risk on

executive compensation (No. w6634). National Bureau of Economic Research.

Agrawal, A., & Mandelker, G. N. (1987). Managerial incentives and corporate investment and financing decisions. Journal of Finance, 823-837.

Bebchuk, L. A., & Fried, J. M. (2003). Executive compensation as an agency problem (No. w9813). National Bureau of Economic Research.

Bebchuk, L. A., Fried, J. M., & Walker, D. I. (2001). Executive compensation in America: optimal

contracting or extraction of rents? (No. w8661). National Bureau of Economic Research.

Conyon, M. J., & Murphy, K. J. (2000). The prince and the pauper? CEO pay in the United States and United Kingdom. Economic Journal, F640-F671.

Croissant, Y., & Millo, G. (2008). Panel data econometrics in R: The plm package. Journal of

Statistical Software, 27, 1-43.

Hall, B. J., & Liebman, J. B. (1997). Are CEOs really paid like bureaucrats? (No. w6213). National bureau of economic research.

Jensen, M. C., & Murphy, K. J. (1990a). Performance pay and top-management incentives. Journal

of political economy, 225-264.

Jensen, M. C., & Murphy, K. J. (1990b). CEO incentives: It's not how much you pay, but how.

Mehran, H. (1995). Executive compensation structure, ownership, and firm performance. Journal of financial economics, 38(2), 163-184.

Murphy, K. J. (1985). Corporate performance and managerial remuneration: An empirical analysis. Journal of accounting and economics, 7(1), 11-42.

Murphy, K. J. (2002). Explaining executive compensation: Managerial power versus the perceived cost of stock options. The University of Chicago Law Review, 847-869.

Ofek, E., & Yermack, D. (2000). Taking stock: Equity‐based compensation and the evolution of managerial ownership. The Journal of Finance, 55(3), 1367-1384.

Yermack, D. (1995). Do corporations award CEO stock options effectively? Journal of financial

(26)

Appendix

Table 5: Model (1) and (3a) show the output of the Pooled linear regression model with firm performance as explained variable. Model (2) and (4) show the output for the First differences model with firm performance as explained variable. All explaining variables are in absolute amounts. The first number is the regression coefficient while the second number indicates the p-value. *, **, *** indicate levels of 10%, 5% and 1 % significance respectively.

CFO Model (1) Model (2) Model (3a) Model (4)

Intercept - (α) 0,868 (0,2233) -0,024 (0,7371) 1,098 (0,1380) -0,015 (0,8366) (7)ln(Total compensation) (β7) -0,082 (0,3139) 0,170 (0,2164) (1)ln(Salary) (β1) -0,149 (0,1751) -0,156 (0,4894) (2)ln(Bonus) (β2) -0,005 (0,6543) 0,015 (0,6436) (3)ln(Options) (β3) 0,005 (0,7307) 0,001 (0,9816) (4)ln(Stock) (β4) 0,004 (0,8265) 0,029 (0,3990) Observations 198 198 198 198 Adjusted R2 0,005 0,009 0,012 0,009

(27)

Table 6: Model (5a) shows the output of the Pooled linear regression model with firm performance as explained variable. Model (6) shows the output for the First differences model with firm performance as explained variable. The total compensation variable is in absolute amounts. All other explaining variables are percentages of total compensation. Model (7) and (8) show the output for Pooled linear regression models with firm performance as explained variable controlled for equity holdings. The first number is the regression coefficient while the second number indicates the p-value. *, **, *** indicate levels of 10%, 5% and 1 % significance respectively.

CFO Model (5a) Model (6) Model (7)a Model (8)a

Intercept (α) 0,476 (0,0769)* -0,014 (0,8475) 2,277 (0,1194) 0,635 (0,1897) (7)Total compensation (β7) -0,247 (0,1413) (10)Percentage Salary (β10) -0,475 (0,5419) -1,958 (0,0562)* 0,160 (0,9299) (11)Percentage Bonus (β11) -0,370 (0,2991) -0,644 (0,4123) -0,484 (0,4212) (12)Percentage Options (β12) -0,407 (0,3085) -0,774 (0,3281) -1,103 (0,1410) (13)Percentage Stock (β13) -0,405 (0,1926) -0,664 (0,2379) -0,761 (0,1434) (16)Equity holdings (β16) 0,366 (0,7425) -0,055 (0,9608) Observations 198 198 121 121 Adjusted R2 0,011 0,022 0,018 0,026

(28)

Table 7: Model (1b), (3b) and (5b) show the output for the lagged pooled linear regression models with firm performance as explained variable. The lag consists of one time period. For model (1b) and (3b) the explaining variables are absolute amounts. For model (5b) the explaining variables are percentages of total compensation. The first number is the regression coefficient while the second number indicates the p-value. *, **, *** indicate levels of 10%, 5% and 1 % significance respectively.

CFO Model (1b) Model (3b) CFO Model (5b)

Intercept (α) 2,489 (0,0018)*** 1,199 (0,1500) Intercept (α) -0,031 (0,9173) (7)ln(Total compensation) (β7) -0,267 (0,1055) (1)ln(Salary) (β1) -0,130 (0,2905) (10)Percentage Salary (β10) 1,574 (0,0659)* (2)ln(Bonus) (β2) -0,011 (0,4160) (11)Percentage Bonus (β11) -0,050 (0,8988) (3)ln(Options) (β3) -0,001 (0,9758) (12)Percentage Options (β12) -0,022 (0,9610) (4)ln(Stock) (β4) -0,020 (0,3306) (13)Percentage Stock (β13) -0,044 (0,8976) Observations 171 171 Observations 171 Adjusted R2 0,049 0,018 Adjusted R2 0,033

Referenties

GERELATEERDE DOCUMENTEN

Dat zou een belangrijke stap zijn, want de student-onderzoekers tonen aan dat buurtbewoners die contact hebben met studenten zich veiliger voelen in de buurt dan degenen die

Aan de hand van het vraagstuk: “waar dreigen archeologi- sche monumenten in de bodem – de boot in de illustratie - te worden beschadigd door schommeling van grondwa- terstanden?”,

To test the fourth hypothesis, The positive relationship between the amount of social media platforms on the obtaining of the funded aim, funded total and the amount of

Using video analyzed from a novel deception experimen t, this paper introduces computer vision research in progress that addresses two critical components to

Economic performance is defined as income, whereas artistic performance is set up according to the selection system theory, divided in market, peer and expert performance.. This

The increased use of equity-based incentive plans, which already inherit a performance measure by nature, namely share price performance, has increased the importance

The results show a negative significant coefficient on environmental and social performance, meaning that firms with high levels of integrated thinking generally

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