The relation between CEO compensations and performances of USA listed companies

(1)

Master Thesis

The relation between CEO compensations and performances of

USA listed companies

Name: Antonio Hilhorst Student number: 10441999

E-mail: antonio.hilhorst@student.uva.nl University: University of Amsterdam

Program: Master Accountancy and Control Field: Control

Number of credits: 15

Supervisor: Georgios Georgakopoulos Second Supervisor Sanjay Bissessur

Date: June 21th 2014

Abstract

The relationship between CEO compensation and company performances is a topic which was and is very popular under researchers, politician and public. There are studies who found a positive relation between CEO compensation and company performances and there are also studies who found a negative or non relation. This study will investigate whether the total compensations of CEO’s are not proportional to the performances of the companies before the crisis and during the crisis. Finally it will compare the results of this study with the results of the study of Mahmoud et al (2008) and other studies. The sample used data from Compustat and ExecuComp. The total sample size contains 833 companies from 25 industries listed on the USA stock exchange during the whole research period. I use regression analyses to test my hypothesis. The results of my study show that we cannot support my hypothesis namely that total compensations of CEO’s are not proportional to the performances of the companies.

Furthermore, the results will be compared with the results of Mahmoud et al (2008) and other researches. It appears that the result of this study is in line with the papers of Mahmoud et al (2008), Lilling (2006), Hogan et al (1998), and other papers A limitation of this research is that it cannot explain how much influence the specific events had on the results of this research. It could be very interesting to analyze what the effect of SOX and, for example, the bankruptcy of Lehman Brothers had on the relation between CEO compensation and performances. Furthermore, there is a lack of independent variables in the formulas of this research.

(2)

Content

1. Introduction ... 3

2. Literature review and hypotheses ... 6

2.1 CEO compensation, agency theory and the managerial power theory ... 6

2.2 CEO compensation and company performance ... 7

2.3 CEO compensation and company size ... 8

2.4 CEO compensation in different industries ... 9

2.5 Hypothesis ... 10 3. Research methodology ... 11 3.1 Research period ... 11 3.2 Data source ... 11 3.3 Measurement of variables ... 12 3.4 Research method ... 12

3.5 The sample size ... 14

3.6 Descriptive statistics ... 16

3.7 Correlations ... 22

4 Results ... 26

4.1 Results hypothesis period 2002-2012 ... 26

4.2 Result hypothesis period 2002-2007 ... 27

4.3 Result hypothesis period 2008-2012 ... 29

Conclusion ... 34

References ... 36

(3)

1. Introduction

CEO compensations is becoming more and more a subject in which the public is interested in. In the public opinion, compensations are becoming extraordinary and politicians want to make rules to lower the compensations (Dillon, 2009). All these facts influences the way compensations are set.

But the traditional accounting says to achieve compliance behaviour between a CEO and shareholders to act in the best interest for the company depends on the percentage of ownership of the CEO. Furthermore, the compensations are future measures like granting stocks, stock options and other equity incentives, which are very important to stimulate the CEO to act in the best interest for shareholders, namely increasing the company value (Jensen & Meckling, 1976). By giving these incentives this should happen.

In past years the researchers also paid intention to this subject. There are a lot of studies about the relation between CEO compensations and company performances. These studies focused largely on large listed companies in the USA (Carpenter and Sanders 2002; Brick et al. 2006; Lilling 2006). Interesting is that none of those researches came with the same conclusions. They all came with different results. The study of Carpenter (2002) concluded that there is not a direct influence of CEO compensation on company

performance, but that the influence is indirect. The evidence of the study of Brick et al. (2006) suggests that excessive CEO compensation has an effect on firm performance. Another study found a strong relation between CEO compensation in relation to company performance (Lilling, 2006).

Outside the USA this research also took place in other parts of the world, for instance in Japan where they also looked at large listed companies (Kaplan 1994; Kato and Kubo 2006). Kaplan found that CEO turnover is less sensitive to poor stock performance, but also to poor sales growth.Kato and Kubo’s (2006) study tends to support the general perception that Japanese CEO compensations are less sensitive to stock market performances and concludes that the bonus system in Japan makes CEO compensations more responsive to firm

performances.

Some researchers compared this subject between nations. For the USA, it was done in comparison with Japan (Kaplan, 1994). There are much research done about CEO

compensations tied to company performances. The latest research I could find about this subject (which includes all kinds of industry) is the research of Mahmoud et al (2008). It was published in 2008 and used data for the period 1996–2002 (Mahmoud et al 2008).

(4)

Based on this information, the following research question arises: What is the relation between CEO compensations and the performances of USA listed companies?

The motivation for my research is that there hasn’t been done any research of this kind in the last 6 years. The data that has been used by Mahmoud et al (2008) are from the period 1996-2002, so this research is outdated. Since 2002 many changes occurred in the economic and financial world. Moreover, there is also a change in companies listed now and in the research period of Mahmoud et al (2008). It is interesting to see if there has been a change. I think that there will be a change, because of the developments in the economic environment. By doing this research I add new information to the research of CEO compensation and performances.

By using a period of the previous11 years we can look what impact the economic crisis, SOX rules, and other factors had on the listed companies, and also whether

compensations changed, referred to performances. By doing this research I can provide updated information for further research.

The scientific relevance of this research is to investigate how the total compensation has developed during the past 11 years. This research contributes to extant literature by using more up to date data from 2002-2012. Prior research has been outdated (latest one published in 2008) and most of the research has been done in a positive economic climate. It is

interesting to see how the total compensation has developed during the economic- and financial crisis. By looking at a pre crisis period and a crisis period we get a good view how and in what way the total compensation has developed during these periods.

The societal motivation for this research is that on basis of the research information (what is the effect of CEO compensations on company performances) shareholders of USA listed companies can see whether compensations are in line with the performances. If the total compensation is not in line with performances they have a stronger negotiation position on how to build up the total compensation of CEO’s to achieve that the company will perform well, by looking at short- and long term compensation possibilities. If the total compensation is in line with company performances, than there is no reason for discussions about the outraged compensations. In this way the results of this research can help to improve the discussion about CEO compensation compared to the related performances.

The structure of the thesis is as follows: The first part is the literature review and hypothesis. I will use theories from prior research which are important for my thesis and explain my hypothesis. To make it more convenient to read I made use of sub-headings to explain the different theories. The second part is about the methodology. To make it more

(5)

clear I also use sub-headings here where every part of the methodology will be explained. The third part is the analysis and description of the results. The fourth part is the conclusion.

(6)

2. Literature review and hypotheses

In this part of my thesis I will describe some theories and findings of other academic literature which gives an impression about CEO compensation related to different aspects. For the clarity I have made 5 different sub headings in which a part of the subject is elaborated on. The first part covers the main theories related to compensation. The second part is related to CEO compensation and company performance. The third part focuses on CEO compensation and company size. The fourth part focuses on CEO compensations in different industries. In the last part my hypothesis will we defined on basis of the prior literature.

2.1 CEO compensation, agency theory and the managerial power theory

The agency theory of Jensen and Meckling (1976) is well known and can be used in different situations. In this thesis I will use this theory to explain how compensation helps to align the interest of the CEO with the interest of the shareholders. The most common components of CEO pay packages are base salary, annual bonus plan (dependent on accounting

performance), stock options and long term incentive plans (Jensen and Murphy, 1990). The agency theory of Jensen and Meckling (1976), focuses on the different interests and behaviour of the agent (CEO) and the principal (shareholders). Jensen and Meckling (1976) define an agency relationship as a contract which is used to make the agent to act on behalf of the principal, which implies granting a certain level of authority to the agent. They argue that to achieve compliance behaviour between a CEO and shareholders to act in the best interest for the company the compensations, it is needed to use future measures to stimulate the CEO to act in the best interest for shareholders, namely increasing the company value. This is supported by Agrawal and Mandelker (1987), who stated that when top management holds stock options in a company it will reduce management incentive problems, which lead to more alignment between agent and principal. Another research supporting the agency theory is that of Akhigbe et al (1995) who found little support that executive compensation reduces agency costs and improve company value.

Furthermore, compensations help to align the behavior between the CEO and shareholders. There are some dissidents, like Core et al (1999) who found that CEO’s of companies with great agency problems and bad performance demand and get more

compensation than CEO’s of companies who have low agency problems. This is supported by Bebchuk et al. (2002), who came with the managerial power theory. This theory elaborate that executive compensation is not solely seen as a remedy to align the behaviour of CEO and

(7)

shareholders. In the contrary, certain features of executive compensation are seen as part of the problem itself. Under the optimal contracting, the board designs compensation

arrangements exclusively for the purpose to align behaviour between shareholders and

executives. In contrast, under the managerial power approach part of the behaviour problem is that executives use their compensation to provide themselves with rents (Bebchuk et al. 2002).

2.2 CEO compensation and company performance

As stated in the first part, the agency theory mentioned that shareholders have to give future compensation measures to the CEO’s in order to align their interest with that of the

shareholders/company, namely creating company value.Kahn and Scherer (1990) and Banker, Lee and Potter (1996) found that a pay for performance compensation leads to improved company performance, because by attracting, retaining more talented executives and motivating them provide greater congruence between their actions and the companies interests.

Lewellen, Loderer & Martin (1987) found a positive relationship between

compensation structures and company performance. This is also supported by Murphy (1985), Jensen and Murphy (1990) and Joskow and Rose (1994), whereby Jensen and Murphy (1990) also focused on the intensity of the pay-performance relation. Jensen and Murphy compute a pay performance sensitivity of $ 0.00325 for USA listed companies. This means that the CEO compensation increases with $ 3.25 by an increase of $ 1000 shareholders value. Although this shows a weak relation between alignment of executive and shareholders, it can still have a great influence on the behaviour of the CEO, for example when shareholder return is very high. In the research of Lambert et al (1991b) it is shown that when executives see that there is a sufficient chance that the options are going to finish in the money, these executives stock option can increase management risk aversion, so the executives will avoid extreme risks which can harm the company. But there are also studies who did not find any relationship between CEO compensation and company performance, for instance in the research of Zhou et al. (2011). Scheafer (1998) found that larger companies choose smaller pay-performance sensitivities.

Furthermore, the percentage of ownership were also taken into consideration. By giving the CEO stock compensations he will be more inclined to act in the best interest of the company, because he will also gain from it (Jensen and Meckling, 1976). However,

(8)

Himmelberg, Hubbarda and Palia (1999) found an insignificant relations between managerial ownership (stock compensations) and the performance of the company.

Further Fahlenbrach and Stulz (2009) did not find any relation in the bank sector. When CEO compensations where aligned with the interest of the shareholders these companies does not performed better than companies where the compensations were not aligned with shareholders interest. They even found that companies where the compensation was aligned with shareholders interest performed even worse compared to those were there was less alignment. They think this is because CEO’s with alignment of interest take more risk to get profitable projects.

2.3 CEO compensation and company size

The allocation theory of control assumed that in a market equilibrium the most talented executives occupy top positions in the largest companies (Rosen, 1992). Consequently, it is believed that CEO compensation is higher in big companies than in midsize and smaller companies. Prior studies have investigated the relations between CEO compensation and company size. Rosen (1982) argues that the observed relationship between CEO

compensation and company size can be assigned to larger, more complex companies who hire better CEO’s. Barro and Barro (1990) research shows that compensation is strongly

associated with the company size. Core et al (1999) found that there is a relation between total compensation of the CEO and the size of a company. He found that the total compensation was related with investment opportunities, prior performances and company risk, because larger companies have more investment opportunities, can easier be judged on prior performances and have to carry more risk than smaller companies. As a result, larger companies will pay their CEO a higher compensation. This can be interpreted that larger companies demand higher quality of managerial talent, which is in line with the findings of Rosen (1982).

The research of Madura et al (1996) did not find a strong relationship between company performance and CEO pay in small companies. They think this is because of the lack of monitoring by large outside shareholders.

Canarella (2008), for example, finds evidence that the effect of company size on CEO compensation is more significant after the stock market crash in 2000. Lambert et al. (1991a), Murphy (1985) and Subramanian (2012) finds evidence that there is a significant association between CEO compensation and company performance. However Lambert et al. (1991a) also find that the change in CEO compensation is not explained by changes in company size.

(9)

While Lewellen et al. (1987), Clinch (1991), and Gaver et al. (1993) find that a greater use of stock options in growing companies is needed to prolong the managers time horizon for decision making.

Baker and Hall (2004) also investigate what the effect was, and came with the finding that CEO incentives do not fluctuate dramatically with firm size. Furthermore, CEO

compensations are constant or decline slightly as company size increases.

2.4 CEO compensation in different industries

In this section I will use prior literature of Anderson, Banker and Ravindran (2000), who investigated the technology industry and concluded that they used more stock options as compensation, but that the total level of compensation is not higher than those in other industries.

Furthermore, Houston and James (1995) founded that compensation packages are differently structured in the banking industry than in other industries. They find that CEO’s on average receive less cash compensation and less participation in stock option plans. Whereby cash compensation is more sensitive for company performance than in other industries and whereby stock options receivable are a small percentage of their total compensation compared with other industries.

Carroll and Ciscel (1982) investigated the differences in compensation between regulated and unregulated industries. They found that regulation has an effect on the rewards received by CEO’s, namely that in regulated industries the basic annual salary of the CEO is lower than those in the unregulated industries. The possible reasons for this are the reduction of financial risks, compensations are reviewed by the government, and price regulation forced regulated companies to focus on costs which is reflected in CEO’s salary (Carrol and Ciscel, 1982). This is supported by Hubbard and Palia (1995) who found a stronger pay-performance relation in deregulated banks compared to regulated banks.

Arya and Sun (2004) found a significant increase in compensation by a change from salary to bonuses and long term incentive compensation. They found this by comparing the construction of the compensation before and after deregulation.

Another research that investigated the components of CEO compensation is the research of Zhou (2000), whereby he looked at four different kinds of industries. He found that the financial service industry has the highest pay followed by manufacturing, resources, and utilities. Whereby utility is in line with the study of Carrol and Ciscel (1982). Further Zhou made a distinction between four components: salary, bonus, benefits and options.

(10)

Moreover, he looked per industry how these components developed over the research years. The results of his research have shown that the total amount of compensation increased over the years and that the use of options increased and the amount of salary is stable or slightly decreased over the years in all investigated industries.

2.5 Hypothesis

On the base of the literature review and the paper of Mahmoud et al (2008) there has been a lot of discussions about whether CEO compensations and performances are in line with each other. There are studies that support the relation between CEO compensation and

performances, but there are also studies contrary to this finding. My hypothesis is: • The total compensations of CEO’s are not proportional to the performances of the

companies.

For the performance I looked at market based (total shareholders return) and accounting based (net income and ROA) returns. To get an answer on this hypothesis I will investigate whether in the USA CEO compensations are in line with the performances of the companies. The results will also give a view on the development of the relation between compensation and performances over the past 11 years, and if indeed SOX, the crisis and other factors have influenced the results.

The quality of my research is that I come with new, more updated data and results of USA listed companies, which can be helpful for the discussion about CEO

compensations in relation to performances. Furthermore, these results add new information to extant literature.

(11)

3. Research methodology

In this part I will describe in detail how the research is stated like the research period, sample size and other relevant matters. Furthermore, I will explain how the variables are measured and explain the regression models.

3.1 Research period

I have chosen for the years 2002-2012, because in this period of time there is a pre-crisis period, a crisis period, and hopefully an after crisis period, so this can give a good view how the CEO compensations have developed over these years. Furthermore, it is possible to make good time frames like a pre crisis period (2002-2007) and a crisis period (2008-2012). In these two times frames there has happened a lot. For example in the first part: mid 2002 SOX was introduced which had influence on public listed companies. In the years 2005-2006 all companies listed on the US Stock Exchange had to have fully implemented these regulations (SOX, 2005). Another important event in this time period was the beginning of the financial crisis.

During the second time frame the financial crisis was worldwide at its peak. At the end of 2008 Lehman Brothers filed for bankruptcy and the USA government had troubles with the debt ceiling. All these events influenced more or less companies performances and the

compensations of CEO’s. During the second time frame the stock exchange was wildly fluctuating, which influenced the companies.

Compared to prior research, the time frame of this research is longer. The latest research only covered 7 years, whereas I will investigate 11 years. I believe that 11 years can provide get a better view on the relations I want to investigate. Furthermore, we can see if there is a pattern in economic good times and economic bad times.

3.2 Data source

The data approach I have used for this thesis is an archival research, by using data from ExecuComp and Compustat. The data related to CEO compensation are retrieve from Execucomp database. The data regarding the accounting data are retrieved from the data of Compustat. Both databases provide data for the S&P 500, 400 and 600. The software program I use for the calculation is SPSS.

(12)

3.3 Measurement of variables

The measurements I use are compensation and company performance. For the

indication of compensation I use the following sub categories: salary and cash bonus which is a short term performance compensation and stock options which is a long term compensations measure (Hoitash et all, 2012). These compensations are the most common used

compensations in all listed firms. Furthermore, I will look to total compensation, which is defined as the total amount of earnings of a the CEO per year on all different kinds of components. In the total compensation all the compensations named above are taken into account, plus the other benefits earned by the CEO. By using these categories it is possible to compare compensations. Furthermore, by looking at short and long term compensations we can see which type of compensation is mostly used.

For measuring companies performance I will look at accounting based returns and market based returns (Firth, Fung and Rui, 2006), because in general shareholders are most interested in the creation of company value and profit.

For the market based returns I look at the stock returns of every company. The stock return (TRS) is defined as the closing price at fiscal year end plus dividends divided by closing price of the prior fiscal year end (Mahmoud et al, 2008).

For the accounting based measure I used return on assets (ROA), which is consistent with the paper of Mahmoud et al (2008). ROA is defined as net income divided by average total assets (Mahmoud et al, 2008).

Furthermore, I add the accounting based measure net income as independent variable. The reason for this is that shareholder looks at the profitability of a company, besides the returns on stock, before they will invest. It is often seen that when the profits don’t meet the requirement of the targets set by the market, shareholders are disappointed and sale their shares in the company.

Net income is defined as the fiscal period income or loss reported by a company after subtracting expenses and losses from all revenues and gains. (WRDS, 2014).

Standard Industrial Classification Code (SIC) are codes used by the USA government for company operating processes.

3.4 Research method

For the hypothesis I want to investigate whether the compensation is deserved, based on the performance of the company, by testing the relation between each performance measure with

(13)

the total compensation. To test the relations three models are used, which are regression analyses.

The dependent variable used for these models is total compensation and the independent variables are net income, ROA and TRS. Net income is defined as the fiscal period income or loss reported by a company after subtracting expenses and losses from all revenues and gains (WRDS, 2014). Furthermore, ROA and TRS are index numbers and are calculated as follows: TRS is defined as the closing price at fiscal year end plus dividends divided by closing price of the prior fiscal year end (Mahmoud et al, 2008) and ROA is defined as net income divided by average total assets (Mahmoud et al, 2008).

The three regression models are formulated as follow:

Model 1:

Total compensation= α+β (net income) + ε it

Model 2:

Total compensation= α+β (TRS) + ε it

Model 3:

Total compensation= α+β (ROA) + ε it

These regression models are partly based on the general model used in many pay-performance relationship researches, which is:

PAYit= α+β(SIZEit)+γ(PERFORMANCEit) + ε it

Where α stands for constant i stands for company. t for year.

β for the coefficient of the independent variable Payit is the CEO compensation of a company per year.

Performanceit stands for an observable performance measure of a company per year.

ε it is the error term.

(14)

3.5 The sample size

For this research I want to use the data of listed companies in the USA. Therefore I will use the companies who are listed at the S&P 500, the S&P Midcap 400 and the S&P Smallcap 600 indexes.

The S&P 500 is the stock market index for the 500 biggest USA market capitalization companies who are publicly traded on the USA stock market.

The S&P Midcap 400 represents more than 7% of the available USA capital market. The S&P 400 is developed to measure the performance of 400 midsize companies, whereby market segments with distinctive risk and return characteristics are reflected.

The S&P Smallcap 600 represents around 3% of the available USA capital market (Standard and Poors, 2013). By using the S&P 500, 400 and 600 I have a wide range of companies represented in different industries in my data sample.

By adding them together we have 1500 companies for our research. This total will diminish for several reasons, namely:

• It is possible that in the past 11 years new companies came on the stock exchange. To get a good view and good comparison I will exclude all companies which have not been listed for the full last 11 years.

• If data are not available companies will be excluded.

After excluding the samples based on the aforementioned reasons, the total sample size contains 833 companies, which results in a total of 9163 company year observations. Table 1 presents the number of companies that meet these requirement. These companies are used for this research.

With regard to the outliers I have used winsorization. This is one of the method to deal with outliers. This strategy modifies the original dataset, imposing and upper and lower bound on outliers by setting extreme data points to be equal to a specified percentile of the

distribution of values for each variable (NIST, 2003).

All the variables used in this research are winsorized to keep as much valuable

information as possible in the sample, instead of dropping (trimming) these data. By using 90 percent winsorization, the top 5% and bottom 5% are winsorized.

(15)

Table 1 Sample size

Sample selection Number of companies

Starting point of the sample 2002-2012 1500

Minus: companies who were not active for the full period 618

Minus: missing data for study period 49

Final sample size 833

Further I have divided the companies into different industries, based on the two digit Standard Industrial Classification Code (SIC) range of 01-99 as shown in table 2 on page 16. I have chosen to use the same range of industries as the research of Mahmoud et al (2008). As we can conclude from table 2, we see that the service industry is the largest sample with 78 companies and a percentage of 9.4, followed by electrical equipment with a sample size of 72 companies and a percentage of 8.6. The industry with the smallest sample size is the toy manufacturing industry with only 4 companies and a percentage of 0.5, followed by printing and publishing with 7 companies and a percentage of 0.8 and mining with 8 companies and a percentage of 1.

The advantage of this sample is that all kinds of industry are taken into account. Also, it contains all different company sizes like small, medium and large, which is also useful for getting a diversify sample.

As shown in table 3 the S&P 500 companies are the dominating group with 46%, followed by the S&P 400 and S&P 600. This is logical because large companies are more steady than small companies. More specifically, in economic down turn or scandal large companies are in a better condition to carry the burden of bad times than small companies.

By only looking at the companies listed in the whole period 2002-2012 I have selected the companies who have survived the changes in economical and political perspectives.

(16)

Table 2

Industry composition Industries

Two digit

SIC Number of companies Percentage

Mining 10, 12, 14 8 1,0

Gas and oil and petroleum refining 13, 29 33 4,0

Construction 15-17 16 1,9

Food 20, 21, 54, 58 50 6,0

Clothing and footwear 22, 23, 31, 56 26 3,1

Forest product, paper 24, 26 19 2,3

Furniture 25, 57 11 1,3

Printing and publishing 27 7 0,8

Chemicals 28 54 6,5

Rubber, plastic, stone, clay and glass 30, 32 12 1,4

Primary and fabricated metal 33, 34 29 3,5

Industry machinery 35 47 5,6 Electrical equipment 36 72 8,6 Transportation equipment 37 26 3,1 Instruments 38 44 5,3 Toy manufacturing 39 4 0,5 Transportation 40, 42-47 26 3,1 Telecommunication 48 13 1,6 Utilities 49 65 7,8 Wholesale trade 50, 51, 99 32 3,8 Retail trade 52, 53, 55, 59 32 3,8 Banks 60 39 4,7

Insurance, other financial services 61-64, 67, 69 63 7,6

Services 70-79 78 9,4

Healthcare professional services 80, 82, 83, 87 27 3,2

Total 833 100,0

Table 3 Stock exchanges

Listed on exchange Number of companies Percentage

S&P 400 232 28

S&P 500 385 46

S&P 600 216 26

Total 833 100

3.6 Descriptive statistics

In table 4 (page 20) the descriptive statistics of all the dependent and independent variables for the period 2002-2012 are presented. Table 5 gives the descriptive statistics for the first research period 2002-2007. Table 6 contains the descriptive statistics for the second research period 2008-2012. Besides the standard percentiles I have added two other percentiles,

(17)

namely the 5th percentile and the 95th percentile. This is because 90 percent winsorization is applied to all the variables used in this research. Furthermore, the tables contain the number of observations, standard deviation and the mean.

On the overall research period the mean salary is $830,757.-, the bonus is $450,323.-, the value realized on option exercise is $1,901,540.- and total compensation is $6,352,138.-, which are all higher than the median. Comparing the total compensation with the prior literature of Mahmoud et al (2008) the results show that the total compensation is much higher. For the whole research period my results show a total compensation of $6,352,138.- and the total compensation in the paper of Mahmoud et al (2008) shows a total compensation of $4,453,000.

The mean of the net income is in this research $486,397,179. However: the ranges between the 5th and the 95th percentile is very large. The net income for the 5th percentile is $-118,269,600.- and for the 95th percentile $3,064,230,000.-.

The other accounting based returns, the ROA, shows a mean of 0.0558. Comparing this mean with the research of Mahmoud et al (2008) is shows that the ROA of this study is higher than that of Mahmoud et al (2008). In Mahmouds research it is 3.66 percent; in my research it is 5.58 percent.

For the market based return we see that the mean is 1.08. Consistent with the literature of Mahmoud et al (2008) the stock market returns has a higher volatility than the accounting based returns measure ROA. The volatility is measured by the standard deviation.

Table 5 (Page 20) shows the descriptive statistics for the pre crisis period.

For the pre crisis period the descriptive statistics are very different than for the total period. In this period, the mean salary is $771,825.-, which is $58,932 lower than the total period. For the bonus the mean is $740,498.-, which is much higher compared to the total period where the value of the bonus was $450,323. So the bonus in the pre crisis period is $290,175 higher. The mean value realized on option exercise is 2,148,379.-, which is also higher compared to the total period, where the mean is $1,901,540. The total compensation in the pre crisis period is $6,020,483.-, which is lower compared to the total period. The mean of the total compensation in the pre crisis period compared with the mean of Mahmoud et al (2008) is becoming more closer to each other now the difference between my result and their result is only $1,567,483.

The mean of the net income is in the crisis period $447,937,682, which is lower compared to the total period. Further the ranges has shrunken, where the 5th percentile now is $-64,939,600 and the 95th percentile is $2,880,200,000.

(18)

The mean of ROA is 0.0598. Here the result show that ROA is higher than in the total period, but it is also higher than the ROA of the study of Mahmoud et al (2008).

The mean of the market based returns show a slightly increase from 1.08 to 1.09. The volatility for stock market returns are still higher compared with the ROA.

Table 6 (page 21) shows the descriptive statistics for the crisis period.

The table shows some divergences in all the means compared to the total period and the pre crisis period. In the crisis period the salary has grown compared to pre crisis period and the total period. In the crisis period the mean salary of a CEO is $896,861. In the pre crisis period this was $771,825. So in the past 5 years the salary strongly increased. This suggest that CEO have increased their salary to compensate the lower returns from their bonus and stock options. When the results of salary from the crisis period are compared with the total period, the results also show a increase but this is much lower compared with the pre crisis period.

The mean of the bonus is $100,852. Comparing these results with the pre crisis period it shows that it dropped very strongly, around $350,000. When the results are compared with the total period it dropped even stronger (around $640,000) than in the crisis period.

For the value realized on option exercise this is less strongly the case. The mean in the crisis period is $1,617,628. When this result is compared with the pre crisis period it shows a decrease of approximately $500,000. When the result of the crisis period is compared to the total period, the results shows a small decrease of around $300,000. So this suggest that the crisis have some effects on the bonuses and on the value realized on options exercised, because of the dropped returns on these two components.

The mean for the total compensation is in the crisis period $6,789,307. This is higher than the total compensation in the pre crisis period and in the total period. The difference between the crisis and the pre crisis period is around $700,000, which is enormous. When the results are compared with the total period it shows that it is also higher. Only here the increase is less enormous (around $400,000). Which suggest that the crisis had a small influence on the total compensation. When the results of the crisis period are compared with the results of Mahmoud et al (2008), it shows that the difference has grown. Where Mahmoud et al (2008) had a total compensation of $4,453,000, my results shows a total compensation of $6,789,307.

The mean of the net income during the crisis period has increased to $533,028,507, compared to the pre crisis period where the net income was $447,937,682. If the net income results are compare with the total period, it shows that the net income ($486,397,178) still have increased but that it is little strong compared to the pre crisis period. Furthermore, the

(19)

ranges has grown compared to both time periods, where the 5th percentile now is $-207,192,800 and the 95th percentile is $3,315,000,000.

The other accounting based returns, the ROA, shows a mean of 0.0507. Comparing this mean with other time periods it shows a small decrease. Compared with the research of Mahmoud et al (2008) is shows that the ROA of this study is higher than the research of Mahmoud et al (2008), but compared with other periods it is closer to the mean of Mahmoud et al (2008). In Mahmouds research it is 3.66 percent; in my research it is 5.07 percent.

For the stock market return the mean is 1.07, which is lower than in the pre crisis period and the total period where it was 1.09 and 1.08. Still the volatility of the stock market returns is much higher compared with the ROA.

(20)

Table 4

Descriptive statistics period 2002-2012

Mean Std. Deviation Skewness Kurtosis

Percentiles 5 25 50 75 95 Salary $830,757.00 $308,645.69 0.385 -0.502 $345,000.00 $598,269.00 $803,087.00 $1,000,000.00 $1,500,000.00 Bonus $450,323.24 $755,273.41 1.806 2.202 $0.00 $0.00 $0.00 $613,600.00 $2,662,383.30 Stock options $1,901,540.58 $3,746,328.57 2.232 3.878 $0.00 $0.00 $0.00 $1,667,513.00 $13,870,326.20 Total Compensation $6,352,138.23 $6,357,001.97 1.527 1.476 $629,164.90 $1,814,042.00 $3,937,547.00 $8,331,404.00 $24,095,343.90

Net income (Loss) $486,397,178.72 $805,455,585.88 2.184 3.872 $-118,269,600,00 $40,375,000.00 $144,600,000.00 $510,865,000.00 $3,064,230,000,00

TRS 1.0805 0.31660 0.200 -0.436 0.5168 0.8651 1.0753 1.2738 1.7441

ROA 0.0558 0.05327 0.165 -0.391 -0.0494 0.0195 0.0509 0.0909 0.1629

Table 5

Percentiles 5 25 50 75 95 Salary $771,825.32 $295,758.66 0.343 -0.650 $308,943.75 $533,779.25 $750,000.00 $995,016.00 $1,400,000.00 Bonus $740,497.65 $941,407.45 1.456 1.246 $0.00 $0.00 $374,160.00 $1,107,350.00 $3,337,450.00 Stock options $2,148,379.23 $4,110,365.71 2.190 3.691 $0.00 $0.00 $0.00 $1,991,471.75 $15,157,009.30 Total Compensation $6,020,483.41 $6,674,427.06 1.683 1.954 $543,665,80 $1,465,583.25 $3,273,078.50 $7,714,036.00 $25,329,271.10

Net income (Loss) $447,937,681.87 $739,898,904.19 2.282 4.363 $-64,939,600.00 $40,970,500.00 $134,593,500.00 $464,087,500.00 $2,880,200,000.00

TRS 1.0883 0.29977 0.235 -0.447 0.5622 0.8843 1.0774 1.2748 1.7189

(21)

Table 6

Percentiles 5 25 50 75 95 Salary $896,861.42 $302,860.52 0.341 -0.555 $400,254.40 $666,114.00 $882,308.00 $1,090,000.00 $1,522,555.80 Bonus $100,851.99 $283,166.03 2.903 7.093 $0.00 $0.00 $0.00 $0.00 $1,123,242.40 Stock options $1,617,627.66 $3,310,043.30 2.287 4.130 $0.00 $0.00 $0.00 $1,288,044.00 $12,369,197.60 Total Compensation $6,789,306.66 $6,042,434.19 1.402 1.149 $870,512.20 $2,397,235,50 $4,723,398.00 $9,073,710.50 $23,225,536.00

Net income (Loss) $533,028,506.79 $888,017,399.63 2.081 3.433 $-207,192,800.00 $39,945,000,00 $159,000,000.00 $586,873,000,00 $3,315,000,000.00

TRS 1.0703 0.33518 0.161 -0.468 0.4724 0.8426 1.0713 1.2735 1.7626

ROA 0.0507 0.05540 0.047 -0.227 -0.0657 0.0150 0.0472 0.0847 0.1601

(22)

3.7 Correlations

In this part the results of the correlations will be discussed. A correlation coefficient is a measure of strength of association between two variables (Field, 2009) The correlation coefficient has a range of -1 to 1. Whereby -1 indicated a highly negative relationship

between two variables and 1 a highly positive relation between two variables (Field, 2009). 0 indicates that there isn’t a relation between two variables, which means that the statistical relation is random.

Table 7 shows the statistical relation between the variable used in this research for the whole period. The results shows that salary is strongly correlated with total compensation, as expected. The table show that the r is 0.54, which is a strong relation. Also the other

compensations components are medium and strongly related. So has bonus compensation has a r of 0.276 and stock options has a r of 0.772 in relation with total compensation.

For the company performance in relation with the total compensation the following results occurred; net income has a r of 0.551, ROA has a r of 0.215 and TRS has a r of 0.064. All these correlations are positive and significant (at p value of 0.01) related to the dependent variable.

This partly correspond with the research of Mahmoud et al (2008). The correlation between TRS and total compensation is 0.0796, which is slightly higher than my results. For the correlation between ROA and total compensation the paper of Mahmoud et al (2008) shows a result of 0.0514, which is lower than my results.

Further the correlation between the independent variables, which are used for this research are weak to moderately correlated, because of these correlations, the results show that there are no serious multicollinearity problems.

Table 8 shows the correlation for the pre crisis period. In this period the correlation between salary and total compensation is also here strongly and positive related. This result compared with the total research period shows that the salary (0.512) in the pre crisis period is slightly lower than in the total research period (0.54). The other compensation components are in the pre crisis period medium to strongly correlated to total compensation. So is the bonus in the pre crisis period with 0.413 correlated and the stock options with 0.81 correlated, which is very high. If we compare these results with the total period the results show that the correlation of bonuses and stock options are higher. The correlation of the bonuses have almost doubled. For the stock options the increase was less compared to the bonus but rise from 0.772 to 0.810.

(23)

Furthermore, the company performances in relation with the total compensation shows that the correlations of the different performances have decreased. Where net income had a correlation of 0.511 it now has 0.488. For ROA it was 0.215 and now is 0.206 and for TRS it was 0.064 and is now 0.058. Still all these correlations are positive and significant (at p value of 0.01) related to the dependent variable.

The correlation between the independent variables, which are used for this research are still weak to moderately correlated. So also here there aren’t any serious multicollinearity problems.

Table 9 shows the correlation for the crisis period. This period shows interesting results. To start with the salary, the correlation with total compensation has increased compared to the pre crisis period where it was 0.512 it is now 0.568, which is strong. The correlation between bonus and the total compensation is more than halved in the crisis period (0.138) compared to the pre crisis period (0.413). Such a decrease also appeared in the descriptive statistics. The correlation between stock options and total compensations shows a lower decrease. In the pre crisis period the correlation was 0.810 and in the crisis periods it was 0.735. Still it is very strongly correlated. When the results are compared with the total period is shows that the correlation between salary and total compensation increased slightly with 0.028. For bonuses the correlation with total compensation is still more than halved from 0.276 to 0.138. The correlation between stock options and total compensation has decreased in the crisis period, where the correlation was 0.772 in the total research period it is in the crisis period 0.735.

The correlation between total compensation and the performance measures shows that the correlation is increased in the crisis period compared with the pre crisis period. So is the correlation of net income in the crisis period 0.536 and in the pre crisis period 0.488. For the ROA the correlation is in the crisis period 0.240 and was in the pre crisis period 0.206. For the TRS the correlation is in the crisis period 0.075 and was in the pre crisis period 0.058. All these correlations are positive and significant (at p value of 0.01) related to the dependent variable.

The correlation in the crisis period between the independent variables, which are used for this research are still weak to moderately correlated. So here there are no any serious multicollinearity problems.

(24)

Table 7

Correlations period 2002-2012

Salary Bonus Stock options

Total Compensation Net income (Loss) TRS ROA Salary 1 Bonus .204** 1 Stock options .230** .154** 1 Total Compensation .540** .276** .772** 1

Net income (Loss) .527** .266** .235** .511** 1

TRS .002 .084** .056** .064** .018 1

ROA .054** .105** .239** .215** .265** .101** 1 **. Correlation is significant at the 0.01 level (2-tailed).

(25)

Table 8 Correlations period 2002-2007 Salary Bonus Stock options Total Compensation Net income (Loss) TRS ROA Salary 1 Bonus .397** 1 Stock options .254** .194** 1 Total Compensation .512** .413** .810** 1

Net income (Loss) .534** .405** .243** .488** 1

TRS -.001 .110** .043** .058** .019 1

ROA .062** .117** .228** .206** .223** .061** 1 **. Correlation is significant at the 0.01 level (2-tailed).

Table 9 2008-2012Correlations Salary Bonus Stock options Total compensation Net income (loss) TRS ROA Salary 1 Bonus .072** 1 Stock options .233** .037* 1 Total compensation .568** .138** .735** 1 Net income .517** .145** .234** .536** 1 TRS .013 .017 .071** .075** .023 1 ROA .068** .028 .250** .240** .319** .144** 1

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

(26)

4 Results

In this part of the thesis the results from the regression for the different period are analyzed, presented and discussed. First I will start with the total period, followed by the pre crisis period and lastly the crisis period. With regard for the presenting and discussion of the results it should be mentioned that for analyzing the relationship between company performances and total compensation, for some company performances, namely ROA and TRS uses the

standardized coefficients beta. This is more appropriate, because these values are indexes. For the net income used the unstandardized coefficients beta, because here the value in presented in dollars.

4.1 Results hypothesis period 2002-2012

The regression analysis is first conducted for the total sample size. With 9163 observations to test if the total compensations of CEO’s are not proportional to the performances of the companies, by using three different performance measures: net income, TRS and ROA. Models 1 till 3 shows the effect of these different performance measures on the total compensation. The results for the total period are shown in table 10.

The R² of the first model 0.261, which explains how much the variability is shared by total compensation. So 26.1% is explained by this variable. The other 73.9% will be explained by other variability. The results of R² and the adjusted R² shows that there aren’t any

differences between these two, because there isn’t any shrinkage. This means that if the model were derived from the population rather than a sample it will be the same variance in the outcome.

The unstandardized coefficient B for the net income is 0.004, which tells the

relationship between total compensation and net income. So when the net income rise with $1 the total salary will rise with $0.004. The result shows a weak positive and significant

relationship.

The F value of the F test is very high and significant, with a value of 3233.044. The F test tells how much the model contributes in predicting the outcomes. As shown above the F value is very high, which indicates that this model contributes highly in predicting the outcomes.

In short, from the results it can be concluded that the hypothesis can be rejected for net income because it moves parallel to the performances of the company. This suggests that the total compensations of CEO’s are proportional to the performances of the company.

(27)

The next model, model 2, shows a R² of 0.004. So this models only explains 0.4% of the variability. When the R² and the adjusted R² are compared is shows a comparable magnitude as the first model with no shrinkage, which also means that if the model were derived from the population rather than a sample it will be the same variance in the outcome.

The standardized coefficient β for TRS is positively and significant related. The results for this research show a 0.064 standardized coefficient β, which indicates that if TRS goes up with 1 standard deviation, the total compensation goes up by 0.064 standard deviation.

The F value of the F test is high and significant, with a value of 37.775. So this model also contributes in predicting the outcomes.

The results show that the hypothesis can be rejected for TRS because it moves parallel to the performances of the company.

For the latest model for the total period, model 3 shows a higher R² of 0.046 compared with the R² (0.004) of model 2. This model explains less than the net income but still explains 4.6%. The R² and the adjusted R² also show no differences.

The standardized coefficient β for ROA is positively and significant related. The results for this research shows a 0.215 standardized coefficient β, which indicates that if ROA goes up with 1 standard deviation, the total compensation goes up by 0.215 standard

deviation.

The F value of the F test is very high and significant, it has a value of 444.922, which indicates that this model contributes in predicting the outcomes.

The results show that it can be concluded that the hypothesis can be rejected for ROA, because it moves parallel to the performances of the company.

In conclusion, the results show that for both performances measures the hypothesis can be rejected. However, it should be taken into account that the R² in some models are low, this indicated that the models explain a little of the variability that is shared by the total compensation.

In the next section, the results for the pre crisis period will be discussed.

4.2 Result hypothesis period 2002-2007

The regression analysis for the pre crisis period is conducted with 4998 observations to test if the total compensations of CEO’s are not proportional to the performances of the companies. Again, the three different performance measures are used: net income, TRS and ROA. Models

(28)

1 till 3 shows the effect of these different performance measures on the total compensation. The results for the pre crisis period are shown in table 11.

The R² of the first model shows a value of 0.238, so 23.8% is explained by this variable. When this result is compared with the R² of the total period it shows that it is lower. This suggest that this model explain less of the variability between the net income and the total compensation compared with the total period. In the pre crisis period the results of R² and the adjusted R² shows that there aren’t any differences. This was also the case in the total research period.

The unstandardized coefficient B for the net income is 0.004. As shown in table 11 there is a positive and significant relation. The results show the same relationship between net income and total compensation as for the total research period. So when net income rise with $1 the total salary will rise with $0.004.

The F value of the F test is very high and significant, with a value of 1558.459.

Compared with the F value of the total value it show a decrease of more than a halve, but still the F value is very high, which indicates that this model contributes highly in predicting the outcomes.

In short, the results concluded that the hypothesis can be rejected for net income, because it moves parallel to the performances of the company. This suggests that in the pre crisis period the total compensations of CEO’s where proportional to the performances of the company.

The second model of the pre crisis period shows a R² of 0.003. This is not very high, it means that only 0.3% of the variability by this model is explained. Like the other models, the R² and adjusted R² are the same.

The standardized coefficient β for TRS is positively and significant related. The results show a 0.058 standardized coefficient β, which indicates that if TRS rise with 1 standard deviation, the total compensation will rise by 0.058 standard deviation. Comparing this result it shows that this is lower in the pre crisis period than in the total research period. It has shrank a little (0.006).

The F value of the F test is significant and normal, with a value of 16.673. So this model contributes in predicting the outcomes, however it is less compared with the total research period.

The results shows that the hypothesis can be rejected for TRS.

(29)

The last model for the pre crisis period shows a R² of 0.042, which is slightly lower than that of the total research period. The R² and the adjusted R² have the same result which is in line with the model of the total research period.

The standardized coefficient β for ROA in the pre crisis period is positively and significant related. The results show a 0.206 standardized coefficient β, which indicates that if ROA goes up with 1 standard deviation, the total compensation goes up by 0.206 standard deviation. This is 0.009 lower compared to the total research period.

The F value of the F test is halved compare to the total period. Still it is very high and significant, with a value of 220.665, which indicates that this model contributes in predicting the outcomes.

The results show that it can be concluded that the hypothesis can be rejected for ROA. In conclusion, the results show that for both performances measures the hypothesis can be rejected for the pre crisis period this suggest that in this period the total compensations where aligned to the company performance measures.

However, it should be taken into account that the R² in some models are low, this indicated that the models explain a little and that there are more factors which have influence on the total compensation.

The results for the crisis period will be discussed in the next section.

4.3 Result hypothesis period 2008-2012

The regression analysis for the crisis period has a total of 4165 observations, to test if the total compensations of CEO’s are not proportional to the performances of the companies. The three different performance measures used are: net income, TRS and ROA. Models 1 till 3 shows the effect of these different performance measures on the total compensation. The results for the crisis period are shown in table 12.

As expected the results are very different compared to the pre crisis period. For a good overview the result will first be compared with the pre crisis period and afterwards with the results of total research period.

The R² of the first model is 0.287, which tells that 28.7% is of the variability will be explained by this variable. The R² in the pre crisis period was 0.238. Here it shows that the net income variable explain more during the crisis period, than in the pre crisis period. If the R² of the crisis period is compared with the R² of the total research period, it shows that the R² (0.287) in the crisis period is higher than R² (0.261) of the total research period. The adjusted

(30)

R² are comparable to the models as previously discussed; there is no difference between the R² and the adjusted R².

The unstandardized coefficient B for the net income is in the crisis period 0.004, which is positive and significant. When this result is compared with the pre crisis period and the total research period it shows that this has not been changed during all the period. This suggest that the net income is resistant for economic down turn and economic ascent and, all other factors where companies are exposed to.

The F value of the F test is very high and significant, with a value of 1674.977 it is higher than the F value of pre crisis period. This indicates that this model contributes more in predicting the outcomes than the model does during the pre crisis period. When the results are compared to the total period the F value is than twice as high than the F value of the crisis period.

From the results it can be concluded that the hypothesis can be rejected for net income. This suggests that the total compensations of CEO’s are proportional to the performances of the company.

The second model in the crisis period shows a increase of the R² (0.006) compared to the R² (0.003) the pre crisis period. When this will be compared with the total period it shows that the R² of model 2 is still higher than the R² of the total compensation. Furthermore, for the first time the R² and the adjusted R² are slightly different, which means that if the model were derived from the population rather than a sample it will be less variance in the outcome.

The standardized coefficient β for TRS in the crisis period shows a positively and significant relation. The results show a 0.075 standardized coefficient β. This is a strong increase compared to the standardized coefficient β during the pre crisis period where it was 0.058. The standardized coefficient β is also higher in the crisis period compared with the total research period.

The F value of the F test has risen compared to the pre crisis period from 16.673 to 23.549, which indicates that the model in the crisis contributes more in predicting the outcomes than for the pre crisis period. The F value is lower if it is compared to the total research period, so the total period has an F value of 37.775 and in the crisis period it is 23.549.

From the results it can be concluded that the hypothesis can be rejected for TRS. This suggests that the total compensations of CEO’s are proportional to the performances of the company.

(31)

The last model for the crisis period shows a R² of 0.058, which is higher compared with the pre crisis period. Like the previous models for the crisis period it shows that ROA explain more during the crisis period than in any other period. Furthermore, for the first time the R² and the adjusted R² are slightly different for ROA.

The standardized coefficient β for ROA in the crisis period is positively and significant related. Comparing the results of the crisis period with the pre crisis period, it shows that standardized coefficient β has increased. Where it was 0.206 in the pre crisis period it is now risen to 0.240. When the result are compared with the total research period it shows that the crisis period have a higher value than the total research period.

Comparable to the other models, the F value have also risen compared to pre crisis period. Where it was 220.655 it is now 254.116, which is very high.

In short, the total results show that for both performances measures the hypothesis can be rejected for the crisis period

The results shown that all the performance measures explain more in the crisis period than in the pre crisis period and the total period. Furthermore, in all the models during the crisis the hypothesis has been rejected, which suggest that during the crisis the total compensations are in line with the performances of the company.

On overall, if the results are compared with prior research there are some similarities and some differences. First of all it must be mentioned that no paper uses all three different performance measures. There are paper which use TSR and net income, but there are also papers which use TRS and ROA.

Compared to the results of Mahmoud et al (2008), the market and accounting returns seems to be a good explanatory variable for the total compensation.

For net income the results for all the three periods shows a comparable pay performance sensitivity. While Jensen and Murphy (1990) compute a pay performance sensitivity for shareholders value of $ 0.00325 for USA listed companies. I have found a pay performance sensitivity for the net income which has almost the same amount, so when the net income rise with $1000 the CEO will earn $4.00.

If the net income coefficient result is compared with the coefficient result of Hogan et al (1998) it shows that it was much higher in his research period compared to this research period. The net income is significant in this research as well as in the research of Hogan et al (1998).

(32)

For TRS the results are consistent with the paper of Core et al (1999), which shows a positive and significant coefficient. Comparing the TRS of this research with the TRS of their research it shows that it is much higher in their research. When the TRS results are compared with the TRS results of the research of Lilling (2006) there are some similarities, so is the coefficient positive and is it significant at a 1% level. The only difference is the higher

coefficient of the research of Lilling (2006) compared to the results of this research. Hogan et al (1998) researchers shows that the TRS is higher in his research period compared to the results of this research, moreover it not significant in his research period.

For ROA the results were not consistent with prior research of Core et al (1999) and Lilling (2006). Core et al shows that the coefficient for ROA was positive but not significant. Also, the coefficient of their research was much higher than this research. Lilling (2006) shows that the coefficient was negative and not significant. By looking at all these papers it can be concluded that in the past year some changes has occurred which have influence the coefficients and its significations.

Table 10 Dependent variable: Total compensation 2002-2012

Independent variables Model 1 Model 2 Model 3 Constant 4,391,483.568*** 4,961,871.603*** 4,919,545.481*** 0.000 0.000 0.000 Net income 0.004*** 0.000 TRS 0.064*** 0.000 ROA 0.215*** 0.000 R² 0.261 0.004 0.046 Adjusted R² 0.261 0.004 0.046 Durbin-Watson 1.140 0.880 0.907 F test 3233.044*** 37.775*** 444.922*** 0.000 0.000 0.000 Observations 9163 9163 9163

***Indicate significance at the 1% level

(33)

(34)

Conclusion

In the previous years, there are much criticism and discussion by the public and politicians about the compensations that CEO’s receives. Also, researchers found this topic interesting. A lot of papers are written about this subject and the relation between compensation and

performances in global sense and not specific per industry. The latest research about this subject was conducted in 2008 and used data till 2002. The data are therefore outdated. By investigating the past eleven years, this research provides updated evidence on the

relationship between CEO compensation and company performances for three time periods, namely the total research period, the pre crisis period (2002-2007), and the crisis period (2008-2012).

The descriptive statistics shows some interesting results. It appears that mean salary has strongly increased during the crisis period compared to the pre crisis period. A possible explanation for this event is that CEO’s have increase their salary to compensate for the lower returns from their bonus and stock options. Furthermore, it shows that the mean bonus

compensation strongly decreased with approximately $600,000. The unexpected outcome, which I did not expect, was the rise in the total compensation during the crisis period, because of poor performances of the companies and the low stock market returns.

The aim of this research is to investigate whether the total compensations of CEO’s are not proportional to the performances of the companies.

The results show that for each period and each model the hypothesis must be rejected. The results did not show that in crisis period the CEO compensations are not proportional to the performances of the company based on the performance measures net income, ROA, and TRS. This suggests that the compensations are in line with the performance of the companies. All the performances measures are positive and significant related with the total

compensation. The results shows that the performance measures had a higher coefficient during the crisis period compared to the pre crisis period. Only the coefficient of net income remained the same for all the three the periods, which suggest that the economic downturn and other factors did not have any influence on the coefficient.

A limitation of this research is that it cannot explain how much influence the specific events had on the results of this. Furthermore, there is a lack of independent variables in the formulas of this research.

It is recommended to investigate the effect of SOX and, for example, the bankruptcy of Lehman Brothers on the relation between CEO compensation and performances, this may

(35)

show some interesting results. Another recommendation is to add more independent and control variables to the formulas to get more accurate results of the pay performance relation. A possible research question could be: Does the pay performance relationship for top

management significantly changed after the implementation of Sarbanes Oxley Act?

(36)

References

Agrawal, A. and Mandelker, G. (1987), Managerial incentives and corporate financing and investment decisions, Journal of Finance, 42 (4), 823-838.

Akhigbe, S., Madura, J. and Tucker, A. (1995), Reexamining the link between executive compensation and corporation performance: a note, American Business Review, 13 (2), 83-89.

Anderson, M., C., Banker, R., D., and Ravindran, S. (2000). Executive Compensation in the Information Technology Industry. Management Science, 46, 530–547.

Arya, A. and Sun, H.-L. (2004), Impact of deregulation on CEO compensation: the case of electric utilities, American Business Review, 22, (1), 27-33.

Baker, George P., and Brian J. Hall, (2004), CEO incentives and firm size, Journal of Labor Economics 22 (4), 767-798.

Banker, R.D., S.Y. Lee, and Potter, G. (1996). A field study of the impact of a performance-based incentive plan. Journal of Accounting and Economics 21 : 195-226.

Barro, J.R. and Barro, R.J. (1990). Pay, performance and the turnover of bank CEOs. Journal of Labor Economics 8: 448-481.

Bebchuk, L. A. Fried, J. M., and Walker, D. J. (2002). Managerial power and rent extraction in the design of executive compensation. University of Chicago Law Review, 69, 751-846.

Brick, I. E., Palmon, O., Wald, J.K. (2006). CEO compensation, director compensation, and firm performance: Evidence of cronyism? Journal of Corporate Finance, 12 403-423. Canarella, G. (2008) New insights into executive compensation and firm Performance

Evidence from a panel of ‘‘new economy’’ firms, 1996-2002. Managerial Finance,(8), p.537-554.

Carpenter, M. A., Sanders, W.M. G. (2002). Top management team compensation: the

missing link between CEO pay and firm performance? Strategic Management Journal, 23, 367-375.

Carroll, T., M., Ciscel, D., H. (1982). The Effects of Regulation on Executive Compensation. The Review of Economics and Statistics, 64, 505-509.

Clinch, G. 1991. Employee compensation and firms’ research and development activity. Journal of Accounting Research 27: 59-78.

Core, J., Holthausen, R. and Larcker, D. (1999), Corporate governance, chief executive officer compensation, and firm performance, Journal of Financial Economics, 51, (3),