Performance vs remuneration : what is the effect of the performance of retail companies in the US on the total level of remuneration of a Chief Executive Officer?

(1)

UNIVERSITEIT VAN AMSTERDAM

Performance vs Remuneration

What is the effect of the performance of retail companies in

the US on the total level of remuneration of a Chief Executive

Officer?

Christopher Seraus, 10579257 Supervisor: Stephan Jagua BSc Economics and Business

Finance and Organization 29th July 2016

This study researches the effect of performance of retail companies in the US on the total level of remuneration of a CEO. Although prior literature argues that the performance has a positive effect on the remuneration, the results in this paper show that the performance has no significant effect on the remuneration, regardless the performance indicator. The

performance is measured as return on assets and return on equity. However, the return on assets does have a significant effect on the base and bonus salary and the return on equity has a significant effect on only the bonus salary. The results are estimated with the ordinary least squares method. And the model is validated with the Chow test and the F-test.

(2)

1

This document is written by Christopher Seraus 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)

2

1 Introduction

The fact that a chief executive officer (CEO) earns a lot is a phenomenon. For instance, the best paid CEO according to Rushe (2013) is Mark Zuckerburg, CEO of Facebook, with a remuneration of 2.27 billion USD in 2013. Mishel and Sabadish (2013) claim that the remuneration of CEO's has increased since 2000. What would be the reason that those CEO’s earn such a large remuneration? Is this due to their ability, their social network or rather thanks to the performance of the company.

It seems plausible that if a company performs better, they can afford to increase the remuneration of a CEO. This study concentrates on the effect of the performance of

companies on the remuneration of their CEO's. The performance of a company is measured as the return on assets (ROA) and the return on equity (ROE). There already exist literature that investigates this relationship. However, most of these studies take data from the financial crisis years of 2007-2011 not in consideration. Furthermore they do not focus their study on the retail sector of the US, they focus for example on the S&P 500 and industrial companies. The research question in this paper is: what is the effect of the performance of retail companies in the US on the total level of remuneration of a Chief Executive Officer? In order to answer this research question, we first review the existing literature on firm

performance and its effect on the remuneration. Secondly, we discuss our methodology and introduce the econometric model we use. As third, we provide an overview of our dataset. Next, we present the estimation results and the validation of the econometric model. Lastly, we provide an answer on the research question .

(5)

4

2 Literature Review

A vast amount of literature tries to determine the effect of firm performance on the remuneration of CEO's. This section reviews some important examples of such studies. In general, the existing literature tends to conclude that all types of performance measures have a positive effect on the remuneration.

According to Borjas (2013) the CEO compensation and the firm economic performance each depend on the other. He indicates that there is a positive correlation between the

compensation and the firm performance. But the CEO compensation has a small effect on the rate of return to shareholders. He argues that this effect is too small to impose real constraints on the behavior of the CEO. Furthermore, Borjas discuss a study of 16000 managers at 250 large companies which suggests that if the salary and bonuses depended more on firm performance, this would improve the profitability of the firm. He claims that this indicates that when the CEO receives a bonus for the performance, the return of the shareholders increases.

Jensen and Murphy (1990) used the shareholders wealth as performance indicator, in the first part of their study. Their performance measurement ignores capital payments. Jensen and Murphy use an executive compensation survey in Forbes from 1974 to 1986, with a total of 2213 CEO’s. This sample contains a total of 7750 observations. Their findings are that there is a positive relation between the compensation of the CEO and firm

performance. In their study they also estimate a model with the total compensation, base salary, bonus salary, value of restricted stock, savings and thrift plans and other benefits as dependent variable. In this estimation they conclude that the total compensation has a positive relation with the firm value. Jensen and Murphy's third finding is that bonuses account for an average of 50 percent of the total salary, but that they are not sensitive to firm performance, measured as changes in market value of equity, earnings or sales. This contradicts the argument of Borjas (2013).

The study of Leonard (1990) analyzes an executive and managerial compensation survey of 1981 to 1985. In his study, Leonard first analyzes the effect of returns to schooling, sorting and experience on the executive pay. Leonard argues that the executives have the most investments in their firm-specific human capital, so therefore he claims that their pay should

(6)

5

be more flexible. He finds that the executive pay has a significantly positive effect on profits, but the effect is rather small. Likewise in another estimation using sales instead of profits, Leonard finds similar results. The only difference being that the effect size is smaller. Conversely, changes in assets, equity or employment have no significant effect on the compensation of CEO's. Leonard compares the bonus salary to the base salary and finds that the correlation between those two is 0.23. Furthermore, he argues that the bonus salary is more flexible over time than the base salary.

In Peck and Conyon (1998), the role of board control, remuneration committees and top management pay is investigated. The data they use in their study are from the U.K. Financial times top 100 companies by market value for the period 1991 to 1994. They measured the management compensation as base salary, bonus salary and other

miscellaneous earnings. In their paper they use an estimation with the shareholder return as one of many independent variables. We consider the shareholder return as the

performance indicator of this paper. According to the results of the estimation, Peck and Conyon argue that the shareholder return has a positive significant effect on the dependent variable, what is defined as management compensations.

Bebchuk and Grinstein (2005) discuss the relationship between the compensation of CEO's, firm size, firm performance and industry classification. They use data of three indexes of 1993-2002: S&P500, Mid-Cap400 and Small-Cap600. These indexes reflect compositions of various companies in the US . As performance indicators they use the ROA, sales and firm returns. Bebchuk and Grinstein conclude that all three performance indicators have a significant positive effect on the CEO compensation.

(7)

6

3 Methodology

This section presents the econometric model that will be estimated in this paper. Throughout the analysis the ordinary least squares specifications will be used. The

expectation is that the performance of a company, the performance indicator, has a positive influence on the salary; especially on the bonus part of the salary. The argument for this expectation, will be explained later in this section. The model that is used in this paper is shown below. In section 5, we will validate the inclusion of the crisis and year dummies, with the Chow breakpoint test and the F-test along with the results of the model estimation. Model 1: 3.1 Dependent Variables

For the dependent variable (salary part), we consider three different salary components in this paper: base salary, bonus salary and all other salary. On top of this, we also estimate our model with the total salary of CEO's as the dependent variable. The base salary is the foundation of the remuneration of a CEO. It is determined due to a contract which is the reason why we expect it to be rigid over time, barring some exceptions. A contract cannot be renegotiated while it is running. As pointed out by Leonard (1990), another part of the rigidity of the base salary is also due to the inflexible nature of the wage-bargaining process between trade unions and firms.

The bonus salary is the flexible part of the remuneration. The expectation is that this part varies a lot over time and that the firm performance will have a notable influence on it. Leonard (1990) confirms this argument in his study. When a company performs well, the CEO could get a higher compensation for the effort he exerts (cf. again Leonard(1990)) All other salary contains salary components like: tax reimbursements, life insurance premiums, debt forgiveness and imputed interest.

The choice to make a distinction between the base salary, bonus salary, all other salary and total salary is to see in the empirical part on which part of the salary the performance

(8)

7

indicator has the most effect, or no effect. It will provide an argument if the hypothesis about the bonus salary is valid or not.

3.2 Performance Indicators

In this paper the performance indicators ROE and ROA are the key explanatory variables. As a performance indicator, it means that these two variables are the measurement of the performance of a company. If the ROA and/or ROE are high, the assumption is that the performance of that company is high.

Stolowy, Lebas and Ding (2013), call the ROA and the ROE profitability ratios. A profitability ratio shows the ability of a firm to generate profit. They claim that the ROA measures the efficiency of the firm to create profits using their assets while the ROE, as they argue, emphasizes the return to shareholders. It measures how efficient the firm has used the invested capital of the shareholders to generate profits.

The choice to make use of two different performance indicator lays in the fact that both are a valid performance indicator. When showing estimates of two performance indicators, the conclusion and answer on the research question will be more informative than showing only one performance indicator.

(9)

8

3.3 Overview of Independent Variables

Table 1 shows all independent variables and their definition.

Table 1 independent variables and definition

Variable Definition

ROA

ROE

Age Age is the age of the CEO in years

Size The size of a company as measured by the total

book value of the assets.

Employees The total number of employees of the company

Male dummy 1 - means that the CEO is a male

0 - means that the CEO is a female

Year dummies Dummies for every year from 2005-2013,

excluded 2005, to prevent multicollinearity. This means that the regression constant measures the effect of the year 2005.

Crisis dummy 1 – When the year lays in the financial crisis

(2007, 2008, 2009, 2010,2011) excluded 2007 to prevent multicollinearity.

0 – When the year doesn’t lie in the financial crisis (2005, 2006, 2012 and 2013)

ROA*Crisis Interaction variable between ROA and Crisis

ROE*Crisis Interaction variable between ROE and Crisis

4 Data Descriptive

For collecting the data the combined database WRDS is used. Of WRDS, Compustat and Execucomp are used to collect all the relevant data for this paper. Table 2 lists the

originating database for each variable that is used. The data is collected from 2005 till 2013. We use this time frame in order to get a complete picture of the performance development of retailers over the whole crisis period and including post and pre-crisis years. According to Elliot (2011)the crisis period can be operationalized as the years from 2007-2011. To

compare this with the performance of the period when there’s no a financial crisis, two years before and two years after the financial crisis are included. The effect of a crisis should be visible, because retail companies have in general a high Beta, which means that they are

(10)

9

risky. To combine the database of Compustat and Execucomp, SIC codes are used as

identifiers. The SIC codes used are 5200-5990, which corresponds to the complete US retail sector. The resulting dataset used has 6269 observations with 197 different companies.

Table 2 Variables with specific database

Variable Database

Base salary Execucomp

Bonus salary Execucomp

All other salary Execucomp

Total Salary Execucomp

ROA Compustat

ROE Compustat

Age Execucomp

Size Compustat

Employees Compustat

Male dummy Execucomp

4.1 Correlation

In table 3 the correlations between the independent variables are shown. The correlation between the ROA and the ROE is weak, which seems counterintuitive since the ROA and ROE are both functions of the net income. So the expectation was set on a high correlation, but the table shows that there is a correlation of 0.1959. However, given that the assets of a company are equal to the sum of its liabilities and shareholders’ equity, the liabilities

included in the ROA, but not in the ROE could yield systematic differences between the two performance measures. Retail companies have to innovate more than other sectors,

because they have to keep up with other companies (Vijf trends die de Retail sector

domineren in 2016, 2016). Due to their high need for innovation and related expenses, they need to take up high liabilities. This could be a reason why the correlation between the ROA and the ROE is low. Another correlation that stands out is the correlation between the size and employees. This correlation is strong at 0.7473. This should not surprise us much: when a company is bigger, the quantity of employees is often also higher. So this could give bias results in the empirical part. But this problem will be discussed in section 5.

(11)

10

Table 3 Correlation table

Correlation table

ROA ROE Size Male Age Crisis

1 ROA 0.0482 1 ROE 0.0063 0.1959 1 Size 0.7473 0.0254 0.0072 1 Male 0.0433 0.0046 0.0198 0.0441 1 Age 0.0475 0.0176 0.0030 0.0392 0.1336 1 Crisis 0.0288 -0.1085 -0.0350 -0.0471 -0.0042 -0.0578 1 4.2 Summarize Data

Table 4 provides some common descriptive statistics for the variables used. As shown in the table, the average base salary is $537,714.60, the average bonus salary is $102,187.70, the average of all other salary is $196,172.80 and the average total salary is $836,075.10.

Table 4 Summarized table in USD

Variable Mean Standard

deviation

Minimum Maximum

Base salary 537,714.60 299,306.1 30,105 2,894,231

Bonus salary 102,187.70 444,904.4 0 14,800,000

All other salary 196,172.80 1,194,692 0 69,900,000

Total salary 836,075.10 1,382,738 44,640 73,000,000

Age 52 7.429 26 87

Male 0.843 0.363 0 1

Size 5,679,120 16,100,000 21,548 205,000,000

(12)

11

An interesting preliminary exercise is to examine how the average total salary and its components develop over time. Figure 1 shows the average total salary for all years in our dataset and tracks the composition of total salaries. As seen in the figure, 2005 has a relatively higher total salary than later years. The figure also shows that in 2008, 2009 and 2010 average salaries are at their lowest. This makes sense, because these years are in the core of the financial crisis. Looking at the salary composition in the table, we first notice that the base salary is quite rigid over time. See section 3.1 for the discussion of why the base salary is persistent over time. The bonus salary displays much more variation. And are again in the years 2008, 2009 and 2010 on their lowest. See section 3.1 for an interpretation of these observations. The behavior of bonus salaries over time can be thought to provide a first bit of evidence for our hypothesis that the company performance has a systemic effect on bonus salary. If we assume that retail companies perform less well in the financial crisis, then the drop in bonus salaries as we enter the crisis would capture precisely the relation we are interested in.

To argue the expectation that the performance of a company is lower in the financial crisis, we illustrate the ROA and ROE over time in figure 2. This figure shows that the expectation holds for the ROA, but the ROE differs much. As we can see, in the years of the crisis the ROA is relatively lower than the others years. But when we concentrate on the ROE, this is not the case.

Figure 1 Stacked column of the salary

0.00 200,000.00 400,000.00 600,000.00 800,000.00 1,000,000.00 1,200,000.00 1,400,000.00 1,600,000.00 2005 2006 2007 2008 2009 2010 2011 2012 2013

All Other salary Bonus salary Base salary

(13)

12

Figure 2 ROA and ROE column over years

2005 2006 2007 2008 2009 2010 2011 2012 2013 ROA 0.0708 0.0672 0.0516 -0.004 0.0471 0.059 0.0632 0.0736 0.0699 ROE 0.0947 0.0788 0.0879 -0.133 0.0258 0.1365 -0.238 0.0783 0.1279 -0.3 -0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 In pe rc en t

(14)

13

5 Empirical Research

This paper is built on the effect of the performance of a company on the level of the salary of a CEO. First we use the Chow breakpoint test to validate the use of a crisis dummy. After this, we test the significance of all the years dummies with an F-test. After those two tests, we estimated the model and discuss the results. As last, we test the years outside the crisis separately from the years inside the crisis and consider to estimate a different model with the information we obtain from the last test.

5.1 Results

Given that the years between 2007 and 2011 are the years of the financial crisis, it might be that the relationship between performance measures and salary components undergoes a fundamental change as we move from theses crisis years to non-crisis years and vice versa. I.e., the start of the crisis might constitute a breakpoint in the data. To test for presence of a breakpoint, we run the Chow breakpoint test. For this test, we treat the date from the whole period outside the financial crisis, so 2005-2006 and 2012-2013, as one dataset and the financial crisis data form, 2007-2011, as another dataset and estimate the regressions separately for both subsets. The formula for the chow test according to van Ophem (2014) is

. Table 1 of Appendix A shows the Chow test results. This table

shows significant evidence for a breakpoint for all salary components with a significance level of 1%, irrespective of which performance indicator is used. This means that we should introduce the crisis dummy in the model that we will estimate.

After the Chow Break-point test, we test the significance of the years dummies with the F-test to validate the dummies in the model. According to Stock and Watson (2015), the F-F-test is used for testing joint hypothesis about regression coefficients. Table 1 of Appendix B shows the results of the F-test. Both tests yield a significant result at a level of 1%, what validates the use of year dummies.

After validating the breakpoint and the significance of the year dummies, we can introduce the year dummies and crisis dummy to the model and estimate it. Also the interaction variable ROA*crisis or ROE*crisis are included in the model. The reason for this interaction variable is that the performance indicator differs relatively a lot in times of the financial crisis, what is shown in section 4.2. There is also argued that the ROE has not the same

(15)

14

change as the ROA in times of crisis. But to be consistent, we will have an interaction variable for both performance indicators.

Table 5 shows the results of the estimations of the model with the ROA. In this table, we control for age, size, employees and gender. As we can see, the ROA has a significant effect only on the base and bonus salary. The effect on the bonus salary is negative in contrast to the effect of the base salary. The ROA has a negative effect on the bonus part, what goes against the expectation. Cable and Vermeulen (2016) discussed in their study, why companies should not give a higher bonus salary when the company performs better. According to them, there are five explanations for the non-positive effect of ROA on the bonus salary. The first one is that when the tasks of the CEO are not standard and can vary, offering a higher bonus salary would give a negative effect on the performance of a CEO. The second argument is that fixating pay for performance, this could weaken the

performance of a company. In their study they discuss studies that show that fixating for acquiring new skills and learning some new things improve the performance of the CEO. The third argument is about intrinsic and extrinsic motivation. Intrinsic motivation means being motivated to do tasks out of genuine interest in completing it. Extrinsic motivation means being motivated by monetary incentives – in our case the bonus. A bonus salary triggers extrinsic motivation, but a company should also be interested in making the intrinsic motivation of employees work in its favor. Fourth argument is that if there is more pay for performance, in other words the bonus salary, CEO’S could fake achievements according to whatever performance measure is used or stop putting effort in as soon as the bonus hits its upper bound. This might negatively affects the performance of the company as whole. A last argument is provided by the fact that any performance measure is imperfect and therefore will always miss some dimensions of what would be an optimal incentive scheme for the CEO. All of these reason help explain why the bonus salaries does not have to relate

positively to the CEO remuneration to any significant extent, while the base salary does. But the negative effect of the ROA on the bonus salary has no obvious explanation. In the end, the ROA has a positive effect on the base salary, which means that the base salary increases if the ROA increases. When a company performs better, in terms of the ROA and ROE, they could and in this case will give the contracts that are signed at that year a higher salary. The crisis has a significant effect on all salary components. On base, bonus and total salary the

(16)

15

crisis effect is negative, in contrast with all other salary where the effect is positive. This could be due to the fact that there is more debt in times of crisis, which could imply that there will be more debt forgiveness, which is part of all other salaries as explained in section 3.2.1. The board of a company could forgive debt that the CEO has in the company. This increases the component, all other salary, which explains the positive effect.

Table 5 Regression with ROA; n=6,269

Dependent variables Independent variables Log (Base

salary)

Log (Bonus salary)

Log (All other salary) Log (Total salary) ROA 0.578 (0.131)*** -7.100 (1.29)*** -0.708 (0.620) 0.122 (0.156) Age 0.017 (0.001)*** -0.035 (0.009)*** 0.047 (0.004)*** 0.019 (0.001)*** Log (Size) 0.074 (0.005)*** -0.114 (0.049)** 0.255 (0.024)*** 0.079 (0.0059)*** Log (Employees) 0.088 (0.005)*** 0.038 (0.051) 0.274 (0.024)*** 0.114 (0.006)*** Male dummy -0.009 (0.018) -0.276 (0.177) -0.052 (0.084) -0.026 (0.021) Crisis dummy -0.092 (0.040)** -8.578 (0.399)*** 0.644 (0.191)*** -0.488 (0.048)*** ROA*Crisis -0.381 (0.144)*** 5.960 (1.426)*** 1.322 (0.681)* -0.016 (0.172)

Year dummies included YES YES YES YES

Adjusted R-squared 0.2443 0.0848 0.1195 0.2412

***significance at 1%, **significance at 5%, *significance at 10% Between parentheses are the standard errors

Table 6 shows the model with the ROE as performance indicator. As shown in the table, the ROE is not positive significant for any salary part, however it has a negative significant effect on the bonus salary. The reason for the non-positive effect can be argued on the same way as explained in the estimation when the ROA is used as performance indicator.

(17)

16

Table 6 regression with ROE; n=6,269

salary)

Log (All other salary) Log (Total salary) ROE 0.104 (0.025) -0.868 (0.245)*** 0.100 (0.117) -0.045 (0.029) Age 0.018 (0.001)*** -0.035 (0.009)*** 0.047 (0.004)*** 0.019 (0.001)*** Log (Size) 0.074 (0.005)*** -0.105 (0.049)** 0.256 (0.024)*** 0.079 (0.006)*** Log (Employees) 0.089 (0.005)*** 0.032 (0.051) 0.274 (0.025)*** 0.114 (0.006)*** Male dummy -0.010 (0.018) -0.272 (0.177) -0.049 (0.084) -0.027 (0.021) Crisis dummy -0.180 (0.039)*** -7.434 (0.389)*** 0.732 (0.185)*** -0.496 (0.047)*** ROE*Crisis -0.009 (0.025) 0.841 (0.248)*** -0.071 (0.118) 0.049 (0.030)*

To refer back to the argument about the high correlation between size and employees in section 4.1. We can see in table 5 and 6 that both variables have a significant effect on the salaries, irrespectively which performance indicator is used. This means that leaving one of them out, would harm the estimation results of the model. So, including both variables does not give bias results in contrast to the argument made in section 4.1.

The estimations above are estimated with the crisis dummy and the year dummies.

However, it is not obvious to see that adding the year dummies to the estimation will result in a better estimation than a estimation without year dummies. Because of the fact that the validation of the breakpoint, also means that the years in the F-test should be significant, because these years are particularly the same. To validate the use of the year dummies, we will test the years outside the crisis separately from the years inside the crisis. This means testing the significance from the years 2006, 2012 and 2013 and testing the significance of the years 2008-2011. The years 2005 and 2007 are left out, in order to fix for

(18)

17

crisis and outside the crisis are significant at a level of 1%. This validates the argument that the year dummies improve the model without the year dummies. So the model used in table 5 and 6 is the richest and at the same time the best model to estimate the effect of the performance indicator on the salaries.

5.2 Robustness check

In section 5.1 homoscedastic standard errors are used. This means according to Stock and Watson (2015) that the variance of the error term is constant and does not depend on the independent variable. Otherwise the standard error is heteroskedastic. To check if the results differ if heteroskedastic standard errors are used, we estimate model 1 again. The reason to do this is to take in account the outliers in the dataset. An outlier is a value that is very different from other values in the dataset. In other words, the observations are far away from the other observations. Appendix C shows the results of the new estimations. Whenever there is a difference for the significance of a variable, the result in appendix C is shown in bold. As we see, there are two differences according the significance of the performance indicators. In table 1 of Appendix C the interaction variable between the ROA and crisis switched from a significance level of 10% to a significance level of 5% on all other salary. In table 2 of Appendix C the interaction variable between the ROE and crisis switched on the same way as above, but now for the total salary part. So, we can argue that there are no big differences when using the heteroskedastic standard errors.

(19)

18

6 Conclusion

This study is based on the effect of the performance of retail companies on the

remuneration of the CEO. In the introduction the research question was set as “What is the effect of the performance of retail companies in the US on the level of remuneration of a Chief Executive Officer? “. To answer this research question, two models are estimated with the use of OLS. In section 5.1 it is proved that the crisis dummy and the year dummies have a significant effect, so the inclusion of those dummies are validated. With the inclusion of those dummies, there are two different approaches to answer the research question. The approaches differ in the performance indicator, what could be ROA or ROE. When looking at the ROA as performance indicator table 5 shows that the ROA has an effect on the base and bonus salary, but no effect on the total salary. Therefore, we can conclude that the

performance of a company, measured in ROA, has no effect on the total remuneration. However, when looking at solely the base and bonus salary it does. When choosing for the ROE as a performance indicator, table 10 shows that only the ROE has an effect on the bonus salary. So the conclusion is likewise as the conclusion when looking at the ROA as performance indicator. But in contrast with the ROA, the ROE has no effect on the base salary. This is what we expected during this study, because the base salary is rigid. When combining the two conclusions about the different performance indicators, the research question can be answered. When looking at this paper, the performance of retail companies in the US has no significant effect on the total level of remuneration of a Chief Executive Officer.

However there a limitations to this research. It is questionable to argue that the ROA and the ROE are the right performance indicators for a company. Also not all companies used in the dataset, were existing during the whole time period. So, the remuneration level and distribution can be affected by having the information that a company goes bankrupt. This problem occurs also with the tenure of the CEO's. Not all CEO's stay more than one year at a company, so this could also affect the estimation. And as lasts, it is debatable to claim that the effect of the performance indicators affects the remuneration; it could be the case that the remuneration has an effect on the performance of a company as studied by Borjas (2013), this is called inverse causality.

(20)

19

7 Reference list

Bebchuk, L., & Grinstein, Y. (2005). The growth of Executive Pay. Oxford Review of Economic Policy, 21(2), 283-303.

Borjas, G. J. (2013). Labor economics. New York, United states: McGraw-Hill.

Cable, D., & Vermeulen, F. (2016, February). Stop paying Executives for Performance. Hardvard Business review. Retrieved from Hardvard Business review website: https://hbr.org/

Conyon, J. M., & Peck, I. S. (1998, April). Board Control, Remuneration Committees, and Top Management Compensation. The Academy of Management Journal, 41(2), 146-157. Elliot, Larry. (2011). Global financial crisis: five key stages 2007-2011. The Guardian.

Retrieved from https://www.theguardian.com

Jensen, C. M., & Murphy, K. J. (1990, April). Performance Pay and Top-Management Incentives. The Journal of Political Economy, 98, 225-264.

Leonard, J. S. (1990, February). Executive pay and firm performance. Industrial and labor relations Review, 43, 13-29.

Mishel, L., & Sabadish N. (2013, June). Ceo pay in 2012 was extraordinarily high relative to typical workers and other high earners. Economic Policy Institute, 367.

Ophem van, H. (2014). Syllabus Research Project Economics and Business. Amsterdam, Nederland: Universiteit van Amsterdam

Rushe, D. (2013, October). US CEOs break pay record as top 10 earners take home at least $100m each. The guardian. Retrieved from: https://www.theguardian.com/business/ Stock, J. H., & Watson, M. W. (2015). Introduction to Econometrics. Essex, England: Pearson. Stolowy, H., Lebas, J. M., & Ding Y. (2013). Financial accounting and reporting. Hampshire,

England:Cengage Learning EMEA.

Vijf trends die de Retail sector domineren in 2016. (2016, January). Retrieved from http://www.emerce.nl/

(21)

20

Appendix A Chow Breakpoint Test

Table 1 Chow test results

Dependent variable

Total SSR SSR of group in the crisis

SSR of group outside the crisis

F Statistic With the ROA

Base salary 1623.8 1120.97 464.79 75.142***

Bonus salary 168670.96 104617.92 63805.39 4.606***

All other salary 36168.08 24211.26 11554.06 35.276***

Total salary 2336.76 1576.18 717.71 58.543***

With the ROE

Base salary 1623.64 1125.13 470.43 55.128***

Bonus salary 169190.97 104709.21 64295.18 3.458***

All other salary 36182.56 24224.27 11557.75 35.065***

Total salary 2337.29 1577.29 716.35 59.614***

(22)

21

Appendix B F-Tests

Table 1 F-test on year dummies

Dependent variables Log (Base salary) Log (Bonus

salary)

Log (All other salary)

Log (Total salary) With the ROA

F-test 12.67*** 66.65*** 5.61*** 21.89***

With the ROE

F-test 14.06*** 65.69*** 5.54*** 22.45***

***significance at 1%, **significance at 5%, *significance at 10%

Table 2 F-test on years outside the crisis, except 2005

salary)

F-test 3.05** 142.22*** 6.77*** 33.23***

With the ROE

F-test 2.85** 141.89*** 6.77*** 33.22***

***significance at 1%, **significance at 5%, *significance at 10%

Table 3 F-test on years inside the crisis, except 2007

salary)

F-test 20.64*** 4.33*** 5.83*** 9.94***

With the ROE

F-test 21.77*** 4.25*** 5.69*** 10.39***

(23)

22

Appendix C Robust Regressions

Table 5 Robust regression with ROA; n=6,269

salary)

Log (All other salary) Log (Total salary) ROA 0.578 (0.141)*** -7.100 (1.746)*** -0.708 (0.719) 0.122 (0.182) Age 0.017 (0.001)*** -0.035 (0.009)*** 0.047 (0.004)*** 0.019 (0.001)*** Log (Size) 0.074 (0.006)*** -0.114 (0.049)** 0.255 (0.023)*** 0.079 (0.006)*** Log (Employees) 0.088 (0.006)*** 0.038 (0.052) 0.274 (0.026)*** 0.114 (0.007)*** Male dummy -0.009 (0.018) -0.276 (0.177) -0.052 (0.082) -0.026 (0.020) Crisis dummy -0.150 (0.038)** -7.797 (0.440)*** 0.644*** (0.250)** -0.488 (0.058)*** ROA*Crisis -0.381 (0.159)*** 5.960 (1.863)*** 1.322 (0.792)** -0.016 (0.198)

(24)

23

Table 6 Robust regression with ROE; n=6,269

salary)

Log (All other salary) Log (Total salary) ROE 0.104 (0.024) -0.868 (0.293)*** 0.100 (0.134) -0.045 (0.026) Age 0.018 (0.001)*** -0.035 (0.009)*** 0.047 (0.005)*** 0.019 (0.001)*** Log (Size) 0.074 (0.006)*** -0.105 (0.049)** 0.256 (0.023)*** 0.079 (0.006)*** Log (Employees) 0.089 (0.006)*** 0.032 (0.052) 0.274 (0.026)*** 0.114 (0.007)*** Male dummy -0.010 (0.018) -0.272 (0.182) -0.049 (0.082) -0.027 (0.020) Crisis dummy -0.180 (0.037)*** -7.434 (0.403)*** 0.732 (0.250)*** -0.496 (0.056)*** ROE*Crisis -0.009 (0.025) 0.841 (0.295)*** -0.071 (0.135) 0.049 (0.026)**

Performance vs remuneration : what is the effect of the performance of retail companies in the US on the total level of remuneration of a Chief Executive Officer?