The effects of CEO-pay-ratio on company performance : a study on 48 European publicly listed companies

The effects of CEO-pay-ratio on company performance

A study on 48 European publicly listed companies

Bachelor Thesis in Economics and Finance

University of Amsterdam

Faculty of Economics and Business

Author:

Alberto Maccario

Student nr: 10437444

Date:

February 2, 2016

Field:

Finance

Supervisor: Andro Rilović

Abstract

The CEO-pay-ratio is the ratio between the compensation of a company’s CEO and the median compensation of its employees. This paper examines the relationship between the CEO-pay-ratio and company performance for 48 European publicly listed companies in the time period 2009-2014. The goal of this study is to determine whether the size of the gap between CEO pay and average employee salary is a significant factor for a company’s performance. This is to help shareholders make an informative decision when having to determine executive compensation packages. The results of the analysis show that a positive relationship is present between the two variables, but that the effect size of CEO-pay-ratio on company performance is almost null.

1

Statement of Originality

This document is written by Alberto Maccario 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.

2

Table of Contents

3.2 Sample and Data ... 10

3.3 Measures ... 11 3.4 Research Model ... 13 3.5 Descriptive Statistics ... 16 4. Empirical Results ... 18 4.1 Robustness Check ... 20 5. Conclusion ... 21 References ... 23 Appendix A ... 25 Appendix B ... 26 Appendix C ... 26 Appendix D ... 27 Appendix E ... 27 Appendix F ... 28

3

1. Introduction

Over-paid top executives have been consistently characterized as an issue by the popular press, with particular emphasis placed on the increasingly high difference between the compensation of CEOs relative to that of average employees. The American Federation of Labor and Congress of Industrial Organization, better known as the AFL-CIO, showed that in 2015 an average Walmart worker paid at a $9/hour rate would need to work a total of 1,036 hours, or 130 days, to equal a single hour of Walmart CEO Doug McMillon’s pay (Executive Paywatch, 2015). Empirical studies suggest that the CEO-to-employee compensation ratio for the top 350 U.S. firms ranked by sales was estimated to be 20-to-1 in 1965, peaking at 383.4-to-1 in 2000, and 295.9-to-1 in 2013 (Davis and Mishel, 2014). Smith and Kuntz (2013) report a similar CEO pay ratio across the S&P 500 Index, with estimated values of 204-to-1 in 2013 and an average of the top 100 companies almost reaching a 500-to-1 ratio, making it very clear that a disparity in pay within organization is present.

This issue forced the U.S. Security and Exchange Commission (SEC) to adopt a regulation on the 5th of August 2015 requiring all US public companies to disclose the ratio between compensation of its CEO and median compensation of its employees, what is now known as the CEO-pay-ratio, on top of the CEO compensation which was required to be disclosed already. Despite the socio-ethical implications related to this topic, the SEC’s decision is focused on the added value that this information gives to shareholders, stating, in their press release, that the new rule “provides companies with flexibility in calculating this pay ratio, and helps inform shareholders when voting on executive compensation packages” (SEC, 2015).

On top of the usefulness to shareholders when voting on executive compensation packages (“say on pay”) studies have shown that CEO-pay-ratio may be related to company performance. In their study on the banking sector, Crawford, Nelson and Rountree (2014) found that bank performance, measured using return on assets as proxy, is increasing in CEO-pay-ratio, though not in a linear manner. Similar results are found in Faleye, Reis and Venkateswaran (2013) where a change in relative pay was found to be positively associated with a change in Tobin’s q – their measure of company performance. However, Faleye et al. (2013) also find no relationship between relative pay and firm productivity. The presence of these contradictory findings suggests that the effect of relative pay on company performance is still unclear, calling for the necessity of further research to establish the real influence it has. In fact, if through further empirical research one can show that a relationship exists between relative pay and company performance, the value to shareholders of such a figure would

4

significantly increase as it would provide a justification of the choice of the compensation package.

The purpose of this paper is to tackle this issue by conducting an empirical analysis on the relationship between CEO-pay-ratio and company performance. While the vast majority of empirical tests on pay ratios and company performance have been conducted on U.S. companies, this research will focus only on European companies. This adds a substantial relevance to the paper, as it represents one of the first empirical analysis conducted on this group of companies. In addition, this paper leverages the unique CEO compensation data available from the online platform DirectorInsight to provide some of the first European company-specific evidence on CEO-pay-ratio and its relation to company performance.

The remainder of the paper is organized as follows. Section 2 reviews the existing literature and theory on CEO-pay-ratio and firm performance. Section 3 formulates the hypothesis based on findings from Section 2, discusses the research strategy and describes the data and the method used. Section 4 presents the empirical results and Section 5 summarizes the findings and concludes.

2. Literature Review

The existing literature makes a clear distinction between two theories in which the relationship between CEO-pay-ratio and company performance may fall in, specifically Tournament Theory and Equity Fairness Theory.

Tournament theory postulates that pay disparity within corporations improves performance because it motivates workers to exert more effort to achieve promotions and reach a higher pay bracket (Lazear and Rosen, 1981; Green and Stokey, 1983). The term tournament refers to a labour contract in which employees are paid a share of a monetary reward, instead of a salary or hourly wage or the typical performance based compensation (Maloney, 2003). To maintain a feeling of competitiveness within the organizational hierarchy, the size of the reward must be increasing at each subsequent stage in the tournament, with a significant difference between the winners reward (i.e. the CEO) and all other rewards given in previous stages (Rosen, 1986). Thus, higher productivity can result from greater pay ratios as average employees strive to win the promotion tournament (Faleye et al., 2013). Proponents of this theory include, above all, Lazear and Rosen (1981) whose research demonstrates that a tournament results in higher performance as opposed to hourly wages, by proving that an optimally structured tournament yields results identical to piece-rate pay, or receiving

5

compensation only on what is produced, which in turn is also proven to be superior to hourly wages.

The opposing view, equity fairness theory, suggests that a large pay difference is counterproductive to a firm’s performance because it generates feelings of inequity, deprivation, and resentment among employees, resulting in decreased effort and cooperation or, in a more extreme case, acts of sabotage aimed purposely at damaging the organization (Akerlof and Yellen, 1990; Lazear, 1989). When employees compare their efforts and relative pays to the substantially higher one that executives receive, it influences their behaviour and attitude, by instilling in them a feeling of unfairness (Wade, O’Reilly, and Pollock, 2006). Proof for this argument can be found in the works of Akerlof and Yellen (1990) and Lazear (1989) where it is found that pay dispersion leads to disharmony, resulting in lower productivity and subsequently poor performance.

Consistent with the implications of tournament theory, the research conducted by Faleye et al. (2013) on 450 US firms retrieved from the S&P 1500 concludes that in a setting where tournament incentives are plausibly stronger, such as when the firm has fewer employees who are informed on executive pay or are not unionized, pay disparity between the CEO and the average employee is positively associated with company performance. In addition, another significant finding of their study is that relative pay is higher among firms in homogeneous industries, where employees are more interchangeable and thus less powerful relative to top executives. This shows that the industry setting plays an important role when investigating the relationship between CEO and average employee compensation, which is why it will be included in this research, by differentiating between the eight industries of focus (see Appendix B).

However, Faleye et al. (2013) also discuss that workers may not be informed on executive compensation or motivated enough to act in response to the difference in pay between the CEO and all other employees, which would result in an unsuccessful tournament and a positive relationship between the two variables might not be observed. In fact, when considering the full sample and if no assumptions are put in place, findings tend to a more null result, where no significant relationship is found. Nevertheless, findings do not uncover that a higher pay differential is associated with reduced productivity and company performance, thus not witnessing the phenomenon of equity fairness.

Although Faleye et al. (2013) do not find any significant evidence on equity fairness in their study, other works do. An example of this can be found in the research performed by Henderson and Fredrickson (2001) where tournaments and equity fairness are analysed through

6

several pairs of rival hypotheses that put these two theories to a stringent competitive test. The researchers define two distinct groups in which supporters of these theories fall into, specifically economists and behavioralists. Similarly to Akerlof and Yellen (1990) and Lazear (1989), it is argued that a behavioralist view contends that large pay gaps weaken coordination by creating a feeling of relative deprivation among employees, who perceive that they receive less than what they deserve, leading to adverse reactions such as absenteeism or strikes (Cowherd and Levine, 1992). For instance, Cowherd and Levine’s (1992) research revealed that in firms where pay gaps between executives and lower level workers were greater, product quality was found to be lower. Thus, establishing a smaller pay gap will promote cooperation and lower the chance of sabotage. In this view, firms requiring greater coordination needs will exhibit smaller pay gaps between executives and employees, resulting in enhanced firm performance (Henderson and Fredrickson, 2001).

Challenging this view, in accordance with the findings of Lazear and Rosen (1981) and Green and Stokey (1983), economists contend that when a high level of coordination is required, large pay gaps are favourable as they provide robust, tournament-like incentives which, as monitoring difficulties increase, reduce the need for costly supervision as they elicit strong individual effort (Lazear and Rosen, 1981; Rosen, 1986). Thus, the economic view predicts that pay gaps will be larger in firms where higher coordination needs are necessary, and the combination of higher needs and larger pay gaps will enhance firm performance (Henderson and Fredrickson, 2001).

Henderson and Fredrickson’s (2001) study focuses on the interaction between a company’s coordination needs and the pay gap between its managers, and how the combination of the two affects company performance. An interesting finding of their research is that a balance is present between the economic and behavioural views as predictors of firm performance. The reason being that coordination needs are measured using different variables. In a context where the variables which, according to the researchers, are likely to affect coordination needs are relatedness and the number of vice presidents in a company, the interaction between larger pay gaps and greater coordination needs was associated with higher performance, thus supporting tournament theory. To explain this conclusion, the researchers speculate that in firms with many vice presidents and a high level of relatedness, peer-to-peer monitoring may be stronger than superior-subordinate monitoring. In such firm, they say, large CEO pay gaps may be effective because they represent prizes that overcome problems related to shirking, free riding and outright sabotage as employees feel a higher level of competitiveness, resulting in higher overall performance (Henderson and Fredrickson, 2001).

7

However, if instead the variables which are likely to affect coordination needs are a company’s capital investment activity and the number of businesses it owns, Henderson and Fredrickson (2001) found that higher performance was associated with the interaction of smaller gaps and greater coordination needs. Here, the researchers explain this finding by speculating that employees in firms with multiple businesses, which, according to them, are characterized by high capital investment activities, feel a higher level of accountability to superiors. A high level of accountability already promotes strong effort and tournament competition would only encourage sabotage, which is why in this case large CEO pay gaps would be detrimental (Henderson and Fredrickson, 2001).

This shows 2 significant results. Firstly, the outcome of a test on the relationship between CEO and employee pay gap and company performance may unfold two opposing conclusions, based on the variables utilized and the setting which the test is performed in. Secondly, since tournament theory is found to be as strong as equity fairness theory, it might be the case that the two forces cancel out, thus resulting in a null absolute effect on performance. This possibility calls for further research on the topic, with a keen interest on how the choice of setting may affect the strength of either theory, thus challenging Henderson and Fredrickson’s (2001) study.

In fact, in the research conducted by Burns, Minnick and Starks (2013) an analysis of cross-country differences in the tournament structure of executive compensation is utilized to address the degree to which the CEO tournament structure is affected by cultural factors, and the success of such structure in terms of superior firm value. The results of this research show that tournament structure is steeper in U.S. companies as compared to foreign companies and that the steeper the tournament structure is the higher the performance of the company, thus supporting the view of economists. Burns et al. (2013) suggest that this can be explained by cultural influences, which include the desirability of competition, the acceptability of power and the belief that income differentials are fair. Overall, Burns et al.’s (2013) analysis supports the hypothesis that tournaments enhance firm performance as they represent an effective incentive mechanism for motivating firm managers to always strive to reach a higher position, and pay bracket, within the company. And, most importantly, it gives a setting in which this conclusion can be witnessed the most, specifically in countries that value competition, power, and differences in income due to hierarchical structures. Here, the effectiveness of steeper tournament structures in improving firm value will be higher (Burns et al., 2013).

It is important to note that, while North America is the region where Burns et al.’s (2013) conclusion is more significant, the analysis shows that the same conclusion can be

8

witnessed in the other 3 regions, specifically Europe, represented by Western and Northern countries only, Oceania and Africa (South Africa), with observations on Europe exhibiting the closest results to the US. The fact that these 4 specific regions show significant results, while Asian and Middle Eastern countries do not, is somewhat rational, as they all share a cultural aspect within the business world which is very typical in Western civilizations – a capitalistic mind-set. It is surprising, however, to see that Nordic countries do not exhibit this kind of behaviour, as their business mentality is similar in many ways to the Western countries one. It is also important to note that no Eastern European country was included in the analysis. If they were, then we would expect to see different results, mainly due to cultural differences. Western European countries and the US have a higher tolerance for pay disparity, as individuals are driven by the view that working hard is beneficial and a higher difference in pay will be more rewarding for their efforts (Conyon and Murphy, 2000). Given these findings, this paper assumes that the relationship which will be uncovered between CEO-pay-ratio and company performance will be a positive one.

In summary, the question whether CEO-pay-ratio affects company performance remains an open one. This is mainly due to sample selection and the choice of variables used to analyse the data. Where Faleye et al. (2013) find that in a setting where tournament incentives are plausibly stronger pay disparity between the CEO and average employee is positively associated with company performance, Henderson and Fredrickson (2001) uncover that, if capital investment activity and number of businesses are taken into account, the interaction of smaller pay gaps and greater coordination needs is associated with higher performance. Furthermore, the possibility of finding no relationship is also a plausible one, as demonstrated, again, by the works of Faleye et al. (2013) and Henderson and Fredrickson (2001).

However, while research has been conducted on CEO-pay-ratio for North American companies to reach the conclusions presented above, studies putting a greater or complete focus on European companies are still not present, which is how this paper differs from the existing research. In fact, the intention of this analysis is to build upon the existing research by examining whether a significant relationship can be found between CEO-pay ratio and company performance for a sample of European based firms. While most of the existing empirical tests often focus on the amount the executive can potentially receive, assuming all performance criteria are met, this study will focus on realized compensation. Farmer, Archbold and Alexandrou (2013) discuss that the reasons for the often-inconclusive findings about pay and performance is the way that academic researchers measure pay. In their view, focusing on

9

the actual realized pay rather than the potential one will give a more insightful understanding of the relationship between pay and performance. Following this conclusion and the link it has with the goal of this paper, realized compensation will be utilized.

3. Research Methodology

3.1 Hypothesis Development

This study examines the relationship between CEO-pay-ratio and company performance, in the period ranging from 2009 to 2014, for a sample of 48 European listed companies active in 6 countries and 8 different industries. These, together with the corresponding CEO compensations, have been retrieved from DirectorInsight – an online platform developed by AMA Partners which provides users with detailed information on corporate governance and executive pay for European companies. This adds significant value to the paper, as DirectorInsight uses a unique method to calculate CEOs compensation, analysing each company’s award policies and both long and short term outstanding awards, in order to construct the true annual total granted compensation.

Therefore, this paper focuses on the following research question: To what extent does CEO-pay-ratio affect company performance?

In order to answer the research question, and following what was discussed in the previous section, the hypothesis of this paper is set to test the assumption that pay ratio and performance are positively related, against the possibility of finding no relationship at all. The hypothesis is tested over the time period 2009-2014 for all 48 companies:

H0: Company performance is not related to CEO-pay-ratio

H1: Company performance is positively related to CEO-pay-ratio

Although the prediction, based on previous studies, is that the relationship will be positive, a negative relationship would have the same impact as a positive one on the value added to shareholders. A positive relationship would incentivize shareholders to opt for a more significant difference between CEO total granted compensation and average employee compensation; while a negative one would produce the opposite outcome. In the case of a null result, shareholders would still regard it as an interesting finding, as it would point to the fact that relative pay is probably not a crucial factor for firm performance.

10

3.2 Sample and Data

The data was retrieved from DirectorInsight and S&P Capital IQ and composes a balanced panel dataset with annual observations for 48 companies and 6 time periods, specifically 2009-2014, for a total of 288 firm-year observations. The total number of European companies available on DirectorInsight that had full information on CEO total granted compensation was 71. However, data relative to average employee compensation retrieved from Capital IQ was only available for 48 of the 71 companies analysed. Thus the decrease in the sample size. The list of companies analysed with the relative industry and country can be found in Appendix A. The 6 countries in scope are Belgium, France, Germany, Netherlands, Switzerland, and United Kingdom (a list of the 8 industries can be found in section 3.4). Data relative to firm performance, risk, and size has been retrieved from Capital IQ.

It is important to note that neither the companies nor the time period were selected from a broader scope or using specific criteria. The platform DirectorInsight is still under development and the total number of European firms for which full information on CEO total granted compensation was available is 71, starting from the year 2009. This represents a limitation to the analysis as, even though involuntarily, it introduces potential sample selection issues. Not only are many European public companies left out, but also the countries they operate in will not be analysed. The positive aspect is that the companies analysed all come from Western European countries which share similarities – relatively stable political and economic structures, comparable educational systems with a high level of educated work force who share similar work habits and, most importantly, comparable pay scales at any given position within a corporation – thus reducing country specific effects. The negative aspect is that the analysis will not include the effect that, for example, Scandinavian or Eastern European companies would have on the final result. The importance of this stems from the differences present between these countries. Although all are part of the same continent and regime, it is renowned that Western, Eastern, Northern, and Sothern European countries share similarities within each group, but differ among each other, and it would have been interesting to analyse how these differences would have impacted the relationship between pay ratio and performance.

Restricting the research to a selected number of countries and relevant firms may increase the bias of the regressor coefficient and alter the significance of the end result. Using the total number of public European companies, and thus all European countries, would reduce the bias and improve the outcome of the analysis. In addition to this, the number of companies analysed had to be reduced from 71 to 48 due to poor information on average employee

11

compensation, adding to the sample selection problem discussed above. However, while the number of firms had to be reduced, the countries analysed stayed unchanged. A detailed explanation of how and why 23 companies had to be excluded is present in the next section.

3.3 Measures

Independent Variable

The independent variable used in this research is CEO-pay-ratio (PayRatio). It is the ratio between CEO total granted compensation (CEOcomp) and average employee compensation (AvgEmplComp) (see Appendix D for compensation statistics). The former has been retrieved directly from DirectorInsight and it is composed of cash base compensation and both long term and short term management incentive plans, including stock options, performance units and share plans, restricted stock, phantom stock and cash bonus, all of which are closely analysed to construct the actual annual realized compensation. Given that European companies are not required to disclose non-executive employee wage, the latter is calculated by taking salary expenses reported on each company’s income statement and dividing it by the total number of employees at the end of the fiscal year. While this represents the most common method used to calculate this figure, as shown by the works of Crawford et al. (2014) and Faleye et al. (2013), it carries a limitation. Companies may have a disparity between the disclosed number of employees and the relative total salary due to unknown factors, yielding an incorrect value for average employee salary. For example, company Randstad, where one of the services it provides is assist people in finding an employment, treats all the people who they found a job for as employees, though not being their salary provider. This results in a very large number of employees and a relatively small total employee salary expense, giving a very small annual average employee salary.

To assess whether the figure for average employee salary was a suitable one, it was compared to the annual minimum wage of the country the company performs in, retrieved from Eurostat. If the average pay was below the minimum wage threshold for any of the 6 time periods analysed, then the company was discarded. Following this method, 23 of the 71 companies had to be removed, which explains the reduction of the total number of firms analysed to 48.

Dependent Variables

Return on Assets (ROA) is the dependent variable in this research. Prior academic studies have shown that ROA is one of the most common proxies to determine company performance, and

12

has been used as such in most of the existing empirical tests on pay-ratio and firm performance (e.g. Crawford et al., 2014; Henderson and Fredrickson, 2001). It is the ratio of net income to total assets and it indicates how profitable a company is relative to its total assets. It shows shareholders how efficient management is at using the available assets to generate earnings.

As part of the robustness test two more regressions are performed, this time using different proxies for firm performance, specifically Return on Equity (ROE) and Return on Invested Capital (ROIC). The former is the ratio of net income to total equity and it reveals how much profit a company generated with the money invested by shareholders. Similarly, the latter is the ratio of net income minus dividends to total capital and it shows how efficient a company has been at allocating its capital to generate profits. The reason for choosing these variables is the high correlation they have with ROA, which equals to 0.8854 for ROIC and 0.6546 for ROE, as shown in Appendix C. In addition, although ROA is generally viewed as a better indicator for firm performance, these 2 proxies are somewhat more related to shareholders as it gives them an idea of where their money has been invested and how profitable those investments were. Given that part of the goal of this research is to uncover the value that CEO-pay-ratio adds to them, ROE and ROIC serve as adequate test variables.

Control Variables

In order for the regression to represent more accurately the relationship between CEO-pay-ratio and company performance, 2 control variables have been included in the model. These will reduce the influence of omitted variable bias in the test which will result in a more accurate relationship.

The first control variable is firm size (Size), measured using number of employees as proxy. The reason for including this variable in the regression is that existing studies (e.g. Crawford et al., 2014) have found a significant relationship between performance and firm size and all prior research analysed includes it, which is why this paper will too. Number of employees is chosen as it has often been used as proxy for firm size (e.g. Cowherd and Levine, 1992; Faleye et al., 2013).

The second control variable is firm risk (Risk), measured as the standard deviation of stock price daily returns. Risk is a variable that most studies include as a control when assessing the significance of pay-ratio on firm performance, and volatility in stock price return is the most common method of capturing this (e.g. Crawford et al., 2014; Faleye et al., 2013).

13 Dummy Variables

Given Faleye et al.’s (2013) finding that relative pay is higher among firms in homogeneous industries, 7 dummy variables (Ind1-Ind7) each representing one of the industries present in the sample have been included in the regression. It is important to note that the total number of industries present in the sample is 8, but the number of dummies is 𝑘 − 1, where k represents the levels of the original categorical variable, which in this case is Industry. This is done so that the coefficients of the variables included in the regression will show the effect an industry sector has on the dependent variable, relative to the omitted one, which is used as benchmark. For this reason it is a rule of thumb to use as the non-coded variable the one with the most observations. Looking at Appendix B, Financials is the sector with the highest number of firms, thus it will be used as the non-coded variable, to which the other 7 industries will compare to. From the sample, the industries which are predicted to have the most skilled workers and thus the least interchangeable ones are primarily Financials and Information Technology, as well as Energy. All the remaining ones are assumed to be more homogeneous.

A robustness test will also be performed by replacing the dummy variables representing the industries, with others representing the country (Ctry1-Ctry5) in which each firm operates (see Appendix F for the respective country per dummy variable). This is to see the interaction between the performance of the companies and the country they operate in, and if any discrepancies are present compared to the previous regressions. The country with the most observations is the United Kingdom (see Appendix B), which will be non-coded in the regression and used as benchmark, to which the other countries will compare to.

3.4 Research Model

As stated above, the data used in this research represents a panel dataset. Panel data, also referred to as longitudinal or cross-sectional time-series data, is a dataset in which the behaviour of entities, in this case firms, can be observed over time (Torres-Reyna, 2007). Panel data provides a solution for tackling the consequences of econometric problems related to the correlation between omitted variables and explanatory variables that often arises in empirical studies. It allows you to control for variables you cannot observe or measure, or that change over time but not across entities, thus accounting for individual heterogeneity (Torres-Reyna, 2007).

The two most used models to analyse panel data are Fixed Effects and Random Effects. While fixed effects assumes that entities have unique time-invariant characteristic which may

14

bias the predictor or outcome variables, random effects assumes that the variation across entities is random and uncorrelated with other variables included in the model:

“The crucial distinction between fixed and random effects is whether the unobserved individual effect embodies elements that are correlated with the regressors in the model, not whether these effects are stochastic or not” (Greene, 2008, p.183).

In order to evaluate which statistical model best corresponds to the data a Hausman test can be performed. The test evaluates whether the covariance between the independent variable (PayRatio) and the error term (𝜀_𝑖𝑡) is equal to zero. If this is the case then it can be said that both the fixed and random effects predictors are consistent, but that the standard error of the random effects predictor is smaller than the fixed effects one, thus suggesting that using the random effects model is more appropriate. If the Hausman test suggests that the covariance is not equal to zero, then only the fixed effect model is suitable.

The outcome of the Hausman test suggests that the random effect model best fits the data. Thus the statistical model used in this paper is the random effects model. Since there is no clear method to test for heteroskedasticity when using random effects, the best way is to control for it by using robust standard errors. In addition, it is important that no perfect multicollinearity is present and that all regressors are exogenous. As shown in Appendix C none of the independent variables have a perfect correlation, which indicates that perfect multicollinearity is not present.

To test the hypothesis, the following regression will be performed:

(1) 𝑅𝑂𝐴_𝑖𝑡 = 𝛼 + 𝛽₁(𝑃𝑎𝑦𝑅𝑎𝑡𝑖𝑜)𝑖𝑡+ 𝛽2(𝑆𝑖𝑧𝑒)𝑖𝑡+ 𝛽3(𝑅𝑖𝑠𝑘)𝑖𝑡+ 𝛽4(𝐼𝑛𝑑1)𝑖𝑡+ 𝛽5(𝐼𝑛𝑑2)𝑖𝑡+ 𝛽₆(𝐼𝑛𝑑₃)_𝑖𝑡+ 𝛽₇(𝐼𝑛𝑑₄)_𝑖𝑡+ 𝛽₈(𝐼𝑛𝑑₅)_𝑖𝑡+ 𝛽₉(𝐼𝑛𝑑₆)_𝑖𝑡+ 𝛽₁₀(𝐼𝑛𝑑₇)_𝑖𝑡+ 𝜀_𝑖𝑡

Where ROA is the dependent variable and PayRatio is the independent variable, measured by dividing total CEO compensation by average employee salary. Size and Risk are the control variables – the former is represented by the number of employees at the end of the fiscal year, while the latter is measured as the standard deviation of daily stock price return. Ind1 to Ind7 are dummy variables representing the 8 different industries. Again, one of the dummy variables has been dropped (Financials) as the other ones present in the model will compare to it. The Betas (β) are regression coefficients of company i at time t and they measure the change in performance per unit change of PayRatio, Size or Risk. The regression coefficients of the dummy variables show the effect an industry sector has on the depended variables, compared

15

to the Financials sector. This means that industries with a positive coefficient perform better relative to Financials, while the ones with a negative one will perform worst. α is the constant and ε is the error term which captures the deviation of the observed values from the theoretical ones.

Table 1

Dummy Variables

The table shows the respective industry sector per dummy variable. Note that Financials is not coded in the regression.

Variable Industry

Ind1 Consumer Discretionary

Ind2 Consumer Staples

Ind3 Energy

Ind4 Industrials

Ind5 Information Technology

Ind6 Materials

Ind7 Utilities

Ind8 (not coded) Financials

For the robustness check, three additional regressions are performed to test the consistency of the results. Regression (2) and (3) will be performed by changing the dependent variable representing performance to see what the impact will be on the initial results. All other variables are kept unchanged. Regression (4) will have the dummy variables representing the industries replaced by the countries the firms operate in.

The updated regressions to test the hypothesis are the following:

(2) 𝑅𝑂𝐸𝑖𝑡 = 𝛼 + 𝛽1(𝑃𝑎𝑦𝑅𝑎𝑡𝑖𝑜)𝑖𝑡+ 𝛽2(𝑆𝑖𝑧𝑒)𝑖𝑡+ 𝛽3(𝑅𝑖𝑠𝑘)𝑖𝑡+ 𝛽4(𝐼𝑛𝑑1)𝑖𝑡+ 𝛽5(𝐼𝑛𝑑2)𝑖𝑡+ 𝛽6(𝐼𝑛𝑑3)𝑖𝑡+ 𝛽7(𝐼𝑛𝑑4)𝑖𝑡+ 𝛽8(𝐼𝑛𝑑5)𝑖𝑡+ 𝛽9(𝐼𝑛𝑑6)𝑖𝑡+ 𝛽10(𝐼𝑛𝑑7)𝑖𝑡+ 𝜀𝑖𝑡 (3) 𝑅𝑂𝐼𝐶_𝑖𝑡 = 𝛼 + 𝛽₁(𝑃𝑎𝑦𝑅𝑎𝑡𝑖𝑜)_𝑖𝑡+ 𝛽₂(𝑆𝑖𝑧𝑒)_𝑖𝑡+ 𝛽₃(𝑅𝑖𝑠𝑘)_𝑖𝑡+ 𝛽₄(𝐼𝑛𝑑₁)_𝑖𝑡+ 𝛽₅(𝐼𝑛𝑑₂)_𝑖𝑡+ 𝛽₆(𝐼𝑛𝑑₃)_𝑖𝑡+ 𝛽₇(𝐼𝑛𝑑₄)_𝑖𝑡+ 𝛽₈(𝐼𝑛𝑑₅)_𝑖𝑡+ 𝛽₉(𝐼𝑛𝑑₆)_𝑖𝑡+ 𝛽₁₀(𝐼𝑛𝑑₇)_𝑖𝑡+ 𝜀_𝑖𝑡 (4)𝑅𝑂𝐴_𝑖𝑡 = 𝛼 + 𝛽₁(𝑃𝑎𝑦𝑅𝑎𝑡𝑖𝑜)𝑖𝑡+ 𝛽2(𝑆𝑖𝑧𝑒)𝑖𝑡+ 𝛽3(𝑅𝑖𝑠𝑘)𝑖𝑡+ 𝛽4(𝐶𝑡𝑟𝑦1)𝑖𝑡+ 𝛽₅(𝐶𝑡𝑟𝑦2)𝑖𝑡+ 𝛽6(𝐶𝑡𝑟𝑦3)𝑖𝑡+ 𝛽7(𝐶𝑡𝑟𝑦4)𝑖𝑡+ 𝛽8(𝐶𝑡𝑟𝑦5)𝑖𝑡+ 𝜀𝑖𝑡

16

3.5 Descriptive Statistics

The total number of firms analysed in this study is 48 with observation ranging from 2009 to 2014, producing a total of 288 firm-year observations. A short analysis of the variables and the firms used in this research will follow.

Table 2

Summary Statistics

This table shows descriptive statistics for key variables. Data has been retrieved from Capital IQ and DirectorInsight.

Variable Observations Mean Std. Dev. Min Median Max

ROA 288 0.04 0.04 -0.07 0.03 0.21 ROE 288 0.11 0.16 -1.09 0.11 0.57 ROIC 288 0.07 0.07 -0.23 0.06 0.35 Pay Ratio 288 81.63 104.15 5.30 39.67 879.44 Size 288 41,245.61 55,978.08 45.00 12,653.00 295,061 Risk 288 0.03 0.06 0.01 0.02 0.60

An interesting figure in Table 1 is the average value of Pay Ratio, equal to 81.63. In the introduction to this paper, previous studies were presented showing that the average CEO-pay-ratio was estimated to be between 204 and 500 to 1 in the years 2000 to 2013. Here, except for the maximum value equal to 879.44, a much smaller ratio is uncovered showing that, with regards to the European companies analysed, the average pay ratio is not as drastic as it was estimated to be for US companies. Looking at the variable Size, it can be seen that most of the companies in the sample are large-size firms as the average number of employees is 41,245.61 with a maximum value of 295,061. No firm is significantly risky as the average value for Risk is 0.03, representing a volatility in the stock price equal to 0.09%. A firm is considered risky when the volatility of its stock is above 50%. Although the maximum value, equal to 0.6, or 36%, is significantly larger than the average one, it is still below 50% thus it is safe to assume that no firm has a significant risk tied to its stock price volatility.

17

Fig.1 Representation of Country and Industry weights for the firms in the sample.

Figure 1 shows an overview of the countries and industries the firms in the sample fall into. 75% of the companies come from The Netherlands and United Kingdom – the reason for this is that the company that developed the platform DirectorInsight is an Anglo-Dutch one, and the first indexes to be covered were the Dutch one, specifically the AEX, and the British one, the FTSE 100. The country with the least observations is France, followed by Switzerland, Belgium, and Germany. Although these countries are all Western European and share similarities which have been discussed above, the unbalanced weights might have an effect on the end results, as no country is the same to another. For this reason, a robustness test incorporating the 6 countries as dummy variables is performed, in order to see what the individual relationship is with company performance. The same holds for the industry sector, as it was discussed that more homogeneous industries, such as consumer staples, materials, or consumer discretionary where workers are on average less skilled and more interchangeable, pay ratio may be higher than in industries such as Financials or Information Technology, where workers are more skilled and thus less interchangeable. Here, the fact that weights differ by a significant amount and that industries such as Energy, Information Technology and Materials have a small amount of observations, representing 2%, 2% and 4% of the total respectively, might have an influence on the end results, with an observed value for Pay Ratio resembling more the one for the Financial, Industrial, and Utilities sector, which combined represent 74% of total observations. To see the number of observations per country and industry please refer to Appendix B.

18

4. Empirical Results

The regressions performed in Table 3 are aimed at testing the hypothesis that a positive relationship is present between CEO-pay-ratio (PayRatio) and company performance against the alternative hypothesis where no relationship is present. Results are reported for the original equation (1) and for the two alternative ones used for the robustness check (2) (3).

Table 3

The table shows the results of the regression for Pay Ratio on ROA, ROE, and ROIC.

ROA ROE ROIC

(1) (2) (3) PayRatio 0.0000732* _0.000308*** _0.000120** (0.0000375) (0.0000837) (0.0000494) Size -0.000000103** _-0.000000303** _-0.000000226*** (4.87e-08) (0.000000133) (8.42e-08) Risk -0.0118 -0.0965 -0.0159 (0.00885) (0.0733) (0.0127) Ind1 0.0830** _0.153*** _0.124** (0.0330) (0.0542) (0.0536) Ind2 0.0451*** _{0.0601 0.0486}** (0.0127) (0.0712) (0.0202) Ind3 0.0556*** _{-0.0175 0.0558}*** (0.00656) (0.0185) (0.0131) Ind4 0.0413*** _{0.0206 0.0448}* (0.0144) (0.0509) (0.0243) Ind5 0.0549*** _0.122*** _0.0530*** (0.00302) (0.0147) (0.0103) Ind6 0.0663*** _{0.0442 0.0691}*** (0.0169) (0.0316) (0.0236) Ind7 0.0274*** _0.0615** _0.0340** (0.00418) (0.0247) (0.0135) Constant 0.00893*** _0.0638*** _0.0319*** (0.00314) (0.0149) (0.0104) Observations 288 288 288 R-squared 0.4673 0.1441 0.3574

Standard errors in parentheses

19 Regression 1

To estimate the relationship, regression (1) uses ROA as a measure for company performance. The results show that PayRatio is statistically significant at a 10% significance level and that the relationship with company performance is positive, supporting the hypothesis.

Thus, at a 10% significance level there is enough evidence that company performance is positively related to CEO-pay-ratio, and the null hypothesis is rejected.

However, the Beta of PayRatio shows that a 1 unit increase in the variable leads to a 0.0000732 increase in ROA. Therefore, the effect that PayRatio has on company performance (ROA) is almost negligible as the coefficient is approximately equal to 0. This means that although the significance and the coefficient sign of PayRatio supports the hypothesis, the extent to which it is related to company performance, which is the research question of this paper, is almost null. In addition, according to the R-squared value, the model can explain 46.73% of the variation in firm performance. Although this represents the highest value between the 3 regressions performed, it shows that the model explains less than half of the variation in firm performance. This may be due to variables which have not been included in the model, but would have been significant for the analysis, resulting in a potential omitted variable bias. Also, it is important to consider that in multiple regression, the higher the number of variables used in the regression, the higher the R-squared value will be. This, however, does not necessarily mean that the variables included actually help to explain a bigger portion of the variation.

The constant is significant at a 1% level of significance. Size is also found to be statistically significant at a 5% significance level, which is consistent with the prediction of the control variable. However, the relationship here is negative and the Beta is approximately equal to 0, suggesting a negligible effect size. Risk is found to be not significant.

All dummy variables representing the eight different industries are found to be significant at a 1% significance level except for Ind1, which is significant at a 5% significance level. The Betas of these variables included in the regression show the effect an industry sector has on the dependent variable, relative to the Financials sector. Since all coefficients are positive, ceteris paribus, all industries perform significantly better than Financials, with Consumer Discretionary (Ind1) being the top performer, with a coefficient of 0.083 and Utilities (Ind7) being the lowest one, with a coefficient of 0.0274. It is however important to consider the imbalance in industry weights as a limitation, which might have shaped the results differently from the true ones.

20

Overall, the findings of the analysis, the goal of which was to test the relationship between company performance measured using ROA, and pay-ratio, have revealed that pay-ratio is a significant factor to company performance, and that the two variables are positively related. Although, statistically speaking, these conclusions support both the prediction and the notions of tournament theory analysed in past research, it is important to note that the effect size of pay-ratio on company performance is almost null. The coefficient of the main independent variable is approximately equal to 0 and the significance is low (10%). This means that, theoretically, a shareholder who would base his decision on this research, would come to the conclusion that altering the CEO-pay-ratio would not result in a significant increase (or decrease) in company performance. Of course, this claim only applies to the findings of this paper – the underlying question remains an open one that only further research will be able to answer.

Still, the works of Faleye et al. (2013) and Burns et al. (2013), who both discovered a positive relationship between the two variables, can help to explain why a positive and significant result is witnessed. Within the business environment, (Western) Europe is characterized by cultural influences which include the desirability of competition and the acceptance of power, which combined generate a belief that income differentials are fair, as predicted by Burns et al. (2013). Top managers are effectively motivated by the tournament incentive mechanism which results in enhanced firm performance.

4.1 Robustness Check

The robustness check is set to test the outcome of the research using different variables to see if the initial results still stand under different conditions. Results for Regression (2) and (3) are presented in Table 3. Results for Regression (4) are presented in Appendix E.

Regressions (2), (3) & (4)

The first robustness check uses ROE as a measure for company performance. As with regression (1), the relationship with company performance is positive and results show that PayRatio is statistically significant at a 1% significance level. The Beta of PayRatio shows that a 1 unit increase in PayRatio leads to a 0.000308 increase in ROE, which, although still very close to 0, is higher than under ROA. This and the higher significance indicate that ROE is more responsive than ROA to changes in PayRatio.

The second robustness check uses ROIC as a measure for company performance. Similar to regression (1) and (2), the relationship with company performance is positive and

21

results show that PayRatio is statistically significant at a 5% significance level. The Beta of PayRatio shows that a 1 unit increase in PayRatio leads to a 0.000120 increase in ROIC, which, just like with ROE, is higher than under ROA. However, the dependent variable here is not as significant as when ROE is used, but the outcome is very similar, since results indicate that ROIC is more responsive than ROA to changes in PayRatio, although not as much as with ROE.

The results of the third robustness check can be found in Appendix E. Nothing new is discovered here, as PayRatio is still significant, at a 1% significance level, and the relationship between the dependent variable and the main independent one is still positive. Again, although the significance level increased compared to regression (1), the effect size is almost negligible as the Beta of PayRatio is approximately 0. No country is found to be statistically significant.

Overall, the robustness check shows that the results still hold under different performance measures and using other dummy variables. Some values differ between the four regressions and variables may have different significance levels, but the key variable, PayRatio, is found to be positive and significant throughout all four tests, which is the most important result. It is interesting to note that PayRatio is most significant in regression (2) where ROE is used as dependent variable, followed by regression (3), which uses ROIC, and regression (1) which uses ROA. However, in regression (4), where the dummy variables are replaced, ROA is significant at a 1% significance level, just as ROE in regression (2).

5. Conclusion

The research question of this paper is: To what extent does CEO-pay-ratio affect company performance? With CEO-pay-ratio calculated by dividing the CEO total annual granted compensation by the average employee salary and performance measured using Return on Assets. In order to answer this question an analysis on 48 European listed companies active in 6 countries and 8 different industries was made, with data ranging from 2009 to 2014. Data is annual and composes a balanced panel data set which was tested using the random effect model. As predicted by the hypothesis, the results of this analysis show that a slightly positive relationship is present between the two variables, supporting the notions of tournament theory analysed in past research. This means that an increase in CEO-pay-ratio leads to an increase in firm performance. However, since the coefficients of PayRatio are all approximately equal to 0, the effect size of this positive relationship is almost negligible, and the information it

22

provides to shareholders is that relative pay is probably not a crucial factor for firm performance.

The most important limitation of this study is the sample – specifically the number of firms and industries analysed and their related countries of origin. This paper only includes companies for which data on CEO compensation was available on the platform DirectorInsight, combined with the data on company performance and average employee salary taken from S&P Capital IQ. This significantly decreased the number of possible firms to be analysed and could have led to biased results, as not all European listed companies where present. This is further supported by the figure of R-squared, equal to 46.73%, which leads to the speculation that omitted variable bias is present. The effect of having more industries and more countries in scope would have most likely changed the outcome of the results. However, it is important to consider the validity of the data used – although DirectorInsight may not have had data for all European listed companies, the data that was present is unique, as it is the analysis performed to arrive to the end figure for CEO total granted compensation. Another limitation is that this study cannot account for the psychological factors affecting both top management and lower level employees and the asymmetric information present within a company. For example, one of Faleye et al.’s (2013) argument is that performance is positively related to pay ratio if a firm has fewer employees who are informed on executive pay. This data was unavailable for the companies in scope.

For further research, it would be interesting to see whether the relationship still holds if all European listed companies and industries where analysed, to see how the presence of Scandinavian, Southern, and Eastern European countries would affect the final results. Also, in order to fully establish the utility of the CEO-pay-ratio figure to shareholders, it would be beneficial to break down the CEO total compensation into its components and create multiple pay-ratios, to see how each individual one relates to company performance.

23

References

Akerlof, G. A., & Yellen, J. L. (1990). The Fair Wage-Effort Hypothesis and Unemployment. The Quarterly Journal of Economics, 105(2), 255.

Burns, N., Minnick, K., & Starks, L. T. (2013). CEO Tournaments: A Cross-Country Analysis of Causes, Cultural Influences and Consequences. SSRN Electronic Journal SSRN Journal.

Conyon, M. J., & Murphy, K. J. (2000). The Prince and the Pauper? CEO Pay in the United States and United Kingdom. The Economic Journal, 110(467), 640-671.

Cowherd, D. M., & Levine, D. I. (1992). Product Quality and Pay Equity between Lower- Level Employees and Top Management: An Investigation of Distributive Justice Theory. Administrative Science Quarterly, 37(3), 524.

Crawford, S., Nelson, K. K., & Rountree, B. (2014). The CEO-Employee Pay Ratio. SSRN Electronic Journal SSRN Journal.

Davis, A., & Mishel, L. (2014, June 12). CEO Pay Continues to Rise as Typical Workers Are Paid Less. Retrieved from http://www.epi.org/publication/ceo-pay-continues-to-rise/

Executive Paywatch. (2015). Retrieved from

http://www.aflcio.org/Corporate-Watch/Paywatch-2015

Faleye, O., Reis, E., & Venkateswaran, A. (2013). The determinants and effects of CEO– employee pay ratios. Journal of Banking & Finance, 37(8), 3258-3272.

Farmer, M., Archbold, S., & Alexandrou, G. (2013). CEO Compensation and Relative Company Performance Evaluation: UK Evidence. Compensation & Benefits Review, 45(2), 88-96.

Green, J. R., & Stokey, N. L. (1983). A Comparison of Tournaments and Contracts. Journal of Political Economy, 91(3), 349-364.

Greene, W. H. (2008). Econometric analysis. Upper Saddle River, NJ: Prentice Hall.

Henderson, A. D., & Fredrickson, J. W. (2001). Top Management Team Coordination Needs And The Ceo Pay Gap: A Competitive Test Of Economic And Behavioral Views. Academy of Management Journal, 44(1), 96-117.

Lazear, E. P., & Rosen, S. (1981). Rank-Order Tournaments as Optimum Labor Contracts. Journal of Political Economy, 89(5), 841-864.

Lazear, E. P. (1989). Pay Equality and Industrial Politics. Journal of Political Economy, 97(3), 561-580.

Maloney, M. T. (2003, December 1). Rank Order Tournaments-Lazear & Rosen. Lecture presented in Clemson University.

24

Rosen, S. (1986). Prizes and incentives in elimination tournaments. American Economic Review, 76, 701-715.

U.S Securities and Exchange Commission. (2015, August 5). Rule Implements Dodd-Frank Mandate While Providing Companies with Flexibility to Calculate Pay Ratio [Press release]. Retrieved from https://www.sec.gov/news/pressrelease/2015-160.html

Smith, E., & Kuntz, P. (2013, May 2). Disclosed: The Pay Gap Between CEOs and Employees. Retrieved from http://www.bloomberg.com/bw/articles/2013-05-02/disclosed-the-pay-gap-between-ceos-and-employees

Torres-Reyna, O. (2007, December). Panel Data Analysis Fixed and Random Effects using Stata. Lecture presented in Princeton University.

Wade, J. B., O'reilly, C. A., & Pollock, T. G. (2006). Overpaid CEOs and Underpaid Managers: Fairness and Executive Compensation. Organization Science, 17(5), 527-544.

25

Appendix A

List of companies, country and industry

This table shows the complete list of companies included in the research, with the respective country and Industry.

Country Company Industry

Belgium Ackermans & Van Haaren N.V. Financials

Ageas N.V. S.A. Financials

KBC Group N.V. Financials

Elia System Operator S.A. Utilities

France GDF SUEZ S.A. Utilities

Germany Commerzbank AG Financials

Deutsche Bank AG Financials

Deutsche Boerse AG Financials

E.ON SE Utilities

RWE AG Utilities

Netherlands Accell Group N.V. Consumer Discretionary

Beter Bed Holding N.V. Consumer Discretionary

Heineken N.V. Consumer Staples

Nutreco N.V. Consumer Staples

Royal Wessanen N.V. Consumer Staples

Sligro Food Group N.V. Consumer Staples

AEGON N.V. Financials

BinckBank N.V. Financials

Eurocommercial Properties N.V. Financials

ING Group N.V. Financials

Aalberts Industries N.V. Industrials

Ballast Nedam N.V. Industrials

Kendrion N.V. Industrials

Royal BAM Group N.V. Industrials

TKH Group N.V. Industrials

Nedap - Nederlandsche Apparatenfabriek N.V. Information Technology

Switzerland Credit Suisse Group AG Financials

Geberit AG Industrials

United Kingdom GKN PLC Consumer Discretionary

BAT - British American Tobacco PLC Consumer Staples

SABMiller PLC Consumer Staples

BG Group PLC Energy

3i Group PLC Financials

Barclays PLC Financials

HSBC Holdings PLC Financials

LSE - London Stock Exchange Group PLC Financials RBS - The Royal Bank of Scotland Group PLC Financials

Schroders PLC Financials

Standard Chartered PLC Financials

Ashtead Group PLC Industrials

BAE Systems PLC Industrials

Bunzl PLC Industrials

BHP Billiton PLC Materials

Mondi PLC Materials

Centrica PLC Utilities

National Grid PLC Utilities

Severn Trent PLC Utilities

26

Appendix B

Table B1 Industries

This table shows the eight different industries in which the firms analysed fall into, the number of firms per industry, and their relative weight.

Industry Firms Weight

Consumer Discretionary 3 0.063 Consumer Staples 6 0.125 Energy 1 0.021 Financials 18 0.375 Industrials 9 0.188 Information Technology 1 0.021 Materials 2 0.042 Utilities 8 0.167 Total 48 1 Table B2 Countries

This table shows the six different countries in which the firms analysed fall into, the number of firms per country, and their relative weight.

Country Firms Weight

Belgium 4 0.083 France 1 0.021 Germany 5 0.104 Netherlands 16 0.333 Switzerland 2 0.042 United Kingdom 20 0.417 Total 48 1 Appendix C Variable Correlations

The table shows the correlations between the main variables of this study

ROA ROE ROIC Pay Ratio Size Risk

ROA 1.0000 ROE 0.6546 1.0000 ROIC 0.8854 0.7110 1.0000 Pay Ratio 0.2239 0.2009 0.1842 1.0000 Size -0.1897 -0.0609 -0.2253 0.4632 1.0000 Risk -0.1485 -0.1316 -0.1367 -0.0409 0.0735 1.0000

27

Appendix D

Summary Statistics Compensation

This table shows descriptive statistics for the level of compensation of CEOs and average employee. Data for CEO compensation has been retrieved from DirectorInsight and it is composed of cash base compensation and both long term and short term management incentive plans, including stock options, performance units and share plans, restricted stock, phantom stock and cash bonus. Data for average employee compensation has been retrieved from Capital IQ and it is calculated by taking salary expenses reported on each company’s income statement and dividing it by the total number of employees at the end of the fiscal year. Data is all in Euro (€).

Variable Observations Mean Std. Dev. Min Median Max

CEOcomp 288 4,978,661.49 5,829,910.16 449,000 3,075,926 36,049,266 AvgEmplComp 288 72,534.31 46,579.22 17,972 61,041 320,613 Appendix E

The table shows the results of the regression for Pay Ratio on ROA, using the 6 countries as dummy variables.

ROA (4) PayRatio 0.0000802*** (0.0000196) Size -0.000000186** (8.26e-08) Risk -0.0128 (0.0207) Ctry1 -0.0286 (0.0197) Ctry2 0.0152 (0.0377) Ctry3 -0.0249 (0.0177) Ctry4 -0.000358 (0.0124) Ctry5 0.0334 (0.0263) Constant 0.0434*** (0.00935) Observations 288 R-Squared 0.2365

Standard errors in parentheses

28

Appendix F

The table shows the respective country per dummy variable used in the robustness test. Note that United Kingdom is not coded in the regression.

Variable Country Ctry1 Belgium Ctry2 France Ctry3 Germany Ctry4 Netherlands Ctry5 Switzerland

29

Internship Thesis Report 1. Introduction

After having completed two and a half years of the Bachelor in Economics and Business, with specialization in Finance, at the University of Amsterdam, I felt like I had a strong theoretical knowledge about the business world, with particular emphasis on the financial sector. The material learned through lectures, tutorials, group works and exams was very detailed and, in some cases, complicated. But all was essential for the shaping of an unknowledgeable individual, as I was, into gaining a strong and vast understanding of concepts related to economics, business, or finance. Of course, all these concepts were learned mainly through books, and even though practical examples were always there to back the theory up, a simple violation of an assumption could have had significant implications in a real life context. For this reason, practical experience is as important as the theoretical one, which is why I decided to put the knowledge gained through university into practice by pursuing an internship during the second semester of my final year as a bachelor student. I was accepted to work at AMA Partners as a research analyst intern. Given the high potential which AMA Partners had as a start-up company within a niche market, and the room for development it offered to its interns, I believed that this would have been a great opportunity for me to see how a small start-up company operates to reach its long term goal of being a well-established firm. To take full advantage of the opportunity, I decided to write my thesis based on the work performed there. This got me more involved in the daily work I performed and helped me to get a deeper understanding of it.

2. Company Description and Strategy

AMA Partners is an independent advisory business, which provides insight on corporate governance and executive remuneration to its users. The target audience is mainly board of directors and investors seeking information on these topics. The company headquarters are situated in Amsterdam and the information provided concerns all of Europe. The reason for this comes from the information which is provided by their main product – an online platform called DirectorInsight. This unique data technology platform tracks the composition of most of Europe’s publicly listed companies’ board of directors, and for each it gives a detailed insight on all board members remuneration, including both the executives and the non-executives directors, starting from 2008. In addition, DirectorInsight doesn’t limit itself to the simple composition of a company’s board of directors, it also gives further information on the individuals past (and future) positions, and it includes both a short biography and a profile

30

picture for each individual. All of this is done to help shareholders, together with a company’s appointments committee, to make an informative decision when having to appoint a new director to the company’s board, or replace an existing one. For example, the responsibilities of a CEO are significantly high and have the power to affect everyone operating in, or with, the company, which make the appointment of it a very delicate action. If a shareholder has access to a candidates past experience, as it would by utilizing DirectorInsight, this action may be simplified and, most importantly, the chances of the action resulting in a successful one would increase.

The major strength of the platform DirectorInsight, however, is the information it provides all its users regarding the remuneration of both the board of directors and the executives of an organization. The reason for this is the unique analysis performed on each company’s remuneration plans. All salaries of the individuals present on Directorinsight are composed of cash base compensation and both long term and short term management incentive plans, including stock options, performance units and share plans, restricted stock, phantom stock and cash bonus, all of which are closely analysed to construct the actual annual realized compensation. This provides a high degree of transparency, which tied to the quality of the data offers the shareholders the perfect scenario to make an informative decision on remuneration packages. The idea here is that performance is not as much absolute as it is relative, just like remuneration. Companies have what they call a ‘peer group’, which constitute a group of companies to which the organisation can relate to when deciding the level of performance desired to attain or the monetary amount to invest in the compensations of its highest representatives. Usually, peer group companies are those with similar size and value of assets, which operate in the same industry. Thanks to DirectorInsight an organisation can compare itself to its peer group to determine both the level of performance and the compensation structures relative to its competitors in the market, giving every user a significant advantage.

Although, during my internship, the total number of European firms in scope was approximately 1,200, the goal of AMA Partners is to eventually cover all the European indexes, including small, mid and large cap, in order to include all listed companies present in Europe. If this was the case, the number of directors present in DirectorInsight, which at the time was roughly 30,000, would significantly increase and so would the transparency within regulated markets. In fact, the reason behind AMA Partners business idea is tightly related to the changes which corporate governance and executive remuneration went through after the financial crisis of 2008. Markets became a lot more regulated and transparency was key. This called for

31

companies to reform their policies, establish a remuneration committee, if one wasn’t present already, and engage in sustainable and responsible pay practices. All this to improve risk management and the relationship between a company and its shareholders. In 2014, AMA Partners saw a great potential within this niche and decided to take advantage of it, starting from the bottom, with a very clear vision in mind: specializing in executive remuneration and corporate governance to provide unparalleled market insight through transparent, easily accessible and unique data. This is what distinguishes it from its competitors.

Well established firms such as Towers or Mercer provide advisory on remuneration as part of a broad range of services which they also provide. The competitive advantage of AMA Partners stands right here. In one word it can be described as specialized. The attention is never drawn away from the goal of the organization. While in the US the popularity of small to medium firms specializing in executive compensation or corporate governance structure has been growing already for years, this is not the case for Europe. AMA Partners represents one of the first European, small-size, firm specialized solely on remuneration and corporate governance. This gives it a significant importance, as they have what in theory is known as the first mover advantage. Gathering and analysing this kind of information, in these quantities, and with this degree of precision can take years, which is why in the past two years AMA has been collecting and processing data on this matter. However, given that well established firms in this field, which go beyond the two sated above, are present, the importance of coexistence within this sector is key for this organizations development.

AMA Partners management structure is very simple, as it is flat. This is the opposite of a hierarchical structure, as there are no levels of middle management between the employees and the managing director. Given that the organisation is a start-up, this is probably the most effective management system it could have implemented, as it helps improve the quality of communication between staff members and management – a crucial aspect for any company, especially a start-up. Moreover, a decentralized decision making process promotes employee involvement and helps create a feeling of equity. Although hired as an intern, I was from the start of the internship directly contributing to the end product, as were all of my colleagues. This feeling of involvement comes with responsibilities, and the combination of these two factors significantly increase the desire for working at the best of your abilities. The research team is responsible for collecting information regarding a company’s remuneration packages and everything related to its corporate governance structure, together with all relevant financial information. A development team is also present, and their duty is to constantly update and improve the platform DirectorInsight. This represents the structure of AMA Partners. However,