The executive to average employee pay ratio, cause and effect : empirical evidence from the Netherlands

(1)

The executive to average employee pay ratio, cause and effect:

Empirical evidence from the Netherlands

Author: Lotte Vermeij – 10004342 Supervisor: S. Dominguez Martinez

University of Amsterdam MSc. Business Economics – Track Organization Economics Thesis credits: 15 ECTS

(2)

Abstract

This research aims to shed light on the determinants and the performance effects of the CEO-employee pay ratio in the Netherlands. With the use of a cross-sectional panel dataset of the 50 largest firms listed in the Netherlands for the years 2010-2014, the research is empirically approached. The determinants are found to be largely dependent on the relative bargaining power of the executive board versus the employee. Using these determinants, the optimally expected pay ratio is predicted. The deviation from the predicted pay ratio is then used to identify its effect on firm performance, in terms of revenue per employee. No significant effect is found. Furthermore, the effect on firm performance of the pay ratio itself is tested. When controls for visibility of the pay ratio are included, the effect of the pay ratio on firm performance is slightly negative, indicating that increasing the gap between the CEO and employee pay worsens firm performance.

List of abbreviations

AEX Amsterdam Exchange Index AMX Amsterdam Midcap Index AScX Amsterdam Smallcap Index CBS Central Bureau of Statistics CEO Chief Executive Officer FTE Full Time Equivalent HHI Hirshman Herfindahl Index ROA Return on Assets

SB Supervisory Board SIC Standard Industry Code

TDC Total Direct Compensation (base salary, short- and long-term bonus) U.S. United States

(3)

1. Introduction

Popular press has been emphasizing the growing gap between the compensation of the chief executive officer (‘CEO’) and the salary of an average employee, referred to as the pay ratio (Dekker and Van Uffelen, 2014). As wage disparity between the top and the bottom of society is growing, questions about its consequences become more prominent. Public disbelief and outrage on this topic may have consequences for firm performance, as employee motivation may be affected by the growing income inequality within the firm.

Growing pay ratios across the world have called for legislation intended to make the differences within the firm more visible to the shareholder, thereby aiming to diminish the pay ratio through ‘public shaming’ (Eavis, 2015). For the past five years, the United States (hereafter ‘U.S.’) has been pushing to pass the Dodd Frank law, approved by the Security Exchange Committee (‘S.E.C.’) just this August, after a long, controversial discussion about its implications (Eavis, 2015). The law requires publicly listed companies in the U.S. to disclose the pay ratio, the ratio of top executives pay compared to the median employee. The European Union has been planning on passing similar laws, intended to give insight and openness on executive compensation practices and the pay ratio within the firm, thereby giving more information to outsiders such as investors or employees on whether compensation is equally distributed within the firm, or concentrated at the top (Eavis, 2015).

However, critics argue that besides reflecting pay practice, the ratio also reflects firm and industry characteristics, making it impossible to compare two stand-alone pay ratio values, as will likely happen after publication of such ratios (Gavett, 2014; Kelly and Seow, 2015). For example, companies from different industries are likely to have different pay ratios due to alternative workforce compositions. However, conclusive evidence on the exact determinants of the pay ratio is lacking, increasing the difficulty to draw conclusions regarding a company’s pay practice from publication of the ‘stand-alone value’ of the pay ratio (Kelly and Seow, 2015).

Besides, as said before, large pay ratios could affect employee morale and therefore motivation, which could lead to lower firm performance. However, there are also proponents that support the notion that large pay differentials increase employee motivation and effort in order to make a desired promotion and thereby a higher salary.

(5)

The increased motivation and effort could in turn lead to increased firm performance (Connelly et al., 2014).

The goal of this thesis is to shed light on the relationship between the executive to employee pay ratio and productivity. The main research question is therefore: To what extent does the pay ratio between top executives and the average worker affect company productivity? The literature provides reasons to believe that the relationship is not linear in the sense that a higher pay ratio always leads to a better or worse performance. But, the relationship is more likely to depend on the deviation from a predicted pay ratio deduced from firms with similar traits. For example, in bigger firms a larger pay ratio might be more generally accepted, and therefore have less effect on the productivity than a large pay ratio in smaller firms. This research will therefore create a model for the determinants of the pay ratio, and use this to predict a pay ratio for each individual firm. This predicted pay ratio is believed to coincide with the generally accepted pay ratio for each firm. The deviation from the predicted pay ratio is then used to describe the relationship between the pay ratio and company productivity.

The study will add to existing literature in several ways. First, it will complement the contradictory results published in the scarce existing body of literature regarding the relationship between the executive to average employee pay ratio and company performance (Connelly et al., 2013; Faleye et al., 2013; Shin et al., 2014). The scarceness of literature on this very specific topic is mostly due to the fact that there is limited public data on employee salaries. Average salaries can be calculated from information published in annual reports, but this information is often only disclosed on a voluntary basis, leading to selection bias. However, in the Netherlands, due to legislation, all companies listed on the Dutch stock exchange are required to disclose total labor force cost and the number of full time equivalents (‘FTE’). This gives the possibility for new, more reliable research. Besides, the three previous studies on this topic are based on data from the U.S. and South Korea, where pay ratios are relatively high compared to the Netherlands (Gavet, 2014). In those cases, higher pay ratios could be more generally accepted, and expected, allowing for different results concerning productivity effects of the pay ratio. Conducting this research in the Netherlands therefore explores a different market, with different cultural attitudes towards pay dispersion. A more in depth discussion of the cultural attitudes is provided in chapter 2, the background section of this thesis.

(6)

Second, this thesis will shed more light on the motivational aspects of the vertical pay disparity, the pay disparity between the average worker and executive, within the organizational environment. Most empirical research on relative pay within companies so far has been done concerning a more horizontal view, such as differences between executives working together in the same organization’s executive board (e.g. Bebchuk et al., 2011; Main et al, 1993) or between companies, not taking in account the different organizational ranks (Hibbs and Locking, 2000). Vertical pay disparity concerns the gap between different organizational levels, which is why the gap can show what an individual can make in the future after, for example, a promotion. A horizontal pay disparity considers functions that are similar in rank, and therefore focuses more on the fairness considerations of difference in pay for employees that are similar to each other. Since vertical pay disparity captures a different effect than the horizontal pay disparity, it is important to fill the gap in literature by analyzing the vertical pay differentials of the average employee to the top executive.

Last, investigating the determinants of the pay ratio will provide valuable insights to both policy makers and shareholders. Policymakers are pushing for mandatory pay ratio disclosure. However, it is unlikely that the pay ratio by itself provides any information that makes it possible to compare, and judge, pay practices between different firms. As explained before, firms with different characteristics could have differently justified pay ratios. It is more likely that the pay ratio needs accompanying information on its determinants such as the percentage of schooled employees, or the size of the company, to make it a useful tool for the comparison of pay practices.

The remainder of this paper is organized as follows: In chapter 2, the background information necessary for the analysis is provided. After that, the theoretical framework will be set out, explaining applicable theories, previous literature and the hypotheses. This is followed by the methodology and descriptive statistics in chapter 4. Afterwards, the results are presented in chapter 5, ending with a conclusion and discussion in chapter 6.

2. Background

To indicate the relevance of research on the pay ratio specifically for the Netherlands, it is important to point out the country specifics of the Netherlands regarding pay practices and its cultural attitude towards them. Rewards play a key role in motivating

(7)

employees and fostering the right behavior into achieving the best possible firm performance (Greckhamer, 2011). Cultural attitudes can lead to different effects of pay differentials on motivation, and consequently performance. These can differ per country, as employees in one country may be greatly motivated by a large pay dispersion, while in another country this effect is diminished due to negative cultural attitude (Chiang and Birtch, 2007; Greckhamer, 2011). This chapter therefore outlines the relative pay practices of executives and the average employee in the Netherlands, followed by a short description of the Dutch attitude towards these pay practices.

Conyon et al. (2011) have conducted a transcontinental analysis of differences in compensation between several countries within Europe and the U.S.1 A sample of their results is included in Table 1, and points to a low to moderate level of CEO compensation in the Netherlands compared to the other countries in the study. Average CEO compensation ranges from EUR 1.27 million in Sweden to EUR 3.78 million in the United States, with the Netherlands at a moderate EUR 1.53 million. Another notable fact from their study is that even though the base salaries for CEOs are similar between Netherlands and the U.S., the variable compensation part causes the remuneration to be more than twice as high in the U.S. as compared to the Netherlands (Conyon et al., 2011). It is thus not only the level, but also the composition of the remuneration package that differs significantly.

Regarding employee pay, in 2010 the Dutch workforce on average earned EUR 38.58 thousand, compared to EUR 42.35 thousand for the U.S. workforce (OECD, 2015). Table 1 presents the complete overview with the same countries included as in the research by Conyon et al. (2011). With only Ireland, Switzerland and the U.S. earning more, the Netherlands is thus on the higher end of compensation regarding employee pay.

The pay ratio focuses on the relative difference between the compensation of the employee versus the executive. However, there is no reliable comprehensive data on this measure for international comparison. Nonetheless, by dividing the average CEO compensation by the average workforce compensation as obtained before, an indicatory pay ratio per country can be created. The large differences in the pay ratio as presented

1_{Year of measurement is 2010. Countries included in the research by Conyon et al. (2011) are:} Belgium, France, Germany, Ireland, Italy, The Netherlands, Sweden, Switzerland, U.K. and the U.S.

(8)

in Table 1 reiterate the international differences in pay practices. With a spread in pay ratio between 37.31 (Belgium) and 99.58 (Italy), the Netherlands has a relatively low pay ratio of 39.55.

Table 1 – Pay practices per country

Country Average CEO _compensation Average workforce _compensation Average pay _ratio

Belgium 1,328.00 35.60 37.31 France 1,522.00 30.23 50.34 Germany 2,606.00 31.44 82.90 Ireland 2,585.00 40.28 64.18 Italy 2,717.00 27.28 99.58 Netherlands 1,526.00 38.58 39.55 Sweden 1,273.00 29.29 43.46 Switzerland 3,636.00 40.97 88.74 United Kingdom 2,016.00 32.32 62.38 United States 3,784.00 42.35 89.35

Compensation variables are in thousands of EUR. The year of measurement is 2010. The average CEO compensation is obtained from Conyon et al. (2011), average wage has been obtained from OECD (2015) and is converted to EUR using an 2010 average exchange rate of 0.75488 (www.oanda.com)

Cultural attitudes towards these relative differences in income are a great factor of importance when examining pay practices and their implication on employee behavior (Bebchuk et al., 2011). Using Hofstede’s (1983) model of culture in organizations, an analysis of four major dimensions can be made: masculinity–femininity (M-F), individualism–collectivism (I-C), uncertainty avoidance (UA) and power distance (PD). These dimensions can be used to examine cultural attitudes towards reward systems, as Chiang and Birtch (2007) have done in their research. In Appendix A, a short description of Hofstede’s model with respect to rewards systems can be found.

Employees in different cultures can thus be motivated by different rewards systems. Hofstede (2001) provides an international survey regarding these cultural attitudes. When placing the Dutch population in an international context, it scores relatively high on collectivism and femininity, and around the median regarding power distance (Hofstede, 2001). According to Hofstede (2001) the Dutch are thus moderately accepting of large differences in power, and value interpersonal harmony, rather than material success. In a compensation framework this is reflected in a lower acceptability of extremely high pay dispersions, however, some pay dispersion is appreciated. The U.S. has a more masculine culture, with a higher acceptance of power distance

(9)

compared to the Netherlands (Hofstede, 2001). This translates into a higher acceptance of large pay dispersions.

Using a worldwide survey amongst 40 countries, Gavett (2014) maps out ideal compared to actual pay ratios across the world. Even though the Netherlands is not included in this comparison, it is notable that there are large differences between countries regarding pay ratios, ranging from 28 in Poland, to 354 in the U.S. (Gavett, 2014). A sub survey asks participants from 16 countries what their preferred pay ratio would be. As expected, the United States has the highest preferred pay ratio, in line with the masculine culture as presented by Hofstede (2001).

3. Theoretical framework and hypotheses

This chapter will provide insight in the theoretical background that is needed to support the quantitative analysis. To get a clear understanding of the pay ratio, empirical evidence on the determinants of the pay ratio are discussed first, followed by the hypotheses regarding these determinants. Afterwards, applicable theories on the relationship between pay dispersion and performance are discussed. Empirical evidence from previous studies will then provide a more in-depth understanding of the theories and problems at hand, making it possible to formulate the hypotheses for the relationship between the pay ratio and performance at the end of this chapter.

3.1. Determinants of the pay ratio

It is essential to have a clear understanding of the determinants of the pay ratio. This helps to distinguish the effect of the pay ratio on performance from any other indirect influences that might affect performance through the pay ratio. Important to bear in mind is that the pay ratio consists of two components2_{, the executive pay and the} employee pay, each affecting the pay ratio in opposite directions. Along this line, Faleye et al. (2013) indicate that the size of the pay ratio depends on the bargaining power of the executive board versus the bargaining power of the employees. Bebchuk et al. (2011) and Conyon et al. (2001) support this notion. A higher bargaining power of

2_{The definition of the pay ratio is as follows: Pay!ratio = !}!"#$%&'(#!!"#$%"!&'()%

!"#$%&''!!!"#$!%&'()$!. Increasing

executive remuneration therefore by definition leads to an increase in the pay ratio, while increasing employee remuneration decreases the pay ratio.

(10)

the executive board increases their pay, thereby increasing the pay ratio. Meanwhile, increased bargaining power of employees raises the employees’ salary, thereby decreasing the gap between an average employee and the board. In the next paragraphs, the factors influencing the bargaining power of both parties are discussed in further detail.

3.1.1. Executive board bargaining power and dispersion

Amongst others, Shin et al. (2014) argue that executive compensation is the primary source of a high pay ratio. High executive compensation is the result of high bargaining power of the board vis-à-vis the supervisory board, the body determining the salary of the executives (Bebchuk et al., 2011; Conyon et al., 2001; Faleye et al. 2013). This is in agreement with the agency theory and the managerial power theory, which will be discussed in section 3.2.

Several determinants increasing the executive bargaining power can be deduced from basic economic theory, which state that skill is one of the key determinants of the level of pay (Bebchuck et al., 2011; Jirjahn and Kraft, 2007). Skill requirements increase with factors such as firm size, firm risk, growth opportunities and the complexity of operations, which is why these factors are in many studies associated with higher executive pay (Faleye et al., 2013; Jirjahn and Kraft, 2007; Kale et al., 2009). Another explanation of why size of the company is said to have a positive effect on executive pay, is found in the tournament theory. The tournament theory is explained further in section 3.2.1., and states that tournament incentives need to be larger when there are more competitors (Shin et al., 2014). Besides, Shin et al. (2014), amongst others, state that good previous firm performance may enhance the bargaining power, since performance may also partially be attributed to the well functioning of the executive board (Bebchuck et al., 2011; Conyon et al., 2001).

On the other hand, the effectiveness of the supervisory board may negatively influence the level of executive pay. A significant body of research finds that stronger supervisory board control has been proven to decrease executive compensation (Shin et al., 2014; Lee et al., 2008). When a supervisory board is able to work more effectively, they are better able to monitor the effort and output levels of the executive board members, making for a more precise method of remuneration. This way, the executive board is not able to misuse its power to bargain for a higher wage (Lee et al., 2008).

(11)

The effectiveness of the supervisory board depends on factors such as board size, amount of meetings and CEO-chair duality, all of which are positively associated with supervisory board effectiveness, and thus expected to negatively influence the executive compensation (Shin et al., 2014).

3.1.2. Employee bargaining power and dispersion

Besides executive board bargaining power, the pay ratio varies with the employee bargaining power versus that of their direct manager (Shin et al. 2014). Compared to the extensive literature on executive compensation, there has been a smaller body of literature on the determinants of employee compensation. However, academics agree that as the employee bargaining power relative to their manager increases, the employee salary increases, ceteris paribus leading to a lower pay ratio (Connelly et al., 2013). On the other hand, higher relative bargaining power of the firm’s managers would lead to lower employee salary, thereby increasing the pay ratio (Shin et al., 2014).

Employee bargaining power increases when they become harder to replace (Faleye et al., 2013). Since highly skilled employees are more difficult to replace than lower skilled employees, this is a key contributor to the bargaining power of the employee (Faleye et al., 2013; Shin et al., 2014). Besides, if their outside opportunities are higher, employees can more credibly threaten to leave the company when their salary is low (Connelly et al., 2013). According to Connelly et al. (2013), outside opportunities increase when there are a high number of firms within the same industry, making it easier for an employee to transfer into a similar job. Moreover, when a large part of the workforce has actively joined a labor union, this would lead to a better bargaining position (Faleye et al., 2013). However, it has been argued that labor unions in the U.S., where most studies on this topic have been conducted, play a more important role in terms of wage bargaining compared to many other countries in the world, including the Netherlands. This makes participation in labor unions a less relevant topic in this research (Schnabel and Wagner, 2003).

Oppositely, the managers increase their bargaining power over the employees that report directly to them when employees are more easily replaceable, which is the case for lower skilled employees, and in a more homogenous industry (Faleye et al., 2013). Thus, in case of a homogenous industry, both the outside option of the employee, as

(12)

well as the outside option of the management increases, which makes it difficult to speculate on the exact direction of the exact effect of this determinant (Faleye et al., 2013).

Figure 1 – Determinants of the pay ratio

Note: As some of the determinants can be correlated with each other, the direct effect of one of the determinants cannot always be distinguished clearly from another. However, the figure is merely a tool for summarizing the determinants and their expected effect as found from theory and previous research.

3.1.3. Hypotheses

In summary, the pay ratio is influenced by two components; executive board bargaining power and employee bargaining power. As concluded from the literature review, increased executive board bargaining power is expected to have a positive relation with executive pay, and therefore be positively related to the pay ratio. Employee bargaining power increases employee pay, thereby decreasing the pay ratio. Determinants entering both employee and executive bargaining power have an ambiguous effect on the pay ratio.

Even though some of these variables are correlated with each other, the individual effects have been studied in previous literature. Figure 1 maps out the determinants of

Supervisory Board Board size Number of board meetings CEO-Chair duality Executive pay Executive bargaining power

Employee pay

Pay ratio

Employee bargaining power

Executive Firm size Firm risk Firm complexity Growth opportunities Manager Outside opportunities Employee Employee skill Outside opportunities Unionization

(13)

the pay ratio, as well as the direction of their expected impact. Taken together, this leads to the following hypotheses regarding the determinants of the pay ratio:

H1: Determinants associated with executive- and manager bargaining power positively

influence the pay ratio.

H2: Determinants associated with supervisory board- and employee bargaining power

negatively influence the pay ratio.

3.2. Theories on pay disparity and performance

The history of motivational pay literature is rich, but mainly builds upon the agency theory (Jensen and Meckling, 1976). This theory is based on the separation of ownership and control, and describes the relationship between principal and agent. An applicable example of a principal-agent relationship is the relation between shareholders and management, or manager and employee (Fama, 1980; Jensen and Meckling, 1976). Incentives are set by the principal to motivate the agent, as the interests between the two are not perfectly aligned. The agency theory hereby describes the basic idea of incentivizing employees in working towards optimal firm performance (Fama, 1980; Frydman and Saks, 2010). While the agency theory is mainly built on an individualistic notion that the employee is motivated by the absolute amount of incentives he is given, Akerlof and Yellen (1990) state that relative pay might be just as important, if not more important in employees’ motivation. Their claim is based on the fact that humans compare their own rewards to those of others, and perceive themselves as being worse off when they receive less than others for the same task (Akerlof and Yellen, 1990). As this thesis investigates the pay disparity between the average employee and executives, theories on relative pay are of most importance, and will therefore be the main focus.

In current motivational pay research on relative pay, two main conflicting theories exist: the equity theory and the tournament theory (Frydman and Saks, 2010). The tournament theory, described in section 3.2.1., predicts that employees are motivated by larger differences in pay, since they are incentivized to exert more effort to work their way up in the organization (Lazear and Rosen, 1981). On the other hand, the equity theory, described in section 3.2.2., predicts that a more equally distributed

(14)

pay fosters cooperation, and feelings of fairness, thereby increasing productivity (Adams, 1965). These two main theories are discussed in more detail, after which a short summary of other relevant theories on motivational pay is given.

3.2.1. Tournament theory

The tournament theory (Lazear and Rosen, 1981) is based on the notion that employees compete with each other for promotion within the company. A promotion entails higher compensation, which can be seen as the prize of winning the tournament. Lazear and Rosen (1981) claim that the higher the monetary prize, the larger the incentive is to make promotion, and thus to work hard. In this sense, it provides similar incentives as individualistic bonus systems that provide monetary rewards when meeting targets (Connelly et al., 2014).

According to the tournament theory, differences in wage spreads within the company are justified because of two tournament characteristics: differences in the number of tournament rounds, and the number of contestants in each round (Lazear and Rosen, 1981). Keeping all else the same, the expected value of entering a tournament becomes higher when there are more potential rounds to be won. Lower in the organization, there are many potential promotions before reaching the top, thus entering the first tournament round has a high expected value. However, at the top of the organization, the amount of tournaments to be won is becoming more limited (Lazear and Rosen, 1981). In order to obtain the same incentives, it is therefore necessary to set higher monetary incentives for each promotion higher in the corporate hierarchy where there are less rounds left (Eriksson, 1999). According to Eriksson (1999), this explains why the salary structure within a company is often convex, with low increments in salary at the bottom of the organizational structure, while the salary gaps at the top are larger.

The second reason why salary gaps may differ is the amount of contestants participating in a tournament round. Increasing the number of contestants decreases the expected value of the tournament, since the chance of winning is smaller (Lazear and Rosen, 1981; Connelly et al., 2013). To compensate for this lower expected value, a higher monetary prize is necessary to reach the same incentive level as in a tournament with a lower amount of contestants (Eriksson, 1999).

(15)

According to the tournament theory, productivity levels rise when wage spreads increase (Lazear and Rosen, 1981). This increased productivity can be attributed to several reasons. First of all, the prize of winning a tournament provides for an individual incentive to work harder compared to a case without such incentives (Lazear and Rosen, 1981). When the wage spread increases, the incentive level thus increases, and more effort is expected to be put forward (Connelly et al., 2014). Larger salary increases accompanied with a job promotion therefore enhance the incentive to reach said promotion. Assuming that effort leads to productive output, higher individual effort levels thus lead to higher overall productivity (Eriksson, 1999).

Second, according to Main et al. (1993), tournaments promote retention of the most productive employees. Only the best performing employees in each tournament round are selected for promotion, making only the most productive employees continue to the highest levels within the organization (Main et al., 1993). The less productive employees will remain in the same position lower in the organization, and may eventually even be fired, as they are relatively the worst performers (Main et al., 1993). This leaves the most productive employees in the company, which causes higher productivity levels than in a case where there are no tournament type incentives (Connelly et al., 2014). Besides, employees that are considered more productive are attracted towards the company with tournament incentives, as their opportunities for rising to the top and thereby reaching a considerable salary increase are increased compared to a non-tournament setting (Connelly et al., 2014).

Third, the firm itself may operate more productively under tournament theory, as monitoring costs are reduced (Lazear and Rosen, 1981). When there is imperfect information about the output or effort levels of an employee, under tournament incentives, it is only necessary to rank each of the tournament round participants in order of their performance (Connelly et al., 2014). An example by Connelly et al. (2014) states that in case there is an external economic shock that affects all employees, there is no need to individually adjust all performance targets, as performance is measured in an ordinal ranking.

However, critics claim that tournament theory incentives lead to lower firm commitment, and feelings of unfairness between different levels within the organization (Main et al., 1993). Feelings of unfairness are the result of the employee perception of top executives being overcompensated compared to the lower levels

(16)

within the organization. This feeling of unfairness could lead to decreased motivation and therefore lower effort provision (Frydman and Saks, 2010). Assuming that effort is productive, lower effort leads to lower productivity. Also, larger wage gaps can create more tension between employees, causing demoralization and reduced cooperation (Connelly et al., 2014). As people are competing for promotion with their peers at the same organizational level, increased promotion incentives can harm collaboration incentives (Eriksson, 1999). If teamwork is important, then this selfish behavior may result in lower productivity due to the absence of teamwork (Hibbs and Locking, 2000). The equity theory embraces the criticisms of the tournament theory to demonstrate the opposite effect of pay differentials.

3.2.2. Equity theory

The equity theory was first mentioned by Adams (1965), but did not receive much attention at the time (Frydman and Saks, 2010). With the gain in popularity of behavioral economics, the theory has found more common ground, and is now one of the main theories on relative pay (Frydman and Saks, 2010). According to the equity theory (Adams, 1965), compressing the wage structure will lead to higher productivity. Two main reasons are put forward for the increased productivity levels. First, when there is a more equal distribution of wages, employees feel treated fairly (Akerlof and Yellen, 1990). All else the same, employees are willing to put in more effort per unit of monetary incentive if they perceive the wage distribution to be fair (Connelly et al., 2014). According to Connelly et al. (2014), employees will increase their individual effort levels, because they feel like they are rewarded for their hard work in a fair way. Akerlof and Yellen (1990) state that it is important to note that the employees perceive the wage distribution to be fair, rather than there being a strict definition of what constitutes fairness. They state that in some companies a higher wage spread could still be perceived as fair, while at others a lower wage spread is accepted. This also links back to the cultural differences, which indicated different wage spreads might be accepted for different cultural attitudes. Moreover, Hibbs and Locking (2000) state that due to feelings of fairness, the company morale is boosted, causing employees to act in the best interest of the company. The common goal of putting the company’s best interest first, aligns each employee and will increase productive output compared to a situation with unequal wages (Hibbs and Locking, 2000).

(17)

Second, because of the compressed wage, the incentive to sabotage co-workers’ efforts is decreased compared to the tournament theory (Connelly et al., 2014). The tournament theory states that promotion is made by outperforming a group. However, the equity theory relies more on absolute performance measures. Since co-workers focus on their own absolute successes rather than being the best within the group, collaboration incentives are increased (Drago and Garvey, 1998). Besides, as the increase in wage when a promotion is made is smaller, the incentives for making promotion are lower. The lower incentives for promotion further enhance teamwork, since it facilitates a more cohesive environment (Drago and Garvey, 1998). A more collaborative environment thus increases productivity through two different channels. Less unproductive efforts are being put into sabotaging coworkers, and more into teamwork, which, assuming teamwork is more effective than the sum of all individual output, leads to higher productivity levels (Connelly et al., 2014).

However, the notion of the equity theory that egalitarian pay increases performance also has its critics. Criticism of the theory includes that egalitarian pay gives incentive for shirking, and thus counterproductive behavior. The agency theory by Jensen and Meckling (1965) states that the right (monetary) incentives should be given to induce desired behavior and high effort levels. When promotion into a higher pay scale leads to a slight increase in salary only, there might not be high enough incentives for the employee to exert a productivity-maximizing effort level (Hibbs and Locking, 2000). Besides, according to Hibbs and Locking (2000), setting a compressed wage scheme could lead to the more productive employees being stolen away by firms with a more competitive pay structure where these productive employees can earn more. When the productive employees leave for companies with more competitive wage structure, this leaves the company with low wage spreads with least productive employees (Connelly et al., 2014).

3.2.3. Additional theories

As concluded from the paragraphs above, the tournament theory and the equity theory have contradicting predictions regarding the relationship between the wage spread and productivity. Even though other theories and motivations try to provide further explanation for the behavior regarding pay differentials within an organizational

(18)

context, frequently their predictions are based on either the tournament- or equity theory (Connelly et al., 2013; Faleye et al., 2013).

More support for the larger predicted wage spread by the tournament theory is found in the marginal productivity theory (Fama, 1980). Fama (1980) states that larger wage spreads occur within a company since employees will be remunerated equally to their contribution to the value of the company. Because the most productive employees make promotions, the most productive employees are found at the top layers of the company, earning significantly more than employees lower in the organization (Fama, 1980). Differences in pay are therefore the result of differences in productivity levels between employees.

The relative deprivation theory (Walker and Pettigrew, 1984) can be associated with the predicted wage levels of the equity theory. The relative deprivation theory, however, takes a more sociological approach, rather than emphasizing the organizational context of the equity theory. According to the relative deprivation theory, individuals feel mistreated when they believe that they have received less than their counterparts. In line with the expectations of the equity theory, the employees will be dissatisfied with unequal pay, and exert less effort at their job (Walker and Pettigrew, 1984).

More recent theories have attempted to combine the predictions of the equity theory and the tournament theory. These theories are the distributive justice theory and the procedural justice theory (Cowherd and Levine, 1992; De Cremer and Tyler, 2007). Both theories predict that wage spreads will only provide positive reinforcement and lead to higher effort and output levels when they are justified.

In line with the equity theory, the distributive justice theory (Cowherd and Levine, 1992) states that wage differences will only lead to more productivity when employees perceive wage differences to be fair. This is the case when there are, for example, obvious observable differences in productivity or ability between the different employees (Cowherd and Levine, 1992). In line with the tournament theory, the distributive justice theory states that the justified pay disparity between job grades is sometimes large, and can thus still create incentives for high effort with a high pay disparity (Cowherd and Levine, 1992).

Where the distributive justice theory looks at whether the final distribution is fair, the procedural justice theory (De Cremer and Tyler, 2007) states that only when the

(19)

process that is used to come to the specific wage is perceived to be fair, employees feel that they are treated fairly. Fair treatment in turn leads to higher productivity levels, as also discussed in the equity theory section above.

3.3. Previous empirical evidence on pay disparity and performance

Empirical evidence on the pay ratio between executives and the average employee is limited and are mixed. To my knowledge, there are only three studies to date that have empirically attempted to investigate the performance effect of the pay ratio between executives and the average employee (Connelly et al., 2013; Faleye et al., 2013; Shin et al., 2014), all three obtaining different results. As stated before, reasons for the scarcity of prior research on this topic can be attributed to the lack of disclosure on the pay of employees3_{. Besides, due to growing attention of the popular press and legislative} offices in both the U.S. and Europe in recent years, the academic interest for this topic grew only recently (Connelly et al., 2013).

Besides the articles by Connelly et al. (2013), Faleye et al. (2013) and Shin et al. (2014), other prior empirical research on the tournament and equity theory has focused on situations where quantitative data is more readily available, such as pay differentials within the executive board (e.g. Bebchuck et al., 2011; Conyon et al., 2001) or situations outside the corporate environment, such as sports teams (Connelly et al., 2014). For relevance of this research, the three articles on pay differentials between the average employee and the executive team are discussed in detail. Besides, articles focusing on the pay disparity-performance relationship in an organizational setting, such as pay differentials within the executive suite are also included. These articles focus on executives only, but are relevant to this research since they are placed in an organizational context.

Considering the large body of literature considering the pay differential between executives, to keep matters concise and relevant, only academic literature from the year 2000 and onwards has been taken into consideration for the empirical review. These studies are most likely to reflect the current situation in a changing organizational context, and therefore are more relevant for this thesis. The remaining body of literature

3_{Approximately 10% of all US listed companies report information on total labor expenses} (Connelly et al., 2013).

(20)

is filtered based on journal quality, leading to a final selection of 8 papers that will give an in-depth understanding of the empirical research to date. A comprehensive overview of the results of these relevant studies can be found in Appendix B – Review of relevant literature.

3.3.1. Tournament theory support

Faleye et al. (2013) study the executive to employee pay ratio in the U.S. They indicate that the pay dispersion between the CEO and the average employee is positively associated with company performance, consistent with the tournament theory. However, their sample solely contains firms that include the total employee pay in their annual report, resulting in a sample of 450 out of 1500 S&P1500 firms. This could possibly create selection bias, as the selected sample also reportedly has higher revenues and includes a higher fraction of financial firms (Faleye et al., 2013).

More convincing support for the tournament theory is found in the literature based on variations within the executive suite. Kale et al. (2009) find evidence in line with the predictions from the tournament theory, where higher pay disparity increases firm performance. Lee et al. (2008) find similar results regarding compensation dispersion of the top management team. They argue that greater pay dispersion leads to higher firm performance since it can mitigate some of the agency problems related to difficulties in monitoring from shareholders. Greater pay disparity aligns the executives to the long-term objectives, in line with the shareholders’ best interest, since long term outperformance could lead to a desired promotion of the executive at hand (Lee et al., 2008).

Kale et al. (2009), suggest that the tournament incentives are stronger, and thus more effective in case the CEO has been in place longer, or when the CEO is close to retirement. Then, a new CEO is likely to be appointed soon and the expected value of the tournament prize increases (Kale et al., 2009). This, however, is not likely to have a significant impact on the performance effect based on the pay ratio between the executive suite and average employee. The reason is that the employees are more than one tournament layer under the executive (Faleye et al., 2013).

(21)

3.3.2. Equity theory support

Shin et al. (2014) have investigated the pay disparity between the CEO and the average work force, and have found empirical evidence in favor of the equity theory. Their results indicate a negative relationship between pay disparity and subsequent firm performance, measured by Return on assets (‘ROA’). Furthermore, they were able to create a model predicting the executive pay multiple per company. This predicted ratio is then used to test the performance effect of deviations from the predicted pay multiple. The results suggest that larger deviations from the predicted pay multiple are associated with poor future performance, giving support to the reasoning that productivity is based on perceived fairness rather than absolute differences (Connelly et al., 2014; Shin et al., 2014).

Bebchuk et al. (2011) measure dispersion as the percentage of the total top 5 executives’ compensation to what the CEO receives. Their conclusion is that a greater dispersion between the CEO and other executives leads to lower firm performance, as measured by Tobin’s q. Bebchuk et al. (2011) argue that their results are in line with theories concerning agency problems, describing opportunistic behavior attributable to high pay dispersion. Bebchuk et al. (2011) state that the opportunistic behavior presented in this case is selfishness, induced by the high monetary incentives associated with a promotion. Even though they do not link this opportunistic behavior directly to the tournament theory or equity theory, many scholars associate this type of behavior with the negative performance effects of the tournament theory (Eriksson, 1999; Hibbs and Locking, 2000).

3.3.3. Further findings

Numerous studies do not find any conclusive evidence for either theory. The research towards tournament and equity theory within the context between the CEO and the average employee is relatively new. The cause of the mixed and inconclusive results can be the consequence of this relatively new field of research, where methods are still being refined further with every new study.

In their sample of 100 U.K. firms, Conyon et al. (2001) have tried to prove the tournament theory within the executive suite. Yet, they find no significant relationship between executive pay dispersion and firm performance. A limitation of their study,

(22)

however, is that they do not include control variables for the different corporate governance structures of firms, which in other subsequent studies have proven to be an important factor in determining the pay ratio (Lee et al., 2008; Faleye et al., 2013). This gives reason to suspect that their results would be different if more of these control variables were to be used. Similarly, Henderson and Fredrickson (2001) are unable to come to a conclusion regarding the performance effect of the wage spread, which they claim might be the result of their outdated data. Their sample runs until 1990, when executive salaries were still relatively equal. After this date, a tremendous change has taken place in the field of executive compensation, which is now more dispersed, according to Henderson and Fredrickson (2001).

Connelly et al. (2013) try to give an explanation for mixed and/or insignificant results of previous research, arguing that coexistence of the tournament theory and equity theory is possible when temporal effects are included. This means a separation is made between the short- and long-term performance effects. In their empirical research, Connelly et al. (2013) use the CEO to median employee pay ratio of the full S&P1500 sample to demonstrate that short-term performance, as measured by annual ROA could benefit from high pay dispersion, while long run performance measured by a 5 year average ROA could be harmed by it. Connelly et al. (2013) might be able to explain contradictory or inconclusive results between studies that use different performance measures and therefore may measure performance on a different time horizon. However, they fail to provide an explanation for different results across studies that each use similar performance measures and time spans.

3.4. Pay disparity and performance hypotheses

As demonstrated in the literature review on the performance effect, empirical evidence is mixed. Some evidence supports the tournament theory, based on a positive relationship between the pay ratio and corporate performance (Faleye et al., 2013; Kale et al., 2009; Lee et al., 2008). On the other hand, there is also support for the equity theory, resulting from a negative relationship between the pay ratio and company performance (Shin et al., 2014; Bebchuk et al., 2011). Furthermore, there are several studies that do not find a conclusive relationship between the pay ratio and company performance at all, or their results show mixed signs (Connelly et al., 2013; Conyon et al., 2001; Henderson and Fredrickson, 2001). These mixed results further indicate the

(23)

need for additional studies on the performance effects of pay disparities, a topic this thesis aims to shed light on.

The equity theory posits a negative relationship between the executive-employee pay ratio and firm performance (Adams, 1965). Some empirical evidence is able to support this theory, leading to the following hypothesis:

H3: The executive-employee pay ratio is negatively associated with firm performance

On the other hand, the tournament theory (Lazear and Rosen, 1981) speculates a positive relation between the executive-employee pay ratio and firm performance. Several studies have been able to prove these implications of the tournament theory. Thus, the fourth hypothesis is as follows:

H4: The executive-employee pay ratio is positively associated with firm performance

The contradictory predictions from theory, together with contradictory results of previous research lead to suspect the linear relationship between the pay ratio and firm performance predicted by the tournament theory and equity theory is not the best fit. The relationship might have an optimum/minimum, taking into account different aspects of the underlying theories. As the equity theory suggests, rather than the absolute level of compensation difference, it could be the perceived fairness of this gap that affects employee productivity (Connelly et al., 2014, Shin et al., 2014). Both the distributive justice theory and procedural justice theory support this notion, by indicating that it is the perceived fairness that leads to increased motivation (De Cremer and Tyler, 2007). Assuming the pay ratio is considered fair when it coincides with the pay ratio that is set for firms with similar traits, it could be the deviation from this predicted pay ratio that leads to inferior performance. Which forms the fifth hypothesis:

H5: A pay ratio that deviates from the predicted harms firm performance

From the fifth hypothesis, a high pay ratio could thus be perceived fair if it coincides with determinants leading to a high pay ratio (a high expected pay ratio). On the contrary, a low pay ratio is fair when it coincides with determinants of a low pay ratio

(24)

(a low expected pay ratio). In both of these cases, employees are able to justify the difference in salary levels accordingly.

However, there might be a difference between a positive and a negative deviation from the predicted pay ratio. If a high pay ratio coincides with determinants for which a low pay ratio is expected, it might harm employee morale, and thereby negatively affect productivity (Shin et al., 2014). On the other hand, a low pay ratio coinciding with the determinants for a high pay ratio could lead to higher productivity levels due to a higher feeling of perceived empathy from the top. These arguments lead to the final hypothesis:

H6: Having a higher pay ratio than predicted has a different effect on performance

compared to having a lower pay ratio than expected

4. Data and methodology

4.1. Sample and variable construction

The initial sample is constructed from AEX and AMX listed companies, the 50 largest Dutch listed companies based on market cap, over the years 2010-2014. This results in an initial cross-sectional panel data set of 250 firm-years4. From this dataset, two companies have been excluded due to the fact that their consolidated financial statements only include top management personnel due to their company structure. It is therefore not possible to generate a pay ratio for these companies. Furthermore, one employment agency (Randstad) has been left out of the sample. This is since the majority of their reported numbers of employees are employed at other companies, rather than at the employment agency itself. The employees are thus not likely to respond to wage differences within the employment agency. The final sample consequently consists of 58 firms with a total of 227 firm-year observations5. The panel data set is unbalanced, since there have been changes in the AEX/AMX composition during the sample period. There were 32 firms that were continuously listed from 2010-2014; the other 26 firms are part of the dataset only in specific years.

4_{Since some regressions include lagged independent variables, all variables except for the pay} ratio, executive pay and employee pay have been collected for the year t-1. This means that these lagged variables include the years 2009-2014.

(25)

An unbalanced panel data set can introduce problems with regression analyses. The main issue is selection bias, where the attrition of companies from the sample is not completely random. It could be the underperforming companies that leave the AEX or AMX. A subsample has therefore been created for robustness purposes. The subsample forms a balanced panel data set of 32 firms with 170 observations in total; it includes the firms that were part of the AEX or AMX in all sample years. This way, subsequent years are more comparable, as they contain the same companies.

However, there might be selection bias regarding the subsample as well. It could be that only the mature and well performing companies remain in the AEX or AMX during the full sample period, while the others that perform worse drop out. This means losing some valuable data of the younger and/or underperforming firms. It was therefore decided to use both samples, the full sample is used for all analyses and the subsample is used as a robustness check. In case the monetary values are denominated in any currency different than the euro, the values are converted into euros using the yearly average ask exchange rate taken from www.oanda.com.

Pay ratio

The main variable of interest in this research is the executive pay ratio. The pay ratio consists of two components: executive pay and average employee pay. Since this study aims to provide insight in the effect of pay disparity on the firm performance through the productivity of the employees, it is of essence to use an executive compensation measure that is most visible to the employee. Actual executive payout levels include complex structures involving the valuation of shares and options, often too complex for a typical employee to come to a conclusion on the total value of rewards. The choice is therefore to include policy levels of remuneration, a more visible measure of executive remuneration. The policy level remuneration includes information on the base salary, as well as the value of the annual- and long-term bonus, which together make up the Total Direct Compensation (‘TDC’). Even though previous research uses the accounting costs of the executive remuneration (Connelly et al., 2013; Faleye et al., 2014), Connelly et al. (2013) have stated to use these due to lack of information on the policy values, as policy levels are usually not included in online databases. Information on executive remuneration in this research has been collected by hand from annual reports,

(26)

together with data from the executive compensation database constructed by Focus Orange6.

In line with the method of Faleye et al. (2013), average employee pay is found by dividing the labor force cost excluding the costs of executive compensation, by the reported average number of full time employees during the year. Faleye et al. (2013) use the total labor force costs, which includes social contributions paid by the employer as well. However, in this research, the social contributions are not included, as this is not a component the employee receives, and can be highly affected by differences in accounting standards across firms rather than actual salary payments. The information needed for the calculation of average pay has been hand collected from annual reports.

To operationalize the pay ratio, the policy level of the total direct compensation of the executive is divided by the average employee pay.

Pay!Ratio = ! _{!"#$%!!"#!!"#"$%!!"#$#!!"#$%&'(#!!"#$%"!&'()%!!"#$# !/!!"#$%&!!"!!"#$%&!!'}!"#$%&!!"#"!!!"!#$%&'!!!"#$%"!&'()%

Performance

Previous research has used many different indicators for firm performance. As some scholars have attempted to prove the performance effect of pay dispersion by using stock returns, others have shown that this measure can be volatile and highly dependent on external market shocks rather than productivity from within the company (Connelly et al., 2013). Besides stock returns, ROA is therefore often used as a measure of company performance (Shin et al., 2014). ROA is a good measure for company performance, but according to Faleye et al. (2013) this measure is less affected by employee productivity than, for example, revenue per employee. The choice is therefore to include revenue per employee as the main measure for company productivity that is most affected by differences in employee output, which is thought to be most affected by differences in the pay ratio.

6_{The Focus Orange database consists of executive compensation data of AEX and AMX listed} firms. For AEX listed companies it comprises the years 2011-2014, for AMX listed companies 2012 and 2013 are included.

(27)

Explanatory variables

All variables used in this research are presented in Appendix D, which includes a description of each variable, and the hypothesized sign on firm performance. The data gathering process of each variable, as well as an explanation of some of the less straightforward variables from previous literature is discussed in more detail below. Previous literature provides a long-list of variables that can be included as control variables. The reasoning behind inclusion of the variables in each of the models is discussed in more detail in section 4.3, which includes a description of the empirical models.

The data is compiled from different sources, including annual reports, Datastream, Central Bureau of Statistics (hereafter ‘CBS’) and LexisNexis. All remuneration data have been hand collected from annual reports, while the explanatory variables excluding news coverage and level of schooling have been obtained from Datastream. Data on the level of education has been obtained from the CBS of the Netherlands. This data is included on an industry level, as there is no information on the individual companies available. The news coverage on executive remuneration of companies used in the sample is obtained from LexisNexis.

The performance effect of the pay ratio is based on the notion that all employees in the organization know the executive pay. Since this is an unrealistic assumption, a proxy is used for the visibility of executive pay. When top salaries are covered in the news, they are assumed to be more visible, and might have a larger effect on employee motivation and productivity (Connelly et al., 2014). Therefore, news coverage on executive compensation is collected from LexisNexis. LexisNexis is a database that includes news articles from 1980 onward. This database has been searched for news articles regarding the executive pay of the companies covered in the sample. A Boolean search is used to search through Dutch news articles for all companies in their respective time period of listing at the AEX or AMX7. The found articles are scanned for relevancy, only articles that state the amount of remuneration of any of the board members is included, in line with the method used by Connelly et al. (2014).

7

The exact Boolean search that is used is: “HEADLINE(Company name) AND salaris OR beloning OR bonus”. This searches for the company name in the title of the article, as well as either salaris (salary), beloning (compensation) or bonus in the full article.

(28)

Consistent with previous literature, the Hirshman Herfindahl Index (‘HHI’) is used to measure industry concentration (Faleye et al., 2013; Shin et al., 2014). The HHI measure is not obtained from a database, but rather calculated from the available market data from the sample. Each company is divided into their respective SIC class, which is based on the first two digits of the standard industry classification code (‘SIC’). The sum of the squared revenues within the industry class then defines the Herfindahl Index. As noted by Faleye et al (2013), this measure does not give a realistic representation of the true industry concentration of the whole market, but is rather a measure of relative density of the industries included in the sample. In the sample, some industries might be overrepresented, leading to a lower HHI. Since the concentration measure is used to indicate the ability to switch jobs, and job switchers are more likely to remain in the same type of firm (a larger, listed firm, within the same industry), this bias does not pose a problem in this research (Faleye et al., 2013).

4.2. Descriptive statistics

The key variable of interest is the pay ratio. In figure 2 below, the development of the pay ratio and its components can be identified.

Figure 2 – Development of the pay ratio and its components from 2010 to 2014

Note: median values are used with 2010 as the base year. For comparability across the years, the data from the subsample has been taken into consideration. This sample includes the same 34 companies in every year of the sample period.

-5% 0% 5% 10% 15% 20% 2010 2011 2012 2013 2014 Cumulative growth Year Employee compensation CEO TDC

CEO employee pay ratio (TDC)

(29)

As seen in the figure, the pay ratio has an upward trend, with some fluctuations over the years. Overall, the pay ratio grew 13.1% from 2010 to 2014, from 31.31 to 35.53. This is mainly due to the increase in CEO pay, rather than employee pay. The total CEO compensation increased with 16.7%, while employee compensation increases with 3% cumulatively. There are some fluctuations in employee count for individual firms, leading to believe the decline in employee compensation in 2012 can be attributed to a changing workforce composition.

The overall statistics for the pay ratio and each of its components is outlined in Table 2 below. Employee compensation is equally distributed across the quartiles, with a median of EUR 49.54 thousand. The median base salary for the CEO is EUR 625 thousand, while the TDC is EUR 1,402.5 thousand. For both executive compensation variables, it is notable that the gap between the median and the third quartile is larger than the gap between the median and the first quartile. The data thus shows that the executive compensation is not equally distributed across the quartiles. Rather, the differences in compensation between firms become larger as compensation levels increase. Consequently, this leads to the pay ratio being skewed towards the higher side as well, with pay ratio differences between firms increasing when the ratio becomes larger. The median revenue per employee is EUR 306.5 thousand, with an approximately equal distribution across the quarters. The median return on assets that is achieved at firms within the sample is 5.15%.

Table 2 – Summary statistics for key variables

Variable N Mean SD Q1 Median Q3

Compensation Employee Compensation 227 53.35 22.54 40.95 49.65 59.69 CEO base 227 715.93 292.97 500.00 625.00 850.50 CEO TDC 227 2,062.53 1,733.16 1,019.50 1,402.50 2,463.44 Pay ratio's

CEO - Employee (Base) 227 15.21 8.25 9.40 13.65 18.80

CEO - Employee (TDC) 227 42.86 36.23 21.91 31.60 47.95

Performance

Revenue per employee 276 614.00 770.81 202.00 306.50 557.00

Return on assets 276 3.68 13.74 0.63 5.15 7.71

The table includes data for the full sample of the years 2010-2014, with each variable as defined in Appendix D. Variables in the compensation panel, as well as revenue per employee are in EUR thousands. Return on assets is presented as a percentage.

(30)

The remaining variables are presented in Table 3. The median company has 8.24 thousand employees, with a market capitalization of 1.99 billion and 2.41 billion in revenues. Similar to the executive compensation, the size distribution of the companies above the median shows to be much wider than the companies below the median.

One of the other variables worth noting is the industry concentration. A higher number represents a more concentrated market served by a few large firms. In the sample, the average Herfindahl index is 0.34. Markets with indices above 0.25 are often considered to be highly concentrated (Jansen et al., 2009). As stated before, this high concentration is most likely the result of the HHI being calculated with firms listed on the AEX or AMX. However, this does not create any problems for its purpose within this research.

The firms listed on the AEX and AMX are all firms with international sales. On average, about 70% of sales is outside the Netherlands. Around the third quartile, 93.48% of sales come from abroad, pointing to very international companies. Other explanatory variables include news and supervisory board bargaining power. In about half of the firm years the executive compensation has been in the news. The average supervisory board bargaining power is 2.5, meaning that the size of the supervisory board is on average 2.5 times larger than the executive board.

Table 3 – Summary statistics for other variables

Variable N Mean SD Q1 Median Q3

Company size Number of employees 227 26.89 51.06 2.45 8.24 25.29 Market capitalization 281 10.61 25.85 0.84 1.99 9.33 Revenue 276 15.14 46.67 0.89 2.41 8.15 Other

Debt to asset ratio 273 26.50 15.65 17.60 24.19 35.09

Industry concentration 282 0.34 0.22 0.18 0.22 0.44

Investment return 274 13.80 36.82 -8.52 12.82 36.56

Level of education 282 35.74 14.79 20.00 37.50 45.59

Market to book ratio 282 1.96 2.87 1.02 1.63 2.48

News 282 0.46 0.50 0.00 0.00 1.00

Percentage of foreign sales 275 69.55 25.94 54.39 76.58 93.48

Price volatility 273 28.16 9.01 21.22 26.79 33.89

SB bargaining power 282 2.51 1.71 1.50 2.00 2.67

The table includes data for the full sample of the years 2010-2014, with each variable as defined in Appendix D. The number of employees is in thousands of FTE, market capitalization and revenue in billions of EUR. Investment return, level of education and percentage of foreign sales and price volatility are in percentages.

(31)

The subsample includes only the firms that are part of the AEX and AMX during the full sample period. The descriptive statistics of the subsample are included in Appendix E. Compared to the full sample, the firms in the subsample have a higher executive salary (the average TDC is EUR 280 thousand higher in the subsample). Consequently, a higher pay ratio can be observed (48.85 compared to 42.86). This could be because the subsample consists of larger firms than the full sample, and larger firms usually have higher executive salaries. Larger firms are more likely to remain in the AEX or AMX during the full sample period because the firms at the small side of the AMX are more likely to be substituted with other firms from the small cap index (‘AScX’).

The executive to average employee pay ratio, cause and effect : empirical evidence from the Netherlands