CEO pay regulation for largest U.S. firms: relevance of industry and bank-level characteristics

CEO pay regulation for largest U.S. firms: relevance of

industry and firm-level characteristics

Name: Sjirk Andela

Student number: 11049049

University of Amsterdam – Bachelor Thesis Specialization – Finance, BSc ECB

Supervisor: Ekaterina Seregina EC’s: 12 EC

Abstract

In this bachelor thesis the effect on CEO compensation for known variables is checked. The relation between CEO compensation and firms’ revenues is researched. Also, effect of the regulation introduced in 2015 on the CEO compensation in the United States is researched. Furthermore, the effect of the regulation on the mean CEO compensation across industry sectors is researched. The sample used in my research consists of 1,403 firms listed on the S&P 500 for the years 2010-2018. For the variables return on assets, Tobin’s Q, revenue, and leverage the effect is confirmed. The results show a positive relation between firms’

revenues and CEO compensation and an existing difference in mean CEO compensation across industries. The introduction of the regulation in 2015 has no effect on CEO compensation. The CEO compensation after the introduction is significantly higher than before, which shows that the opposite is true.

Statement of Originality

This document is written by Sjirk Andela. I declare to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are 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.

§1 Introduction

In the last ten years the compensation a chief executive officer (CEO) receives have been criticized more than the years before. The main reason for this criticism is the wage gap between CEOs and workers. In 2011 the median executive officer in the S&P 500 earnings were $9.6 million (Murphy, 2013). This is substantially higher than in other countries and has increased by 600% since 1980 (Murphy, 2013). This rapid increase in CEO compensation may play a role in the recent rise in income inequality (Piketty & Saez, 2003; Piketty, 2014). The wage gap between high executive officers and workers continues to be wide. According to the Economics Policy Institute in 2018 CEO compensation for the top 350 United States (U.S.) firms rose since 1978 with 1,007.5%, compared with 11.9% for average workers (Mishel & Wolfe, 2019). In other words, CEOs make 278 times as much as an average

worker. During the financial crisis starting in 2007 it dropped from 345 times as much to 195 times. But immediately after the financial crisis the gap started growing again (Mishel & Wolfe, 2019). The development of the CEO/average worker compensation ratio can be seen in figure 1 in the appendix.

This of course raises questions about the payment to executive officers. One question that arises is whether the impact of CEOs is truly so valuable to firms that it justifies the amount of compensation. There are several perspectives regarding this question, with one positing a positive relationship with firm performance and value and the opposite positing a negative relationship with firm performance and value. A second question that pops up is if there are existing differences in CEO compensation across industry sectors. A third question that arises is whether regulation on CEO compensation is able to restrict the continue increase.

In 2014 in the EU banking sector a bonus cap regulation was introduced. Colonello et al (2019) analysed how this regulation affected the structure of the CEO compensation. But also, in the U.S. regulation on CEO compensation has been announced. For example, in 2015 when the SEC launched the CEO Pay Ratio rule. This rule required firms to share to the public the CEO compensation, the salary of the average worker, and the CEO/average worker pay ratio (Securities and Exchange Commission, 2015).

I intend to confirm that the effects of several known variables on CEO compensation are valid for the period 2010-2018. Then I investigate whether the introduced regulation in 2015 has a significant effect on CEO compensation. Also, look for differences in CEO

compensation across industry sectors and see if they became smaller after the additional regulations. In this research I intend to answer the question: What is the effect of CEO pay regulation for largest U.S. firms, the relevance of industry and firm-level characteristics?

For my research I used data on non-financial firms listed on the S&P 500 between 2010 and 2018. The data is found using COMPUSTAT and Execucomp via Wharton Research Data Services. There are several control variables that are known to have an effect on the amount of CEO compensation. The most important ones are included in my research, so I can check their relationship with CEO compensation. To analyse the relationship between variables and CEO compensation Stata is used. In Stata these variables are tested using multiple Fixed Effects (FE) regressions. In my research I stated three hypotheses. The hypotheses are as following:

Hypothesis 1: CEO compensation is positively related to firms’ revenues. Hypothesis 2: CEO compensation differs across industry sectors.

Hypothesis 3: The regulation in 2015 has a significant effect on CEO compensation and the impact is valid for the various industry sectors.

In the second section some prior literature will be discussed. The research methods and results of a couple papers are highlighted. The regulation in the EU banking sector and the regulation on the CEO Pay Ratio in the US are explained more briefly. The third section is used to describe and explain the research method. In the fourth section the descriptive statistics are given. In the fifth section the results are shown and discussed. In the last section the conclusions, limitations, and room for discussion are given.

A lot prior research is done in the past regarding CEO compensations. For the return on assets the effect on CEO compensation is proven many times in prior research (Murphy, 1999). The Tobin’s Q is known to have a negative relation with CEO compensation when used as proxy for firm value (Mehran, 1995). Also, for the revenue the positive relation is researched often (Conyon & Murphy, 2000). Leverage is also used often as control variable for CEO compensation (Coles et al., 2006). Zhou (2000) concluded that interindustry

differences are insignificant. However, Murphy (1999) argued that the industry sector in which the firm is active also has an influence on the CEO compensation. Results from prior research on differences across industries are divided.

The effect of the control variables on CEO compensation is verified. For firms’ revenues a positive relation with CEO compensation is found. What also becomes clear is

that CEO compensation differs across industry sectors. Furthermore, the regulation in 2015 has no effect on CEO compensation and the various industry sectors.

§2 Literature review

The size of the compensation depends on several determinants. The results of prior

research on the CEO’s compensation for firms show us that there are several determinants that have a relationship with compensation. A couple determinants will be discussed below. The results of prior research on differences in industry sectors are given. Also, regulation in the EU banking sector and US Pay ratio are reviewed.

§2.1 Risk

According to Berk and DeMarzo (2014) the cost of capital of levered equity increases with the firm’s market value debt-equity ratio. Financing by debt is riskier than financing by equity because by taking additional debt the firm has an obligation to pay in the future. High-leveraged firms carry more risk since they have a higher debt-equity ratio, but this risk comes with a higher potential return on investments. So, high-leveraged firms have

potentially higher returns than low-leveraged firms. The potential losses are bigger, but also the expected return on investments is higher. Cerasi et al. (2020) looked whether CEO compensation changed after the issuance post-crisis guidelines. The compensation packages for executives are difficult and are built up of fixed compensation and variable

compensation. For the variable compensation CEOs might want to take excessive risk to maximize their own compensation. In the case CEOs their appetite for additional risk grows it is important to understand the determinants and how this might affect the firm’s risk-taking policy and how to regulate this success (Cerasi et al., 2020).

In 2011 the “Principles and Standards for Sound Compensation” (P&S) was adopted by the Financial Stability Board (FSB), to analyse the variation in compensation. The new regulation was meant to relate compensation with prudent risk-taking. The main goal of their research is to study the sensitivity of compensation to measures of performance and risk after this new regulation was implemented (Cerasi et al., 2020). Their focus is on the variable part of the CEO compensation and the conclusion is that implementing P&S has significantly changed the CEO compensation policies. Guo et al. (2015) confirm that the composition of CEO compensation is related to a firm’s incentive to take excessive risk. Also, the paper of Ghysels et al. (2005) focussed at a possible trade-off between conditional variance and conditional mean of the stock market return. A positive and significant relation

between risk and return has been found. Therefore, Ghysels et al. (2005) concluded that there exists a positive and significant relation between risk and return.

§2.2 Firm size

Bebchuk and Grinstein (2005) argue that firm size is the most important determinant of CEO compensation. The most consistent and enduring result from myriad studies of CEO

compensation is that firm size is positively and significantly associated with compensation levels (Conyon, 1997). Edmans and Gabaix (2016) state that the talent of a manager is scalable and therefore the more talented managers tend to work at larger firms. According to them the effect of CEOs on firm value is very large compared to the average employee of that company. That is why Edmans and Gabaix (2016) also argue it is optimal to pay high CEO compensations to attract talented CEOs, which confirms the predictions for the cross-sectional relationship between the compensation for a CEO and the firm size. In their research firm size is used as a proxy for talent. Another finding is the variety in the level of pay for different sizes of firms. The level of pay should increase as the firm size increases (Edmans & Gabaix, 2016). A greater proportion of incentive pay is positively related to firms’ valuation and performance (Guo et al., 2015). However, Murphy (1999) showed us that the height of the compensation is inversely correlated for larger firms. According to him levels of pay are higher in larger firms, but the pay-performance sensitivities are lower.

§2.3 Regulation

§2.3.1 Regulation EU banking sector

Colonello et al (2019) analysed the introduction of the EU bonus cap for banks implemented in January 2014. Their research examined how compensation structure affects the ability of banks to retain their CEOs and the risk-adjusted performance. Colonello et al (2019) used a 100% threshold if the treatment group consisted only of banks affected by no bonus cap. They chose this threshold of 100% because then the number of false positives in the treatment group is minimized. A CEO is classified as treated if their variable compensation to fixed compensation ratio is above the threshold of 100%. They introduced a variable called ‘Treatment Intensity’ which equals 0 for the control group. For an untreated CEO, which equals a variable to fixed compensation ratio below 100%, the variable ‘Treatment Intensity’ also equals 0. For example, if the ratio is 140%, the variable equals 0.4. Colonello

et al (2019) analysed how banks changed their compensation packages around the new regulation and analysed if and how banks compensated their CEOs for the loss in variable compensation. Several dependent variables are used for the measurement of

compensation. Namely, the level of fixed and variable pay, the ratio between the variable and fixed compensation, and the expected pay. The age, tenure, a female indicator, professional experience, and a CEO indicator are used as control variables on executive-level. Bank-level control variables are size, performance measured by ROE, and the number of executives in the board (Colonello et al., 2019). A couple conclusions are drawn.

Regulation in pay practices in the banking sector may have unexpected consequences. It can adversely affect banks’ abilities to retain their most talented CEOs. Also, the consequences from these regulations are far from obvious. The results indicate that banks deal with the regulation by giving CEOs higher fixed compensation and lower the maximum variable compensation. This suggests that banks compensate their CEOs for the introduction of the pay cap. Besides the fact that their research is about the banking sector in the European Union and my research is about the non-financial banking sector for the US, their research method can be seen as an example and can be helpful for my analysis. It also may be the case that the implementation of regulation on CEO compensation has the same effect in the US as in the EU.

§2.3.2 Regulation US Pay ratio

In the US CEO Pay regulations have been appointed before in the past. For example, in Augustus 2015, the U.S. Securities and Exchange Commission (SEC) adopted the CEO Pay Ratio rule that requires public companies to share the CEO compensation, median employee total annual compensation, and the ratio between CEO compensation and the

compensation for the median employee (KPMG, 2017). This new rule required by the Dodd-Frank Wall Street Reform and Consumer Protection Act gives public companies flexibility in the calculation of the compensation for the median employee (Securities and Exchange Commission, 2015). For example, companies are permitted to choose its methodology for identifying its median employee. Also, it permits the median employee determination only once in the three years (Securities and Exchange Commission, 2015). Besides that,

companies are allowed to exclude “leased” workers. In this case leased workers are non-U.S. employees. There are several factors companies can take into account by determining the

median employee. The size and nature of the workforce of the company, the complexity of the organization, and the different types of compensation the workers receive (Richman & Hermsen, 2015). In the disclosure companies have to describe the chosen methodology, the material assumptions, adjustments, and estimates made for identification of the median employee (Wood & Posner, 2013). It also gives shareholders information to help evaluate the CEO’s compensation. These new U.S. disclosure rules were implemented with the

purpose of more transparency on CEO compensation and the pay ratio (KPMG, 2017). Figure 2 in the appendix shows the development of CEO compensation, the figure suggests a change after the introduction of the CEO Pay Ratio regulation in 2015. Therefore, it is likely that the total compensation for CEOs changed significantly after this regulation.

§2.4 Firm performance and firm value

The measurement of firm performance and firm value can be done by using several proxies. Hall and Liebman (1998) researched the relationship between the CEO compensation and firm performance. The results of their research suggested that there exists a positive

relationship. A first proxy that is used often to measure firm value is the Tobin’s Q (Chung & Pruitt, 1996; Mehran, 1995). Mehran (1995) demonstrated a positive relationship between the compensation and Tobin’s Q. A proxy that is often used as measurement for firm

performance is the Return on Equity, known as ROE. Wall Street investors and analysts often focus on the ROE as a primary measure of firm performance. Also, many executives seem to focus heavily on the return on equity since this measure seems to get the most attention from possible investors (Hagel et al., 2010). The measure focuses on the return the shareholders receive. For shareholders this is a good and quick way to make conclusions about the performance of the company you invested in.

However, there are a couple concerns that should be mentioned. A company can divert from their business fundamentals to maintain a healthy return on equity and hide the actual deteriorating performance. Most of the time this is done by letting the debt-equity ratio grow and by performing stock buybacks with cash. With this strategy the investors are kept happy and the CEO gets credits for maintaining a healthy ROE. A decrease of the ROE can be painful for the CEO since it is possible that the stock price drops immediately. On the other side are the risks less immediate and identifiable, therefore it is understandable that CEO’s act in this way. These concerns are the reason many researchers used the return on

assets, known as ROA, as a measure for firm performance (Murphy, 1999; Raithatha & Komera, 2014). ROA receives way less attention from investors and shareholders. It avoids the possible concerns that could occur using the ROE. In the calculation of the ROA the assets that support the business activities are taken into account. Hagel et al. (2010) argue that we rely too heavily on the ROE and that we should use ROA as measurement for the performance of a firm.

§2.5 Industry Sector

Another control variable that is often used in prior research on CEO compensation is the industry sector. Murphy (1999) argued that the industry sector in which the firm is active also has an influence on the CEO compensation. According to Murphy (1999) levels of pay and pay-performance sensitivities are higher for industrial firms and lower for regulated utilities. Also, Houston and James (1995) find in their research that the CEO compensation in the banking industry is lower than in other industries. This has to do with the regulations limiting the scope of a bank’s assets and investment opportunities (Houston & James, 1995). Zhou (2000) looked for interindustry differences. In the financial sector the compensation is the highest and in the utility sector the lowest. The resources and manufacturing industries are in between them. After testing Zhou (2000) concluded that these interindustry

differences are insignificant. Moreover, Anderson et al. (2000) argue that the level of CEO compensation in the information technology industry is the same as in other industries, when performance and other factors are taken into consideration. The findings from prior research are very divided.

§3 Research method §3.1 Sample construction

To construct my sample used in this research, I took the following steps. First, I started looking for data I could use. The databases Compustat – Capital IQ and Execucomp using Wharton Research Data Services provided the data for all firms that are listed on the S&P 500 from 2010-2018. I excluded financial firms before downloading the data because their structures are very different. Secondly, I searched those databases for the variables I needed to run the tests in Stata.

In this research I intend to review the effect of CEO pay regulation for the largest U.S. firms, the relevance of industry and firm-level characteristics for firms on the S&P 500 for the years 2010-2018.

The S&P 500 is a stock index in the U.S. which gives a presentation of the

developments on the stock market in the U.S. The index contains the 500 largest firms of the U.S. measured by their market capitalization. The S&P 500 is composed by the S&P Index Committee that consists of economists and analysts of the credit rating agency Standard & Poor’s. The task of this committee is to set up the index in an independent and objective way. There are many criteria to be listed on the S&P 500, the most important ones are discussed. Firstly, the firm in question must be noted on the American exchange. The second one is that the firm in question must have a market capitalization equal or higher than $5.2 billion. A third important criterion is that a minimum of 50% of the shares must be tradable at the exchange. Also, the firm must be in good financial health and have a liquidity minimum of 0.3. Given all these criteria the S&P 500 gives a good and reliable picture of the performance, value and leverages of firms from the U.S. For that reason, the S&P 500 is an appropriate index for my research.

§3.2 CEO compensation

The dependent variable in this research is the total CEO compensation. According to

Friedman and Jenter (2010), CEO compensation consists of five components: salary, annual bonus, pay-outs from long-term incentive plans, restricted option grants and restricted stock grants. In this analysis, the total value of CEO compensation for the individual year is comprised of the following: salary, bonus, other annual, total value of restricted stock granted, total value of stock options granted (using Black-Scholes), long-term incentive

pay-outs and all other total (Execucomp & COMPUSTAT). This total compensation is used because the data on the components Cash Compensation and Non-Cash Compensation were not available separately. The total compensation can be split up in fixed compensation and variable compensation. The determinants of fixed compensation are salary and other annual. Salary is a fixed cash compensation that will not change on the annual performance of the firm. The same holds for the other annual. The determinants of variable

compensation are bonus, total value of restricted stock granted, total value of stock options granted, long-term incentive pay-outs and all other total. These variables depend on the firm performance. Long-term incentive pay-outs is the amount paid out to the CEO under the firm’s long-term incentive plan. This long-term plan measure firm performance generally over a period of three years. The pay-outs can be seen as a reward for the CEO for

maximizing the shareholders’ value. The amount of bonus, the total value of restricted stock granted, and the total value of stock options granted are performance-pay compensations. This can lead to the CEO maximizing his own value instead of maximizing the value of the firm. These differences in interest are well known as the principal-agent problem. The principal-agent problem is a conflict between the owner of assets and the person who controls the assets and determines how it is used (Agarwal, 2018). It can occur in many situations, but in this case, it is between the stockholders of a firm and the CEO. In this research the CEO compensation is the dependent variable.

§3.3 Regulation

This research intends to view the effects of the implemented regulation on the CEO Pay Ratio in 2015. Two groups are created to test these effects. The first group contains the years 2010 to 2015 and second group the years 2016 to 2018. This way it will be researched if there is a significant change before and after the implementation of the regulation. A dummy variable is added to the regression, which equals 0 for the years 2010-2015 and equals 1 for the years 2016-2018. Based on the literature talked about earlier I expect the regulation implemented in 2015 to have a significant impact on CEO compensation, ceteris

§3.4 Explanatory variables

There are several variables used from which it is likely they influence the compensation and are often used in prior literature. A check for their effect on CEO compensation will be done. The results of these tests will be discussed. In the next section each variable is explained and if needed the formula for the calculation will be given.

§3.4.1 Return on Assets

As discussed in the literature review many researchers looked at firm performance. The proxy most often used is the return on assets (ROA). Therefore, my first explanatory variable is firm performance and is measured by the Return on Assets (ROA). ROA is expressed as a percentage and often used as a proxy to calculate the performance of a firm. Efficiency is extremely important for businesses. ROA gives a manager or analyst a good indication how efficient the assets are performing to generate earnings. A higher ROA means higher efficiency. The relative profitability to its total assets is given by the ROA. Also, when

comparing similar companies or comparing your own company to the efficiency of its assets the year before ROA is often used. Due to prior literature I expect that the return on assets has a negative relation with CEO compensation, ceteris paribus. The Return on Assets is calculated as follows:

ROA = Net Income / Total value of Assets

§3.4.2 Tobin’s Q

A lot of research is done regarding firm value. The proxy that is used most often is the Tobin’s Q ratio. Therefore, the Tobin’s Q is used as measurement of the explanatory variable firm value. Tobin’s Q is the ratio between the physical asset’s market value and its

replacement value. Tobin’s Q ratio is found by dividing the total market value of the firm by the total market value of assets (Berk & DeMarzo, 2014). When the Tobin’s Q is somewhere in between 0 and 1, it costs more to replace a firm’s assets than the firm is worth. If the ratio is above 1 it means that the firm is worth more than the cost of its assets. A firm is overvalued if the ratio is above 1 and a firm is undervalued if the ratio is between 0 and 1. To check whether Tobin’s Q still has an effect on CEO compensation a regression is

performed. Before running the regression, the natural logarithm of the Tobin’s Q is taken to get a normal distribution. Due to prior literature discussed in the literature review I expect

that Tobin’s Q has a negative relation with CEO compensation, ceteris paribus. Tobin’s Q is calculated as follows:

Tobin’s Qi = Total value of Assetsi / Total market value of Firmi

§3.5 Control variables §3.5.1 Firm size

As discussed in the literature view, the firm size is seen by many researchers as one of the most important determinants for CEO compensation. For that reason, I included firm size as a control variable. There exists a well-documented positive relation between total CEO compensation and firm size (Choe et al., 2014). Some prior research used assets as proxy for firm size (Choe et al., 2014; Leone et al., 2006). According to Tosi Jr and Gomez-Mejia (1994) the number of employees measure the firm size. But most of the prior research used sales revenues as a proxy for firm size (Conyon & Murphy, 2000; Dang et al., 2018; Shehata, 1991). In this study as a proxy for firm size the natural logarithm of sales revenue is taken to get a normal distribution. According to prior literature reviewed earlier I expect that

revenue has a positive relation with CEO compensation, ceteris paribus.

§3.5.2 Leverage

Another control variable is the leverage of the firm. In prior literature leverage is used as control variable for the analysis (Coles et al., 2006; Smirnova & Zavertiaeva, 2017). The structure of the CEO compensation gives CEOs the incentive to invest in riskier assets and implement a more aggressive debt policy (Coles et al, 2006). Coles et al. (2006) state that a risky investment will likely have a high leverage ratio. Modigliani and Miller showed that the return of levered equity is higher than then unlevered equity (Berk & DeMarzo, 2014). Ghysels et al. (2005) found a positive and significant relationship between risk and return. A high leveraged investment will therefore have a higher potential and expected return. That is why leverage should be included in the regression as a control variable. For a normal distribution the natural logarithm of leverage is taken. Based on the literature given above I expect there exists a positive relation between the leverage of the firm and the CEO

compensation. In this research the leverage is calculated as follows:

§3.5.3 Return on Equity

Another control variable that is used many times for the measurement of firm performance is the Return on Equity (ROE). Using ROE, you divide the net income by the shareholders’ equity while using ROA you divide income by debt and shareholders’ equity (A=D+E). Therefore, the ROE will always be much higher than the ROA. That is why ROE is also considered as the return on net assets. ROE is considered how effectively management is using the company’s assets to create profit. In my research as proxy for the first explanatory variable firm performance the return on assets is used. Because of multicollinearity return on assets and return on equity can’t be in the same regression. Therefore, I created an interaction variable between return on assets and the leverage. Based on the literature review I expect that the return on equity has a positive relation with CEO compensation,

ceteris paribus. The Return on Equity is calculated as follows: ROEi = Net Incomei / Total value of Equityi

§3.5.4 Industry sector

Another control variable I intend to check is the industry sector. As described in the literature review the conclusions on differences in CEO compensations across industry sectors are divided. A couple sectors are picked to test these differences across industries. Dummy variables for industries are created to separate the observations for sectors. Regressions will be performed for all of those industry sectors to check the effect of the control variables. Based on the literature above I expect that there exists a significant difference in CEO compensation between different industry sectors, ceteris paribus.

§3.6 The model

To give answers on the stated hypotheses, the following regressions are used:

Compensationi,t = b0 + b1X1i,t + b2X2i,t + b3X3i,t + b4X4i,t + (b5X4,i*roe) + ei,t

Compensationi,2010,…,2015 = b0 + b1X1i,t + b2X2i,t + b3X3i,t + b4X4i,t + (b5X4,i*roe) + ei,t

Compensationi,2016,2017,2018 = b0 + b1X1i,t + b2X2i,t + b3X3i,t + b4X4i,t + (b5X4,i*roe) + ei,t

IndustrySectori,t = b0 + b1X1i,t + b3X3i,t + b4X4i,t + b5X5i,t + (b7X5,i*roe) + ei,t

Compensation = total CEO compensation b0 = constant X1 = Return on Assets X2 = lnRevenue X3= lnTobin’s Q X4 = lnLeverage

X4*roe = lnLeverage*ln(Return on Equity)

§4 Descriptive Statistics

In the next sections the descriptive statistics for each variable, group or subsample used are given. For each variable the number of observations (N) is given, the mean, the standard deviation (sd), the lowest value in the sample (min), and the highest value in the sample (max). The whole sample consist of 1,403 observations from companies listed on the S&P 500 for the years 2010-2018. For my research I have created subsamples, groups and I made a distinction in industry sectors to analyse the effect on CEO compensation.

§4.1. Sample

In table 1 the descriptive statistics for the whole sample with all 1,403 observations are given. In my regression with CEO compensation as dependent variable, I added five control variables to the model. In table one the interaction variable between the natural logarithm of the return on equity and the natural logarithm of leverage is left out.

VARIABLES N mean sd min max

ROA 1,403 0.106 0.0591 0.000236 0.300

LnTobin’s Q 1,403 -0.202 0.655 -2.447 1.585

LnRevenue 1,403 8.231 0.865 3.510 10.12

LnLeverage 1,327 -0.666 1.173 -12.95 1.097

Number of year 9 9 9 9 9

Table 1. Descriptive statistics for the whole sample for the control variables for the years 2010-2018.

§4.2 Industry Sectors

Table 2 shows different industry sectors for which firms on the S&P 500 can be listed. For my research I selected some sectors with a relatively large number of observations (>200), some with a bit lower number of observations (<100), and a couple with a number of observations that is in between the two. This way a good subsample is created to test for differences across industry sectors.

Industry Sectors N mean sd min max

Energy 76 7,062 3,196 618.3 13,588 Industrials 254 5,707 3,513 751.09 14,901 Materials 132 5,627 3,069 721.9 14,041 Consumer Discretionary 244 5,159 3,505 31.01 14,910 Healthcare 142 6,773 3,037 742.3 14,121 IT 224 6,195 3,016 49.52 13,379 Communication Services 46 5,048 3,622 1,208 14,458 Real Estate 158 5,371 3,289 355.5 14,213 Number of year 9 9 9 9 9

Table 2. Descriptive statistics for different industry sectors for the years 2010-2018.

Several tests will be performed on industry sectors. First a test with industry dummies is performed to test the significance on CEO compensation. Then a couple regressions will be ran, with the industry sector in question as dependent variable, and the effect of the variables in the model will be checked. Also, the mean compensation before and after the regulation for an industry will be calculated and the difference will be tested on significance. The median for industries is calculated and then subtracted from each outcome in this industry. Then a regression on these deviations is performed to test the effect on CEO compensation.

§4.3 Regulation

Table 3 shows the number of observations before and after the implementation of the regulation on the CEO Pay in 2015. Three fixed effects regressions will be performed. The first one for the whole sample with all 1,403 observations. One regression with 873

observations for the years 2010-2015, and a third regression with 454 observations for the years 2016-2018 are performed. This is done to see the effect of the introduction of the regulation. Table three with an added time dummy shows the observations after the implementation of the regulation on CEO Pay in 2015.

Variables N mean sd min max

ROA 1,403 0.106 0.0591 0.000236 0.300 LnTobin’s Q 1,403 -0.202 0.655 -2.447 1.585 LnRevenue 1,403 8.231 0.865 3.510 10.12 LnLeverage 1,327 -0.666 1.173 -12.95 1,097 Dummy 454 2,017 0.809 2,016 2,018 Number of year 3 3 3 3 3

§5 Results

In this section the results are given and will be discussed. A single * means significance on level p<0.10, a double ** means significance on level p<0.05, and a triple *** means significance on level p<0.01. This holds for all the tables.

§5.1.1 Overall statistics

Table 4 shows three regressions with the control variables for the years 2010-2018. The regressions are used to check the control variables and see if the variables are for a valid reason in the model. To be sure the effects are valid I performed three regressions within each regression other fixed effects. By performing three regressions with different fixed effects the results become much stronger. The first regression is year fixed effects, the second quintile size, and the third industry sector fixed effects. For all three regressions the effect of the return on assets, the natural logarithm of the revenue, and the natural

logarithm of the Tobin’s Q are significant with a significance level p<0.01. Only for fixed effects on quintile size the natural logarithm of the Tobin’s Q is significant with a significance level p<0.05. In all the three models the effect of the natural logarithm of leverage on total CEO compensation is insignificant, just like the effect of the interaction variable.

VARIABLES Fixed Effects Fixed Effects Fixed Effects

ROA -14,932*** -16,742** -16,624*** (1,732) (4,282) (3,732) LnTobin’s Q -1,028*** 1,236** 1,191*** (198.3) (438.7) (273.8) LnRevenue 833.5*** -1,227** -1,412*** (113.6) (326.6) (274.8) LnLeverage 23.09 321.1 368.8 (185.1) (206.4) (304.3) LnROE*lnLev -113.5 -40.91 -75.31 (94.35) (96.22) (82.57) Constant 472.6 -2,588 -2,190 (987.5) (3,836) (2,088) Number observations 1,327 1,327 1,327 #Year/Quintile/Sector 9 5 10 R-squared 0.057 0.060 0.102 Firm FE YES NO NO

Quintile Size FE NO YES NO

Industry Sector FE NO NO YES

§5.1.2 Deviations from median CEO compensation

An extra check on the results for the whole sample is done. A regression with the deviations from the median CEO compensation as dependent variable is performed. The results are shown below in table 5. For the ROA, revenue and the Tobin’s Q the effects show the same significance on CEO compensation as in table 4. The control variable leverage is significant with a significance level p<0.05, whereas earlier it was proven to be insignificant. The results show that the control variables are for a valid reason in the model and show a relation with CEO compensation. For all four regressions the revenue shows a positive relation with CEO compensation, so I conclude that revenue has a positive relation with firms’ revenues. This is in line with my first hypothesis.

VARIABLES Fixed Effects

ROA -13,769*** (1,582) LnRevenue 823.3*** (131.4) LnTobinsq -1,016*** (129.8) LnLeverage 272.6** (102.0) Constant -4,838*** (1,001) Observations 1,327 Number of year 9 R-squared 0.057 Firm FE YES

Table 5. Results on deviations from median for CEO compensation for the years 2010-2018.

§5.2 Regulation

In table 5 the results of three regressions are shown. The first one shows the years 2010-2018, the second one shows the years 2010-2015, and the third displays the years 2016-2018. The third one is after the introduction of regulation, which was implemented in 2015. For the years 2010-2015 the return on assets, the natural logarithm of revenue and the natural logarithm of the Tobin’s Q are significant with a significance level of p<0.01. For the years 2016-2018 the return on assets is significant with a significance level of p<0.05, the natural logarithm of the Tobin’s Q is significant with a significance level of p<0.10, and the natural logarithm of the revenue is significant with a significance level of p<0.01. The effect

of the return on assets and lnTobin’s Q is not very different after the regulation. The lnRevenue is also still significant with significance level p<0.01. However, the effect of the lnRevenue becomes almost twice as strong. What also stands out is the changing effect of lnLeverage. After the regulation is introduced the effect becomes positive and significant with a level p<0.10, whereas before the regulation the effect was negative. A possible reason for this change could be that the test before the introduction of the regulation also includes the years 2010 and 2011. In these years a lot of companies were still recovering from the financial debt crisis in 2007-2008 and the confidence was low. In recent years the economic growth in the U.S. was big. This is accompanied by growth in confidence. So, in the years 2016-2018 firms took on more leverage and increased their debt financing in investments. In sum, there is little change in effects of the control variables on CEO compensation except for the leverage. This suggests that the effect of the introduction of the regulation is very small.

Years 2010-2018 2010-2015 2016-2018

VARIABLES Fixed Effects Fixed Effects Fixed Effects

ROA -14,932*** -14,942*** -13,164** (1,732) (2,326) (2,224) LnTobin’s Q -1,028*** -950.0*** -1,077* (198.3) (145.2) (272.6) LnRevenue 833.5*** 636.3*** 1,184*** (113.6) (144.2) (10.25) LnLeverage 23.09 -68.92 555.6* (185.1) (180.7) (172.0) LnROE*lnLev -113.5 -164.7 140.9 (94.35) (97.90) (108.0) Constant 472.6 1,646 -1,721** (987.5) (982.2) (306.7) Number observations 1327 873 454 Number year 9 6 3 R-squared 0.057 0.047 0.090

Firm FE YES YES YES

Table 6. Results on CEO compensation for the years 2010-2018, 2010-2015, and 2016-2018.

§5.3 Industry Sectors

§5.3.1 Overall results with industry dummies

In the table below are all the industry sectors added as dummies. A regression is performed with industry sector fixed effects. This regression gives a good picture for all the sectors. All industry sectors are significant with a significance level p<0.01.

VARIABLES Fixed Effects ROA -16,624*** (3,744) lnTobinsq -1,412*** (275.8) lnRevenue 1,191*** (274.7) lnLeverage 368.8 (305.4) lnroelnLeverage -75.31 (82.85) Materials -2,561*** (203.3) Industrials -2,246*** (236.8) Cons. Discretionary -2,887*** (271.2) Cons. Staples -2,429*** (308.0) Healthcare -1,043*** (249.3) Financials -277.8*** (78.90) Info. Technology -1,538*** (212.2) Com. Services -2,079*** (95.19) Real Estate -2,162*** (212.5) Constant -253.1 (1,985) Observations 1,327 Number of gsector 10 R-squared 0.1022 Sector FE YES

Table 7. Results on regression with industry dummies for the years 2010 to 2018.

§5.3.2 Significance of control variables on industry sectors

A regression for a particular industry sector with the CEO compensation as dependent variable is performed to check the effect of the control variables. For the industry sectors consumer discretionary, financials, industrials, information technology, and communication services the results are shown below in table 9. The return on assets is for all industry sectors significant. Also, Tobin’s Q and the revenue have for most of the sectors a significant

effect on CEO compensation. The leverage and the interaction variable between return on equity and leverage have fewer significance in the tested sectors. The reason there are differences in effects among industries has to do with what is important for this particular industry. For example, information technology has very few assets compared to the industry sector. Therefore, the effect of an increase in the return on assets for the information technology sector is smaller than for the industrial sector. Also, an increase in leverage in the information technology sector has no significant effect on CEO compensation, while for the industrial sector this has a significant negative relation.

Con. Discr. Financials Industrials IT Com. Services

VARIABLES Fixed Effects Fixed Effects Fixed Effects Fixed Effects Fixed Effects

ROA -7,379** -43,200** -37,515*** -12,973* -31,728* (2,874) (15,770) (5,525) (6,688) (13,923) LnTobin’s Q 122.6 -4,016** -1,940*** -1,148** -2,875** (308.4) (1,425) (567.9) (492.8) (966.2) LnRevenue -304.8 -732.8 2,061*** 1,187*** 3,146 (245.5) (722.1) (270.5) (353.1) (1,047) LnLeverage -512.4* -531.7 -1,432*** 101.1 -1.366 (243.8) (2,044) (278.0) (454.8) (2,097) LnROE*lnLev -731.7*** -553.8 -790.9*** -132.0 -280.8 (173.3) (920.9) (110.7) (216.3) (1,214) Constant 9,361*** 18,018** -8,243*** -2,186 -16,130* (2,052) (7,284) (2,130) (2,591) (7,861) Observations 222 63 248 192 44 Number of year 9 9 9 9 9 R-squared 0.055 0.236 0.239 0.095 0.362

Firm FE YES YES YES YES YES

Table 8. Results on regression industry sectors for the years 2010-2018.

§5.3.3 Deviations from industry median

The median of several industries is collected and for each outcome the deviation from this median is calculated. For eight sectors regressions are performed and the results are shown below in table 9. These regressions are performed to see if the effects of the control

variables are different if you take the deviation from the industry sector median. In every regression the deviation from the industry median is the dependent variable. For the industrial, the financial, and the consumer discretionary sector the results are exactly the same as in table 8. For the communication services the ROA and Tobin’s Q are the same but the effect of revenue differs. What stands out is that every variable for the industrials sector

is significant with a significance level of p<0.01. The differences in results when the deviation from the median is taken is not very large compared to table 8. Based on these results and the results in section 5.3.2 I conclude that for most industries the ROA, lnRevenue, and lnTobin’s Q have a significant effect on CEO compensation.

Sector Materials Energy Financials Healthcare Industrials Com. Services Real Estate Cons. discr.

VARIABLES Fixed Effects Fixed Effects Fixed Effects Fixed Effects Fixed Effects Fixed Effects Fixed Effects Fixed Effects

ROA -6,387 -10,658 -43,200** -15,654 -37,515*** -31,728* -21,893 -7,379** (7,175) (7,440) (15,770) (9,821) (5,525) (13,923) (15,284) (2,874) LnTobin’s Q -1,837** -2,368*** -4,016** -837.3 -1,940*** -2,875** -1,904 122.6 (758.6) (582.5) (1,425) (717.8) (567.9) (966.2) (1,477) (308.4) LnRevenue 1,142** 503.8 -732.8 1,270*** 2,061*** 3,146** 1,170** -304.8 (341.1) (873.9) (722.1) (313.1) (270.5) (1,047) (360.1) (245.5) LnLeverage 953.3 668.9 -531.7 -20.67 -1,432*** -1.366 -527.7 -512.4* (539.1) (754.8) (2,044) (386.8) (278.0) (2,097) (1,904) (243.8) LnROE*lnLev -203.4 -368.8 -553.8 -181.7 -790.9*** -280.8 -908.2 -731.7*** (361.3) (247.6) (920.9) (306.8) (110.7) (1,214) (828.1) (173.3) Constant -8,314** -965.3 12,307 -8,483** -13,184*** -20,344** -6,782** 5,384** (2,695) (6,834) (7,284) (2,834) (2,130) (7,861) (2,041) (2,052) Observations 132 76 63 131 248 44 158 222 Number of year 9 9 9 9 9 9 9 9 R-squared 0.175 0.336 0.236 0.119 0.239 0.362 0.119 0.055

Firm FE YES YES YES YES YES YES YES YES

Table 9. Results of regression on deviations from the median per industry for the years 2010-2018.

§5.3.4 Differences across industry sectors

Prior research suggested that across industries there exist differences in CEO compensation. Therefore, several t-tests on the means for two industry sectors are performed to check if there is a significant difference across industries. If the test shows a significant p-value of p<0.05 does this mean that there is a difference in CEO compensation between these two industry sectors. The next Stata-command is used to run these tests: ttest sector1==sector2,

unpaired unequal.

sector1 obs sector2 obs t-value p-value 2-sided

Energy 76 Real Estate 158 3.7552 0.002***

Energy 76 Con. Discr. 244 4.4284 0.000***

Healthcare 142 Materials 132 3.1052 0.002***

IT 224 Com. Services 46 2.0105 0.049**

Energy 76 IT 224 2.0725 0.043**

Materials 132 Real Estate 158 0.6849 0.494

For the energy industry the mean CEO compensation is significantly higher than the mean CEO compensation for the real estate and consumer discretionary industries with a

significance level of p<0.01. The mean of the healthcare industry is significantly higher than the mean in the materials sector with a significance level of p<0.01. For the information technology sector, the mean is also significantly higher than in the communication services industry with a significance level of p<0.05. On the other side the difference between the mean of materials and real estate is insignificant. For the difference between these two sectors I can’t conclude there is a significant difference in mean CEO compensation.

However, most of the test effects are significant. The conclusions from prior research were divided, but my results are very clear. Therefore, based on the tests performed I conclude there exists a significant difference in mean CEO compensation across industries which is in line with my second hypothesis.

§5.3.5 Industry sector means before and after regulation

To test the effect of the regulation on the mean compensation per industry every industry sector is divided into two groups. One group before the introduction of the regulation, this group contains the observations for the years 2010-2015. The second group after the

introduction of the regulation, this group contains the observations for the years 2016-2018. Then a test is performed on the differences in the means for both groups. The following command is used: ttest Sector1after==Sector1before, unpaired unequal

sector obs total mean before mean after t-value p-value

IT 224 5534.304 7534.85 -5.019 0.000*** Energy 76 6245.934 9200.014 -4.049 0.000*** Real Estate 158 4907.551 6148.153 -2.379 0.019** Materials 132 4920.846 6947.031 -3.692 0.000*** Con. Discr. 244 4342.145 6714.025 -4.804 0.000*** Healthcare 142 6482.056 7381.553 -1.584 0.117 Com. Services 46 5738.226 3468.862 2.717 0.009**

Table 11. Results on t-tests on means within one industry sector before and after the introduction of regulation. Several t-tests from table 11 show a significant effect on the mean CEO compensation before and after the implementation of the regulation on the CEO Pay Ratio in 2015. What is remarkable is that in most cases the mean compensation is bigger after the regulation. Only for communication services the mean before the introduction of the regulation is bigger. Moreover, there is no significant change in the mean CEO compensation for the healthcare

sector. This is not surprising since the test with the health care sector as dependent variable also showed only for one variable significance. A possible reason for this change in mean compensation is the economic growth. In the recent years the economy in the U.S. has grown hard, with as result an increased CEO compensation. This explains the significant increase in average CEO compensation for most of the industry sectors. For most industries the mean compensation is significantly higher after the introduction of the regulation than before. This means that the introduction of CEO compensation regulation has no effect in these industry sectors. The regressions on the whole sample also showed little change in effects on CEO compensation, suggesting that the introduction of the regulation has significant effect. To check this suggestion a t-test is performed on the mean total compensation after the introduction of the regulation and the mean before the introduction. The t-value is 8.81, which means that the mean CEO compensation after introduction is higher than before the introduction with a significance level p<0.01. Based on the results on the mean in industry sectors and the t-test performed for the whole sample I conclude that the introduction of the regulation in 2015 has no significant effect on CEO compensation and the impact is not valid for the various industry sectors. This not in line with my third hypothesis, the opposite is true.

Mean before Mean after t-value p-value

§6 Conclusion and discussion

In this thesis the effect of regulation on CEO pay and the relevance for industry and firm-level characteristics for the largest U.S. firms in the S&P 500 is researched. This research is executed on 1,403 firms listed on the S&P 500 for the years 2010-2018. The next research question is formulated to investigate this: What is the effect of CEO pay regulation for largest U.S. firms, the relevance of industry and firm-level characteristics? The used control

variables are return on assets, revenue, Tobin’s Q, and leverage. An interaction variable between the return on equity and leverage is added to the model. The total CEO

compensation is the dependent variable.

To check the effects for the control variables on CEO compensation four regressions are performed. The first one with firm fixed effects, the second one with quintile size fixed effects, and the third one with industry fixed effects. The dependent variable for the fourth regression is the deviation from the median CEO compensation. For the variables ROA, Tobin’s Q, and revenue their significant effects are confirmed in all four regressions and thus are for a valid reason in the model. The significant effect of leverage is proven in the

regression with the deviations from the median CEO compensation. This shows that the control variable leverage is for a valid reason in the model. Two regressions are performed to check the effects of the introduced regulation. Multiple regressions are performed to test the effects of the control variables on industry sectors. T-tests are performed to test the effect of regulation on the mean CEO compensation for industry sectors and test for differences in CEO compensation across industry sectors.

In my research I intended to find the answers on three hypotheses stated. In all regressions performed, with CEO compensation as dependent variable, the control variable revenue showed a significant positive effect on CEO compensation. For this reason, I

conclude that revenue has a positive relation with CEO compensation which is in line with my first hypothesis. This is also in line with my expectations and research done by others.

The tests on the differences in mean CEO compensation across industry sectors showed in almost every case a significant difference, therefore I conclude CEO

compensation differs across various industry sectors, and this in line with my second hypothesis. This also confirms the second question stated in the introduction. The results from prior research were divided but beforehand I expected there to be differences across industry sectors.

For the third hypothesis regressions are performed before the regulation and after the regulation. The results showed the same effects for the return on assets, revenue, and the Tobin’s Q. Only the effect of leverage became significant positive after the regulation with p<0.10. This suggests that after the introduction of the regulation there was little effect on CEO compensation. To test the change in mean CEO compensation for industry sectors t-tests are performed. The results showed a significant higher compensation after the

regulation and that is why I conclude that the regulation has no effect on CEO compensation and is not valid for various industry sectors. This is not in line with my third hypothesis and is thus not confirmed. The answer on the third question stated in the introduction is no, regulation is not able to restrict the continue increase in CEO compensation.

In sum, the thesis confirmed the effects of the known control variables on CEO compensation for the firms on the S&P 500 for the years 2010-2018. Also, firms’ revenues is positively related to CEO compensation. Moreover, CEO compensation differs across

industry sectors. Furthermore, the regulation in 2015 has no significant effect on CEO compensation and is not valid for various industry sectors.

In my research several explanatory and control variables were added, but not all of them. Possible extensions would include an additional set of control variables. Three

examples of variables that could have been included are the age of the CEO, the tenure, and the size of the board of the firm can influence the total compensation received by the CEO. For that reason, it is possible that my variables are biased because of these omitted

variables. Also, I could have made a distinction between fixed and variable compensation. Furthermore, I could have looked at additional sectors. That is why conclusions can be drawn from my results, but with caution.

For future research I would suggest including more control variables and test their effect on CEO compensation. Also, focus on additional sectors to get more unbiased results. Furthermore, it becomes clear that more regulation is needed to restrict the continue increase of CEO compensation. For example, regulation on the ratio between fixed and variable compensation could be introduced. Another example is that corporate governance could give shareholders more control to limit the CEO’ compensations. Options like these should be researched in the future.

References

Agarwal, P. (2018). The principal-agent problem. The intelligent economist. Retrieved from: https://www.intelligenteconomist.com/principal-agent-problem/

Anderson, M. C., Banker, R. D., & Ravindran, S. (2000). Executive Compensation in the Information Techonology Industry. Management Science, 46(4), 530-547.

Bebchuk, L., & Grinstein, Y. (2005). Firm expansion and CEO pay (No. w11886). National

Bureau of Economic Research.

Berk, J., & DeMarzo, P. (3rd _{edition) (2014). Corporate finance. Location: Pearson.}

Cerasi, V., Deiniger, S. M., Gambacorta, L., & Oliviero, T. (2020). How post-crisis regulation has affected bank CEO compensation. Journal of International Money and Finance, 1-19.

Choe, C., Tian, Y. G. & Yin, X. (2014). International Review of Financial Analytics, 35, 237-248.

Chung, K. H., & Pruitt, W. (1996). Executive ownership, corporate value, and executive compensation: A unifying framework. Journal of Banking and Finance, 20(7), 1135-1159.

Coles, J. L., Daniel, N. D., & Naveen, L. (2006). Managerial incentives and risk-taking. Journal

of financial economics, 79(2), 431-468.

Colonello, F., Koetter, M., & Wagner, K. (2019). Compensation Regulation in Banking: Executive Director Behavior and Bank Performance after the EU Bonus Cap. Conyon, M. J. (1997) Corporate governance and executive compensation. International

Journal of Industrial Organization, 15, 493-509.

Dang, C., Li, Z., & Yang, C. (2018). Measuring firm size in empirical corporate finance. Journal

of Banking and Finance, 86, 159-176.

Gabaix, X., & Landier, A. (2008). Why has CEO pay increased so much? The Quarterly

Journal of Economics, 123(1), 49-100.

Edmans, A., & Gabaix, X. (2016). Executive compensation: A modern primer. Journal of

Economic Literature, 54(4), 1232-1287

Hagel, J., Brown, J. S., & Davison, L. (2010, March). The best way to measure company performance. Harvard Business Review. Retrieved from

Hall, B. J., & Liebman, J. (1998). Are CEOs really paid like bureaucrats? The Quarterly Journal

of Economics, 113(3), 653-691.

Houston, J. F., & James, C. (1995). CEO compensation and bank risk. Is compensation in banking structured to promote risk taking? Journal of Monetary Economics, 36, 405-431.

Ghysels, E., Santa-Clara, P., & Valkanov, R. (2005). There is a risk-return trade-off after all.

Journal of Financial Economics, 76(3), 509-548.

Guo, L., Jalal, A., & Khaksari, S. (2015). Bank executive compensation structure, risk-taking and the financial crisis. Review of Quantitative Finance and Accounting, 45(3), 609-639.

KPMG. (2017, September). United States – New Implementation Guidance from SEC on CEO Pay Ratio Disclosure. Retrieved from

https://home.kpmg/xx/en/home/insights/2017/09/flash-alert-2017-142.html Leone, A. L., Wu, J. S., & Zimmerman, J. L. (2006). Asymmetric sensitivity of CEO cash

compensation to stock returns. Journal of Accounting and Economics, 42(1-2), 167-192.

Mehran, H. (1995). Executive compensation structure, ownership and firm performance.

Journal of financial economics, 38(2), 163-184.

Mishel, L., & Wolfe, J. (2019). CEO compensation has grown 940% since 1978. Economic

Policy Institute. Retrieved from

https://www.epi.org/publication/ceo-compensation-2018/

Murphy, K. J. (1999). Executive compensation. Handbook of labour economics, 3, 2485-2563. Murphy, K. J. (2013). “Executive Compensation: Where We Are, and How We Got There”.

Handbook of the Economics of Finance, 2A, 211-356.

Piketty, T. (2015). Capital in the Twenty-First Century. American Economic Review, 105(5), 48-53.

Piketty, T., & Saez, E. (2003). “Income Inequality in the United States, 1913-1998.” Quarterly

Journal of Economics, 118(1), 1-41.

Raithatha, M., & Komera, S. (2016). Executive compensation and firm performance: Evidence from Indian firms. IIMB Management Review, 28(3), 160-169. Richman, L. D., & Hermsen, M. L. (2015). The SEC’s Pay Ratio Disclosure Rule And Its

Tosi Jr, H. L., & Gomez-Mejia, L. R. (1994). CEO compensation monitoring and firm performance. Academy of Management Journal, 37(4), 1002-1016.

Shehata, M. (1991). Self-Selection Bias and the Economic Consequences of Accounting Regulation: An Application of a Two Stage Switching Regression to SFAS No. 2, The Accounting Review, 66(4), 768-787.

SEC. (2015, August). SEC Adopts Rule for Pay Ratio Disclosure. Press Release. Retrieved from https://www.sec.gov/news/pressrelease/2015-160.html

Smirnova, A. S., & Zavertiaeva, M. A. (2017). Which came first, CEO compensation or firm performance? The causality dilemma in European companies. Research in

International Business and Finance, 42, 658-673.

Wood, A., & Posner, C. (2013). Securities Disclosure. The Corporate & Securities Law Advisor, 27(10), 2-9.

Appendix

Figure 1. CEO/Average worker compensation ratio. Source: Economic Policy Institute

Rule of thumb for control variables 0 .1 .2 .3 .4 .5 D e n si ty 4 6 8 10 lnRevenue 0 .2 .4 .6 .8 D e n si ty -3 -2 -1 0 1 2 lnTobinsq 0 .2 .4 .6 D e n si ty -4 -3 -2 -1 0 1 lnLeverage 0 .2 .4 .6 .8 D e n si ty -4 -3 -2 -1 0 lnroe 0 2 4 6 8 D e n si ty 0 .1 .2 .3 roa

Variables Definitions Level Dependent variable

CEO compensation Total CEO compensation x1,000,000

Firm performance and value

ROA Operating income divided by total assets Ratio

Tobin’s Q Total assets divided by (share price*shares outstanding) Ratio

ln Tobin’s Q Natural logarithm of Tobin’s Q Logarithm

Control Variables

Revenue Total sales revenue x1,000,000

ln Revenue Natural logarithm of revenue Logarithm

Leverage Total debt divided by total equity Ratio

ln Leverage Natural logarithm of leverage Logarithm

ROE Operating income divided by total equity Ratio

ln ROE Natural logarithm of return on equity Ratio

lnROE*ln*Leverage Natural logarithm ROE x natural logarithm leverage Ratio