The effect of the CEO pay ratio on employee satisfaction

(1)

The effect of the CEO pay ratio on employee satisfaction

Bachelor’s Thesis in Economics and Finance University of Amsterdam

Faculty of Economics and Business Author: Janine van der Woude Student number: 10788786 Date: June 26, 2018

Track: Economics and Finance Field: Microeconomics

Supervisor: Patrick Stastra

Abstract

This paper examines the effect of CEO pay ratios on employee satisfaction. CEO pay ratios reflect the ratio between CEO compensation and median employee compensation. Higher ratios can have positive effects on productivity and firm performance and negative effects on employee satisfaction due to inequality faced by the employees, potentially influencing the level of employee satisfaction. This is examined by analyzing 429 S&P500 firms. The results show that there is no effect of CEO pay ratios on employee satisfaction. Limitations of this research are the modification of a Likert scale to a percentage scale and the voluntary participation of employees in Glassdoor’s research. Future research can be done to check whether SEC’s mandatory disclosure has an effect or by matching salaries with employees’ reviews.

(2)

2 Statement of Originality

This document is written by Janine van der Woude who declares 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.

(3)

3

1. Introduction

The wages of top executives are highlighted by well-known newspapers and media. The issue that is emphasized in those articles is the increasingly high levels of CEO pay ratios, which reflect the relative difference between compensation of a company’s CEO to that of the median employee (Gelles, 2018; Anderson & Pizzigati, 2018). Gelles (2018) reported in The New York Times that at Walmart, the median employee - earning $19,177 in 2017 - should work for over a thousand years to earn the $22.2 million salary that was awarded to CEO Doug McMillon in 2017. The levels of CEO compensation are still rising. In America, the average raise of median pay for the 200 highest-paid CEOs was 5% in 2015, increasing to 9% in 2016 and 14% in 2017 (Gelles, 2018).

The U.S. Securities and Exchange Commission (SEC) has been monitoring these CEO pay ratios for years and sets rules and regulation on it. On the fifth of August 2015, SEC revealed the final rules on CEO pay ratio disclosure, which complements the Dodd-Frank Act that promotes accountability and the information availability to investors (Aguilar, 2015). As of early 2018, companies are required to make disclosures on CEO compensation, median employee compensation and CEO pay ratios (SEC, 2017).

These increased levels of transparency give employees an insight in how much they earn relative to their principal. When employees value their wage too low for the effort they make, i.e. to what they value as their fair wage, productivity will be dropped, which lowers firm performance (Akerlof & Yellen, 1990). Additionally, Melián-González, Bulchand-Gidumal and González López-Valcárel (2015) find a positive relationship between employee

satisfaction and firm performance. So when firm performance lowers due to reduced

productivity, employee satisfaction is expected to be lower as well. Research has been done on relationships between these variables, but not all combinations have been researched yet. Therefore, the purpose of this paper is to execute an empirical analysis on the

relationship between CEO pay ratios and employee satisfaction at American S&P500 firms, with the aim of answering the following research question: Does the CEO pay ratio affect employee satisfaction?

To answer this question, 429 S&P500 firms will be examined using employee satisfaction as dependent variable and CEO pay ratio as independent. Productivity,

(5)

5 added in an extended regression formula to check for industry differences. The results of the OLS regression show no significant effect of CEO pay ratio on employee satisfaction. The control variables for productivity and performance show positive relations with employee satisfaction in the original formula, and for all three control variables in the extended formula.

In the following paragraph, employee satisfaction and CEO pay ratios will be discussed in more detail. In paragraph 3 the model will be introduced, including a brief explanation of the variables and the method used. Findings and results are presented in paragraph 4 and the paper is concluded with the conclusion in paragraph 5.

2. Literature review and hypothesis development

In this paragraph previous research results and important findings on employee satisfaction and CEO pay ratio will be discussed and examined.

2.1. Employee satisfaction

Job satisfaction, which is similar to employee satisfaction, is defined by Locke (1976, p.1304) as “a pleasurable or positive emotional state resulting from the appraisal of one’s job or job experiences.” The level in which an employee is satisfied with their job and work

environment influences their performance, productivity and firm value (Edmans, 2012; Huang, Li, Meschke & Guthrie, 2015; Melián-González et al., 2015). Therefore, it is beneficial for a company to have satisfied employees and find factors that can raise the level of

satisfaction, making employees more productive and profitable for a firm.

The relation between employee satisfaction, financial performance and productivity is examined by Melián-González et al. (2015). They used detailed Glassdoor (Glassdoor.com) data to measure employee satisfaction and revenue per employee as a measure for

productivity. The result they found on the influence of productivity on employee satisfaction is positive. Taking a company’s return on assets (ROA) and operating margin as indicators for financial performance, they find a positive relation between employee satisfaction and firm performance, with job satisfaction as independent variable. This result is supported by Huang et al. (2015), who measured performance by the return on assets (ROA) and Tobin’s q – which is the ratio of the market value of assets to the book value of assets –, finding a

(6)

6 positive association between employee satisfaction as independent variable and next

quarter’s Tobin’s q and ROA as dependent variables.

Edmans’ (2012) interest was in the relationship between job satisfaction and firm value. He researched this by linking this year’s performance to previous year’s satisfaction. By doing so, he eliminates reversed causality. To measure job satisfaction, he used the rankings of the Best Companies list for a 28-year time period, which consists of two

questionnaires: the Trust Index employee survey and the survey from Culture Audit, which has to be completed by the management. As performance measure he used future stock return, because this reflects the change in market value of a stock between two years, which gives an appropriate indication of the effect of satisfaction on firm value. His main finding was a positive link between job satisfaction and firm value.

Pfeffer and Langton (1993) did research on wage dispersion within academic

departments at colleges and universities. They investigated the effects of wage dispersion – the difference between wages of all members of an academic department – on, among others, satisfaction and productivity. Pfeffer and Langton (1993) argue that satisfaction is positively related to several factors. For example, an individual’s salary, the number of years in service, whether or not an individual plays an important role within a department, the amount of social contact with colleagues, etc. Their main finding on satisfaction is that there is a significant negative relationship between salary dispersion and satisfaction, especially for the employees with lower salaries. They found that the most unhappy were those who earn less money and face higher levels of salary dispersion, followed by those earning less money but facing lower levels of salary dispersion, more satisfied are those earning more money but facing higher levels of salary dispersion and the most satisfied were the people who earn more money and face low levels of salary dispersion (Pfeffer & Langton, 1993). So it would be more likely to find a negative relationship between satisfaction and pay ratios for those who earn lower salaries and are therefore further from the CEO’s salary and positive relationships for those who earn higher salaries.

The conclusions that can be drawn from the paragraph above, is that there are positive relations between employee satisfaction and productivity, performance and firm value. Furthermore, a negative relationship is found between salary dispersion and satisfaction.

(7)

7

2.2. CEO pay ratio

CEO pay ratio is a subject that has been studied in various researches. According to section 953(b) of The Dodd-Frank Act, CEO pay ratio is defined as “the ratio of the annual total compensation of its CEO to the median of the annual total compensation of all its employees, except the CEO” (SEC, 2015). Faleye, Reis and Venkateswaran (2013) did research on the relationship between CEO pay ratios and employee productivity. They find a positive effect of CEO pay ratios on productivity at firms with stronger tournament incentives, when the firm has fewer employees and when there are higher levels of transparency within a firm – i.e. when employees are aware of their CEO’s wage. This last result is supported by research of Lacmanović (2013), who finds that pay ratio disclosure – i.e. making the wages publicly available – has a positive impact on company performance, employee motivation, behaviour and productivity. This suggests that the new SEC rules (SEC, 2017) for mandatory disclosure can have positive effects on productivity and firm performance, but this goes beyond the scope of this paper.

The positive effect of CEO pay ratios on productivity at firms with stronger

tournament incentives (Faleye et al., 2013) is also explained and supported by Lazear and Rosen (1981), who argue that in these tournaments – in which the best employee wins and gets a promotion – larger salary dispersion provides incentives to work harder to have a chance to win the tournament. So in this case, larger CEO pay ratios may increase productivity. However, when employees think their salary is unfair and less than they deserve, they will work less hard, decreasing productivity to the wage they get paid (Akerlof and Yellen, 1990). This decreases a firm’s productivity.

Pfeffer and Langton (1993) found – besides the results mentioned in paragraph 2.1. – a negative relation between salary dispersion and a researcher’s productivity, or, in other words, his research performance. This result is supported by Bloom (1999), who conducted research on pay distribution at major league baseball to see whether different levels of compensation among team members have an effect on their individual and team

performance. He found that pay dispersions have negative effects on performance and that dispersion may cause feelings of injustice in various forms; psychological, social and

economic.

When concluding the findings in this paragraph, there are several conflicting results found. While there are positive relationships found between CEO pay ratios and productivity

(8)

8 when tournament incentives are strong, a firm has fewer employees and there is

transparency, others found negative relationships between wage dispersion and productivity and performance.

2.3. Hypothesis development

Evaluating the literature above, several conclusions can be drawn. At first, there are positive relationships found between employee satisfaction and productivity, firm performance and firm value (Edmans, 2012; Huang et al., 2015; Melián-González et al., 2015). Furthermore, Pfeffer and Langton (1993) found a negative relation between salary dispersion and satisfaction. Regarding CEO pay ratios, there is a positive relation found between these ratios and productivity when tournament incentives are strong, a firm has fewer employees and there is transparency within the firm (Faleye et al., 2013; Lacmanović, 2013). However, when employees value their wage as unfair and too low for the effort made, they will work less hard, which reduces overall company productivity (Akerlof & Yellen, 1990). Finally, Pfeffer and Langton (1993) and Bloom (1999) found negative relationships between wage dispersion and productivity and performance.

Since SEC (SEC, 2017) set new rules for mandatory disclosure on executive compensation and CEO pay ratios, it can be assumed that there used to be a lack of

transparency. This argument makes the result of Faleye et al. (2013) and Lacmanović (2013) – i.e. the positive relation when a firm is more transparent – less valuable in this situation. The remaining findings regarding CEO pay ratios suggest a negative relationship between salary dispersion and performance and productivity. Since there is a negative relation found between salary dispersion and satisfaction, it is expected that a higher CEO pay ratio will have a negative effect on employee satisfaction, also due to the inequity and lack of fair wages employees will experience (Akerlof & Yellen, 1990). Weighing out the literature above results in the expectance of a negative relationship between CEO pay ratios and employee satisfaction, resulting in the following hypothesis:

(9)

9

3. Research methodology

In this paragraph the sample and data used are explained, then the model will be introduced, followed by the hypotheses that will be tested and an explanation of the variables.

3.1. Sample and data

The data on employee satisfaction and the CEO pay ratios are retrieved from Glassdoor (Chamberlain, 2015). Glassdoor (Glassdoor.com) is a well-known website where employees can review their employer on an anonymous basis. This gives the employees an opportunity to speak freely about overall thoughts on their work environment and job satisfaction.

Until the beginning of 2018, companies were not obliged to report executive compensation and employee wages. Since these numbers were not available, Glassdoor started in 2008 with conducting their own dataset on employee salaries at S&P500 firms, by collecting anonymous reports from employees on a voluntary basis, with the aim of

encouraging pay transparency within companies. To calculate the employee’s median wages, salary reports between 01-01-2009 and 17-08-2015 are used and adjusted for inflation into 2014 dollars (Chamberlain, 2015). The salaries used for the pay ratios, are those of 2014. They obtained the data on CEO compensation from SEC proxy filings. To ensure statistical validity, only companies where at least 30 employee reports were shared are included. This was the case for 447 of the total of 500 S&P500 firms. For 6 firms the SEC proxy filings were unavailable, resulting in a final sample of 441 firms.

The data on employee satisfaction which is available on Glassdoor, is retrieved via their own research as well, by collecting employee satisfaction ratings via surveys and reviews on their employers from 2015. The questions regarding the satisfaction ratings were divided into six categories; overall company satisfaction, career development,

compensation, benefits, the senior management and the balance between work and personal life. The overall company satisfaction rating is used in this paper, because this eliminates, among others, the satisfaction that is caused by higher levels of compensation only and makes it an adequate representation of employee satisfaction. These ratings are based on a 5 point Likert scale. The ratings can take scores between 1.0 and 5.0 with one decimal place, therefore they can be treated like continuous numbers.To use these ratings

(10)

10 for the regression, the scores are divided by 5. This modification is done to make it a

percentage scale (Jamieson, 2004). Since the original ratings are not completely continuous, there is a risk of a potential bias. However, the numbers of Glassdoor’s research take values with one decimal place in their dataset, which assumes that the values can have many possible outcomes and that Glassdoor’s researchers treated the results as interval data (Mangiafico, 2016). Knapp (1990) argues that ordinal data can be treated as interval data when the results and conclusions are scale-free, which is the case in this paper – it only matters if CEO pay ratios will increase or decrease employee satisfaction. According to Jamieson (2004), many researchers describe their data using means and standard deviations and use parametric tests to analyse the ordinal data. According to the Central Limit

Theorem, means are approximately normally distributed when sample size is larger than 10, regardless of the original distribution (Normal, 2010). Norman (2010) states that parametric tests can be used when sample sizes are larger than 10 per group – which is the case for the Glassdoor data with at least 30 responses per employer.

The overall company ratings date from 2015 and the CEO pay ratios from 2014. The reason for this is that satisfaction can, among other things, be based on perceived pay ratios, provided that these are disclosed. This is in line with Edmans’ (2012) reasoning on the

comparison between satisfaction and firm value, where he linked a year’s performance to prior year’s satisfaction. In this paper this reasoning is used the other way around, but because unfair wages can affect productivity negatively (Akerlof & Yellen, 1990), it is useful to use 2014’s pay ratios and 2015’s overall company ratings.

The data for the control and dummy variables are all obtained via Compustat and date from 2014, since these control and dummy variables are expected not to change much within one year. Revenue per employee is used to measure productivity. Both data on total revenue and the number of employees can be found on Compustat. Total revenue is divided by the number of employees to calculate the revenue per employee, i.e. productivity. Return on assets (ROA) is used to measure performance, and is calculated by dividing net income by total assets, which are both obtained via Compustat. Tobin’s q is not used as second

performance measure, which was used by Huang et al. (2015), since the required data was not available. The final control variable is firm size, measured by the number of employees. The data for the industry dummies are available on Compustat as well and divided into ten categories, which are presented in Appendix 1 (United States Department of Labor, n.d.).

(11)

11 Data on the key variables was unavailable for 12 out of 441 companies, leaving with a total sample size of 429 S&P500 companies.

3.2. Research model and hypothesis

To test the hypothesis whether CEO pay ratios affect employee satisfaction, the following regression formula is used:

(1) 𝐸𝑚𝑝𝑙𝑆𝑎𝑡 = 𝛼 + 𝛽_01234564789𝐶𝐸𝑂𝑃𝑎𝑦𝑅𝑎𝑡𝑖𝑜 + 𝛽_3A9B𝑙𝑛𝑃𝑟𝑜𝑑 + 𝛽_62F𝑅𝑂𝐴 + 𝛽_H8IJ𝑙𝑛𝐸𝑚𝑝 + 𝜀

Where α represents a constant in all formulas, which is the value of employee satisfaction when all other variables are null. The betas represent the regression coefficients of the independent and control variables. For example, 𝛽01234564789 indicates the change in employee satisfaction when CEO pay ratio rises with one and is expected to be negative, following the alternative hypothesis.

The control variable lnProd – measuring productivity by the revenue by employee –, ROA – which is the measure for performance – and lnEmp – representing firm size – are included in the formula to control for potential omitted variable bias.

The final component of the regression equation, ε, is the error term, which contains the errors obtained because the relationship between the dependent, independent and control variables is not fully represented by the model.

Additionally, an extended formula (2) is used which includes dummy variables. Nine dummy variables for industry are included in the formula to check whether different industries have different effects on employee satisfaction.

(2) 𝐸𝑚𝑝𝑙𝑆𝑎𝑡 = 𝛼 + 𝛽_01234564789𝐶𝐸𝑂𝑃𝑎𝑦𝑅𝑎𝑡𝑖𝑜 + 𝛽_3A9B𝑙𝑛𝑃𝑟𝑜𝑑 + 𝛽_62F𝑅𝑂𝐴 + 𝛽_H8IJ𝑙𝑛𝐸𝑚𝑝 + 𝛽_M𝐼𝑛𝑑𝑀𝑎𝑛𝑢𝑓𝑎𝑐𝑡𝑜𝑟𝑖𝑛𝑔 + 𝛽_T𝐼𝑛𝑑𝑅𝑒𝑡𝑎𝑖𝑙 + 𝛽_V𝐼𝑛𝑑𝑇𝐶𝐸𝐺𝑆

+ 𝛽Y𝐼𝑛𝑑𝑊ℎ𝑜𝑙𝑒𝑠𝑎𝑙𝑒 + 𝛽]𝐼𝑛𝑑𝑆𝑒𝑟𝑣𝑖𝑐𝑒𝑠 + 𝛽_`𝐼𝑛𝑑𝑀𝑖𝑛𝑖𝑛𝑔 + 𝛽_{__}𝐼𝑛𝑑𝐶𝑜𝑛𝑠𝑡𝑟𝑢𝑐𝑡𝑖𝑜𝑛 + 𝛽_{_a}𝐼𝑛𝑑𝑁𝑜𝑛𝑐𝑙𝑎𝑠𝑠𝑖𝑓𝑖𝑎𝑏𝑙𝑒 + 𝜀

The data will be analyzed using STATA, by using the OLS regression method. The derived data will be further analyzed to be able to answer the research question: Does the CEO pay ratio affect employee satisfaction?

(12)

12 As briefly described in paragraph 2.3, it is expected that a higher CEO pay ratio will have a negative effect on employee satisfaction. The following hypotheses will be tested to check whether there is a negative effect of higher CEO pay ratios on employee satisfaction:

𝐻`: 𝛽01234564789 = 0

𝐻_{_}: 𝛽_01234564789 < 0

3.3. Measures

Dependent variable

Employee satisfaction (𝐸𝑚𝑝𝑙𝑆𝑎𝑡) is used as the dependent variable in this research. The variable can take values of 0-1. It represents the overall company rating reviewed by

employees. The reports on these ratings are submitted by the employees on a voluntary and anonymous basis. This carries a limitation. Since employees can decide for themselves whether to participate or not, it is possible that only very satisfied employees participate or employees who want to damage the company’s image (Huang et al., 2015). Furthermore, it could be possible that employers influence their employees to give positive answers on the survey’s questions or only encourage satisfied employees to give a review. Melián-González et al. (2015) faced this same limitation and conducted a frequency analysis on their sample, finding no evidence for employers manipulating the reviews.

Independent variable

The independent variable used in this paper is CEO pay ratio (𝐶𝐸𝑂𝑃𝑎𝑦𝑅𝑎𝑡𝑖𝑜). This ratio is calculated by dividing the CEO’s wage by the median employee wage (Lacmanović, 2013). This variable is expected to have an impact on employee satisfaction. Similar research found a negative relationship between salary dispersion and satisfaction (Pfeffer & Langton, 1993). As salary dispersion is less extreme than CEO pay ratio – indicating differences between all employees’ wages instead of executive compensation relative to median employee wages – it is expected to yield similar results, because differences are larger with CEO pay ratios.

𝛽_01234564789 represents the regression coefficient and is expected to be negative, following

(13)

13 Control variables

In order to make the regression more accurate, several control variables are included. These variables are included in the formula to control for potential omitted variable bias.

The first control variable that is included is productivity (𝑙𝑛𝑃𝑟𝑜𝑑). Prior research found both positive and negative relations between CEO pay ratio and productivity, and positive relationships between employee satisfaction and productivity. Because these

relations occurred, this variable is needed to be included to control for omitted variable bias. Productivity is measured by the natural log of the revenue per employee, following Faleye et al. (2013). It is calculated by taking the natural log after dividing the revenue by the number of employees.

Performance is the second control variable. The reason for including this variable follows the same reasoning as stated above and it is measured by the return on assets (𝑅𝑂𝐴), following prior research (Faleye et al., 2013; Melián-González et al., 2015; Huang et al., 2015).

The third variable included is firm size (𝑙𝑛𝐸𝑚𝑝). Since Faleye et al. (2013) argued that firms with fewer employees are more productive, and productivity has a positive effect on employee satisfaction, there should be a relation between firm size and employee

satisfaction as well. Furthermore, they found that CEO pay ratios increase with firm size. These interrelationships give enough reason to include firm size as control variable. It is measured as the natural log of the number of employees, following Cowherd and Levine (1992) and Artz (2008).

Dummy variables

Faleye et al.’s (2013) finding that relative pay is higher in homogeneous industries is the reason to include industry as dummy variable. The industries are divided into ten categories following the Standard Industry Classification codes, which are listed in Appendix 1. Nine industries are included in the formula: 𝐼𝑛𝑑𝑀𝑎𝑛𝑢𝑓𝑎𝑐𝑡𝑜𝑟𝑖𝑛𝑔, 𝐼𝑛𝑑𝑅𝑒𝑡𝑎𝑖𝑙, 𝐼𝑛𝑑𝑇𝐶𝐸𝐺𝑆,

𝐼𝑛𝑑𝑊ℎ𝑜𝑙𝑒𝑠𝑎𝑙𝑒, 𝐼𝑛𝑑𝑆𝑒𝑟𝑣𝑖𝑐𝑒𝑠, 𝐼𝑛𝑑𝑀𝑖𝑛𝑖𝑛𝑔, 𝐼𝑛𝑑𝐶𝑜𝑛𝑠𝑡𝑟𝑢𝑐𝑡𝑖𝑜𝑛 𝑎𝑛𝑑 𝐼𝑛𝑑𝑁𝑜𝑛𝑐𝑙𝑎𝑠𝑠𝑖𝑓𝑖𝑎𝑏𝑙𝑒. By omitting 𝐼𝑛𝑑𝐹𝑖𝑛𝑎𝑛𝑐𝑒, the effect that an industry has on the dependent variable can be measured, compared to the omitted one. Since people working in the finance industry are assumed to be the unhappiest (Dishman, 2015), this industry is omitted in the formula.

(14)

14

4. Results

In this paragraph, descriptive statistics are presented. Furthermore, the results of the

performed tests to examine whether CEO pay ratio has an effect on employee satisfaction or not are presented.

4.1. Descriptive statistics and correlations

The standard deviation of employee satisfaction is quite small – 0.0718 – with a mean of 0.6786, meaning that the ratings do not differ much. Appendix 2 shows a skewness of -0.0537, and a kurtosis of 2.9514. Normally distributed data has a skewness of 0 and a kurtosis of 3. The values for employee satisfaction are slightly left from, but approximately equal to these numbers, so it can be assumed that the Likert data on employee satisfaction is normally distributed. Therefore, the OLS regression method can be used. The standard deviations for productivity and firm size are small, which are measured by lnProd and lnEmp respectively. The data for CEO pay ratio are more spread out, with a mean of 204.45 and a standard deviation of 187.4974. Appendix 2 shows that the data on CEO pay ratio is heavily skewed and tailed, so the distribution of this variable varies widely. The minimum and maximum values of this variable support this, with a minimum value of 0 and a maximum of 1951. The minimum value of 0 is explained by the fact that the CEO of Fossil does not pay himself a salary. For the return on assets, the mean is 0.0675 and the standard deviation 0.0594, with minimum and maximums which are far away from these numbers (respectively -0.3163 and 0.3491) and is heavily tailed.

Table 1 – Descriptive Statistics

This table shows descriptive statistics for key variables. Data is obtained from Compustat and Glassdoor.com

Variable Obs Mean Std. Dev. Min Max

EmlpSat 429 0.6786512 0.0718003 0.48 0.9

CEOPayRatio 429 204.4544 187.4974 0 1951

lnProd 429 13.06192 0.8893324 9.945911 16.38045

ROA 429 0.0674897 0.0594467 -0.3163411 0.3490942

lnEmp 429 10.06343 1.331637 4.574711 14.60397

The correlation matrix is presented in Table 2. The correlation between employee

satisfaction and CEO pay ratio is -0.0382. Since correlations between the range of -0.1 and +0.1 are said to have no or a very weak linear relationship and this correlation of -0.0328

(15)

15 falls within this range, it can be concluded that, when analyzing the correlation, there is no or a very weak relation between employee satisfaction and CEO pay ratios.

When comparing the results of the correlation matrix to the prior research analyzed in paragraph 2, several conclusions can be drawn. The finding of Pfeffer and Langton (1993) that salary dispersion and productivity are negatively linked, is supported by the negative correlation coefficient of -0.1178. Faleye et al.’s (2013) finding of the positive effect of CEO pay ratios on productivity when a firm has fewer employees can be supported by the correlation between productivity and firm size, which is -0.5044. This indicates that when the number of employees decreases, productivity increases.

Table 2– Correlation matrix

EmplSat CEOPayratio lnProd ROA lnEmp

EmplSat 1.0000 CEOPayRatio -0.0382 1.0000 lnProd 0.1854 -0.1178 1.0000 ROA 0.0472 0.0641 -0.1730 1.0000 lnEmp -0.0537 0.3933 -0.5044 0.0897 1.0000 4.2 Regression results

Table 3 shows the results for the regressions. Results are reported for the original regression formula (1) in the first column. The second column presents the results for the modified regression formula (2), which includes the industry dummies. The regression results in Table 3 are Stata OLS results, which are two-sided results. Since the hypothesis suggests a negative relation, the result on CEO pay ratio has to be one-sided. Therefore, an extra test is done to get this one-sided result on CEO pay ratio, by dividing the p-value by two. All other results are two-sided.

Regression 1

The result for CEO pay ratio (CEOPayRatio) is not significant (p=0.193, one-sided) and approximately equal to zero. Thus, there is not enough evidence that CEO pay ratio has a negative effect on employee satisfaction. In addition, the 𝑅a_{value, which shows to what} extent the model explains the variation in employee satisfaction, is 0.0447. So 4.47% of the variation in employee satisfaction is explained by the model. This can be due to variables which are not included in formula (1), resulting in potential omitted variable bias.

(16)

16 Table 3 – Regression results

This table provides the regression results of CEO pay ratio on employee satisfaction for the three different regression formulas.

Dependent variable EmplSat

(1) (2)

CEOPayRatio -0.0000173

(1.99e-05) (2.03e-05) -3.85e-06

lnProd 0.0187166*** (0.0045) 0.0238571*** (0.0048) ROA 0.100975* (0.0583) 0.1059377* (0.0606) lnEmp 0.0039597 (0.0032) 0.0062272* (0.0032) IndFinance IndManufacturing 0.0091594 (0.0101) IndRetail -0.0215835 (0.0159) IndTCEGS 0.0201056* (0.0122) IndWholesale -0.0571769** (0.0224) IndServices 0.0455322*** (0.0134) IndMining -0.0083035 (0.0176) IndConstruction -0.0301219 (0.0351) IndNonclassifiable 0.0220321 (0.0496) IndAgriculture 0.0767999 (0.06912) Constant 0.3910818*** (0.0800) 0.2901931*** (0.0827) Observations 429 429 𝑹𝟐 _0.0447 _0.1207

Column (1) shows the results for the original regression formula with the key variables only. In column (2) the industry dummies are included.

Stata regression results are two-sided. To get the one-sided result for CEO pay ratios, the p-values are divided by 2.

*** ** *

Denotes significance at the 1% level. Denotes significance at the 5% level. Denotes significance at the 10% level.

(17)

17 Furthermore, the coefficient of productivity (lnProd) indicates a significant positive relation (p=0.000). This result is in line with Melián-González et al.’s (2015) finding that there is a positive relationship between employee satisfaction and productivity. The same holds for performance (ROA), which is positive (p=0.084), but the significance is low (10%). Research of Melián-González et al. (2015) and Huang et al. (2015) – both finding a positive relation between performance and employee satisfaction – support this result.

Regression 2

The results of the extended model with the industry dummies included are shown in column (2). When adding industry dummies, the results on the key variables change a bit. CEO pay ratio is still insignificant (p=0.425, one-sided). However, productivity (p=0.000) and firm performance (p=0.081) are significantly positive. Additionally, firm size is significant at the 10% level (p=0.054). When analyzing the dummy variables, it becomes clear that different industries have different effects on employee satisfaction. The wholesale trade industry has a significant negative relation with employee satisfaction, compared to the finance industry (p=0.011). There are positive relationships found for the TCEGS (p=0.100) and services (p=0.001) industries, compared to the finance industry. This indicates that employees in the TCEGS and services industries are more satisfied than people working in the finance

industry, which is not surprising since the finance industry has the unhappiest employees. The results indicate that people working in the wholesale trade industry are less satisfied than those working in the finance industry, which is a surprising result. This can be due to large differences in wages along the production chain. The regression results in a higher 𝑅a value of 0.1207. So by adding the dummy variables, the omitted variable bias decreases and more of the variation in employee satisfaction is explained by the model.

5. Conclusion

The increasingly high levels of CEO pay ratios is highlighted by newspapers (Gelles, 2018; Anderson & Pizzigati, 2018), since this ratio keeps on rising. This paper is focused on whether these high ratios have an effect on employee satisfaction. To answer this question, 429 S&P500 companies are analyzed using the OLS regression method. Company ratings of 2015 are used to measure employee satisfaction and the data on CEO pay ratios, productivity,

(18)

18 firm performance and size date from 2014. By using these two consecutive years, reversed causality is eliminated (Edmans, 2012).

When analyzing and weighing out prior literature, there is a negative effect expected of CEO pay ratios on employe satisfaction. Therefore, the hypothesis that is tested is that there is a negative effect of CEO pay ratios on employee satisfaction.

The regression results do not show a significant relation between CEO pay ratios and employee satisfaction in any case. Therefore, the null hypothesis can not be rejected and no relation is found between employee satisfaction and CEO pay ratios. A reason for not finding a relation could be that employees are not aware of their CEO’s wages. Considering the new rules for mandatory disclosure of CEO compensation, median employee compensation and CEO pay ratios (SEC, 2015), it could be assumed that there is a lack of transparency. This lack of transparency can be a cause of the absence of a relation between CEO pay ratios and employee satisfaction. Therefore, further research can be done on this topic in some years, to see whether this mandatory disclosure – which is compulsory as of early 2018 – has an impact on employee satisfaction. Pfeffer and Langton (1993) did find a negative relation between salary dispersion and satisfaction in a research with somewhat similar variables. However, they conducted their research at academic departments of colleges and

universities, which are smaller and colleagues are more likely to know each other than all employees and CEO’s at large firms. Besides that, they used salary dispersion instead of CEO pay ratios, which takes on less extreme values than CEO pay ratios, since it indicates the difference in all wages members of a department get instead of only the department head and median employee. Furthermore, the regression results support the findings of Melián-González et al. (2015) and Huang et al. (2015), where there are positive relations found between employee satisfaction and productivity (p<0.01), firm performance (p<0.1) and firm size (p<0.1).

A potential risk for making the model less reliable is the modification of the company ratings from a 5-point Likert scale to a percentage scale with values from 0-1 and the fact that employees voluntarily participate in Glassdoor’s research on employee satisfaction. Additionally, the R-squared values are low in all three variants, which indicates that there is potential omitted variable bias. Therefore, adding more variables could be an improvement of this research. Potential variables that could be added for future research are: age; the CEO’s age; position within the firm; number of years employed and the amount of contact

(19)

19 with the CEO – i.e. corporate culture. Furthermore, another improvement could be that other measures for the key variables could be used. For example, size could be expressed by the total market capitalization and performance by the stock returns. Additionally, an

interaction variable could be added by multiplying CEO pay ratios by firm performance, since high CEO pay combined with low firm performance could influence employee satisfaction (Belger, 2018). Another suggestion for future research would be to match Glassdoor respondents’ salaries to their reviews and ratings given. By doing so, it could be examined whether people who earn lower salaries within a company and are further away from the CEO’s compensation – salary wise –, face lower levels of satisfaction or not, which is proved to have different effects on satisfaction (Pfeffer & Langton, 1993).

Transparency could make either CEO’s and employees aware of the wages they get paid, which can have positive and negative effects. The positive effect can occur when the transparency leads to a deceleration of the rise in CEO pay ratios. Transparency makes, on the other hand, employees aware of their own wage, their CEO’s wage and their wage compared to their colleagues. The lowest paid employees – i.e. those earning lower than median wage – could become encouraged, leading to lower levels of satisfaction and lower productivity, which in turn reduces firm value and firm performance (Akerlof & Yellen, 1990; Melián-González et al., 2015; Huang et al., 2015). Another negative effect could be that when CEOs know eachothers wages, chances are that they want to raise their wage to a amount higher than their competitor’s wage.

Regulation on mandatory disclosure is introduced in the UK (Belger, 2018) and the US (SEC, 2015). In The Netherlands, this is not yet the case. However, the problem with

increasingly high ratios is present (Kakebeeke & Couwenbergh, 2018). When deciding

whether The Netherlands should implement rules on mandatory disclosure as well, it may be wise to see how the new rules influence the pay ratios and performance in the UK and the US in years subsequent of the introduction of the new rules and wait for the results of further research on this, before implementing such rules in The Netherlands.

(20)

20

References

Aguilar, L.A. (2015). The CEO Pay Ratio Rule: A Workable Solution For Both Issuers and Investors. Retrieved from https://www.sec.gov/news/statement/statement-on- open-meeting-on-pay-ratio-aguilar.html

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

Anderson, S., & Pizzigati, S. (2018). No CEO should earn 1,000 times more than a regular employee. The Guardian. Retrieved from https://www.theguardian.com/business/ 2018/mar/18/america-ceo-worker-pay-gap-new-data-what-can-we-do

Artz, B. (2008). The Role of Firm Size and Performance Pay in Determining Employee Job Satisfaction Brief: Firm Size, Performance Pay, and Job Satisfaction. LABOUR, 22(2), 315-343.

Belger, T. (2018). UK to force companies to justify pay gap between CEOs and staff. The Financial Times. Retrieved from https://www.ft.com/content/db9d0b4c-6b25-11e8-8cf3-0c230fa67aec

Bloom, M. (1999). The performance effects of pay dispersion on individuals and organizations. Academy of Management Journal, 42(1), 25-40.

Chamberlain, A. (2015) CEO to Worker Pay Ratios: Average CEO Earns 204 Times Median Worker Pay. Retrieved from https://www.glassdoor.com/research/ceo-pay-ratio/ - _ftn2

Cowherd, Douglas M., & Levine, David I. (1992). Product quality and pay equity between lower-level employees and top management: An investigation of distributive justice theory. (Process and Outcome: Perspectives on the Distribution of Rewards in Organizations). Administrative Science Quarterly, 37(2), 302.

Dishman, L. (2015). Why People In Finance And Insurance Are The Unhappiest Employees. Fast Company. Retrieved from https://www.fastcompany.com/3046257/why-finance-and-insurance-workers-among-the-unhappiest-employees

Edmans, A. (2012). The Link Between Job Satisfaction and Firm Value, With Implications for Corporate Social Responsibility. Academy of Management Perspectives, 26(4), 1-19. Faleye, O., Reis, E., & Venkateswaran, A. (2013). The determinants and effects of CEO– employee pay ratios. Journal of Banking and Finance, 37(8), 3258-3272.

Gelles, D. (2018, May 25). Want to Make Money Like a C.E.O.? Work for 275 Years. The New York Times. Retrieved from https://www.nytimes.com/2018/05/25/business/highest-paid-ceos-2017.html

(21)

21 Huang, M., Li, P., Meschke, F., & Guthrie, J.P. (2015). Family firms, employee satisfaction, and corporate performance. Journal of Corporate Finance, 34, 108-127.

Jamieson, S. (2004). Likert scales: How to (ab)use them. Medical Education, 38(12), 1217- 1218.

Kakebeeke, P., & Couwenbergh, P. (2018). Ceo Paul Polman verdient 292 keer zo veel als gemiddelde Unilever-medewerker. Het Financieel Dagblad. Retrievew from

https://fd.nl/ondernemen/1246594/ceo-paul-polman-verdient-292-keer-zo-veel-als-gemiddelde-unilever-medewerker

Knapp, R., T. (1990) Treating Ordinal Scales as Interval Scales: An Attempt To Resolve the Controversy. Nursing Research, 39(2), 121-123.

Lacmanović, S. (2013). The Relevance and Effects of the Executive-to-worker Pay Ratio’ Disclosure. Economic Research-Ekonomska Istraživanja, 26, 165-184.

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

Locke, E.A. (1976), “The nature and consequences of job satisfaction”, in Dunnette, M.D. (Ed.), Handbook of Industrial and Organizational Psychology, Rand McNally, Chicago, IL, pp. 1297-1349.

Mangiafico, S., S. (2016). Summary and Analysis of Extension Program Evaluation in R, version 1.13.5. Retrieved from http://rcompanion.org/documents/RHandbook ProgramEvaluation.pdf

Melián-González, S., Bulchand-Gidumal, J., & González López-Valcárcel, B. (2015). New evidence of the relationship between employee satisfaction and firm economic performance. Personnel Review, 44(6), 906-929.

Norman, G. (2010). Likert Scales, Levels of Measurement and the "Laws" of Statistics. Advances in Health Sciences Education, 15(5), 625-632. Pfeffer, J., & Langton, N. (1993). The effect of wage dispersion on satisfaction,

productivity, and working collaboratively: Evidence from college and university faculty. (includes appendix). Administrative Science Quarterly, 38(3), 382.

U.S. Securities and Exchange Commission. (2015). SEC Adopts Rule for Pay Ratio Disclosure [Press release]. Retrieved from https://www.sec.gov/news/pressrelease/2015-160.html

U.S. Securities and Exchange Commission. (2017). SEC Adopts Interpretive Guidance on Pay Ratio Rule [Press release]. Retrieved from https://www.sec.gov/news/press-release/2017-172

(22)

22 United States Department of Labor. (n.d.) SIC Division Structure. Retrieved from

(23)

23

Appendix

Appendix 1

Appendix 1 – Dummy variables

Dummy Industry category IndManufacturing Manufacturing IndRetail Retail Trade

IndTCEGS Transportation, Communications, Electric, Gas and Sanitary Service IndWholesale Wholesale Trade

IndServices Services

IndMining Mining

IndConstruction Construction

IndFinance Finance, Insurance and Real Estate IndAgriculture Agriculture, Forestry and Fishing IndNonclassifiable Nonclassifiable

Appendix 2

Appendix 2 – Detailed descriptive statistics

Variable 𝑸_𝟏 Median 𝑸_𝟑 Variance Skewness Kurtosis

EmplSat 0.64 0.68 0.72 0.0051553 -0.0536763 2.951383

CEOPayRatio 102 157 239 35155.27 3.995868 28.65003

lnProd 12.51055 12.95334 13.61134 0.790912 0.2915736 4.340145 ROA 0.0263362 0.0591467 0.0981679 0.0035339 0.2426408 8.441251 lnEmp 9.169518 9.998798 11.0021 1.773257 -0.0743406 3.343861

The effect of the CEO pay ratio on employee satisfaction