Compensation Consultants and Executive Pay: An
Analysis of Pay Changes around Consultant
Appointments
University of Amsterdam
Amsterdam Business School MSc Business Economics Finance track
Master Thesis
Student Name: Yuling Hu Student Number: 11086211 Thesis supervisor: Dr. Torsten Jochem
June 2016
Statement of originality
This document is written by Student Yuling Hu who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
This thesis estimated the relationship between change in having a compensation consultant and change in top five executives’ compensation; the association between change in number of compensation consultants and change in top five executives’ compensation; and whether CEO power influence firms’ decisions on having a compensation consultant and firms’ change behavior in number of compensation consultants. The answer of this will help firm’s committee board to avoid management rent extraction and solve agency problem. To conduct this analysis, I run four panel regressions. The results show having more consultants reduces the executives’ pay. So I concluded compensation consultants appear an effective vehicle to prevent managerial rent-‐extraction. This finding is consistent with the rent extraction theory.
Content
I. Introduction ... 5-‐12 II. Literature review ...12-‐20 III. Methodology ...20-‐30
i. Hypotheses ...20-‐22 ii. Sample Construction ...23-‐24 iii. Variable Measurement ...24-‐29 iv. Regression Equation...29-‐30 IV. Data and descriptive statistics...30-‐34 V. Results...34-‐39 VI. Conclusion...39-‐41 References... 42-‐43 Table... 44-‐55
I. Introduction
Why focus on executive’s compensation? A group of brilliant executives can benefit the company in many ways. These executives enjoy high reputations in the related field and they have some precious characteristics. They are very experienced and are famous for leadership. They have broader views or more creative minds. Therefore, companies want to hire them because they can solve problems in a more efficient way and they are more likely to make the company to achieve great success. And the compensation is a good tool to attract these executives. The compensation has several components such as the fixed salary, the performance-‐based bonus and the time-‐vesting stock options. Generally, based on various components of compensation, it has three main goals, the first one is to attract the right executives at lowest costs, the second goal is to motivate executives to take actions that increase long-‐term shareholder value and the third one is to retain the right executive at lowest costs. From the other perspective, the executive compensation is a significant issue in corporate governance because it related to the first agency problem. As we know, there are two agency problems in corporate governance. The first one is the misalignment of interest between executives and shareholders. When discussed about the compensation and agency problem, there are two traditional theories. One is the optimal contracting view. This view supports the opinion that executives’ compensation can solve agency problems and help to align the interest between executives and shareholders in a more efficient way, Hall and Murphy
(2003). The other one is the managerial power view (the rent extraction view), this view argued that executive compensation was a part of agency problem and could not solve the agency problem because powerful CEOs affect the process of arranging manager’s compensation and are able to extract “excess” compensation that are larger than economic characteristics justify, Bebchuk and Fried (2003, 2004).
As described above, the set of executive compensation is quite important for a company, but how to set a reasonable pay packages? To do this, the compensation or nominating committee of the board often seeks helps from external compensation consultant to advice. Compensation consultants can use their professional knowledge to design a suitable pay packages. They help to find candidates. In addition, they contribute to reach the attracting and retaining goals by creating a better-‐structured compensation plan. What’s more, they are beneficial to the motivation goal. The compensation will provide incentives to executives to create long-‐term value. Furthermore, they also help to navigate the legal and stock market environment.
However, many executive compensation consultants provide non-‐executive compensation consulting services to the firm as well. For example, they also provide advice on pension plans. So critics argued that these cross-‐selling interests induced compensation consultants to give upward biased advice in executives’ compensations.
consultant affected the executive’ pay. For example, Murphy and Sandino (2010) found that CEO pay was higher in companies that have multi-‐service consultants. Conyon, Peck and Sadler (2009) pointed out that CEOs had higher pay and the amount of equity used in CEO pay package was higher in firms with compensation consultants. Goh and Gupta (2010) also supported that all components of executive compensation were higher in firms with compensation consultants. Voulgaris, Stathopoulos and Walker (2009) showed that the use of a compensation consultant had an increasing effect on the level of total CEO compensation. And Armstrong, Ittner and Larcker (2012) pointed out that the differences in CEOs’ pay between firms with consultant and firms without consultant were not driven by the use of consultant firms. Instead, they claimed that CEOs’ higher pay driven by weaker corporate governance. Following Armstrong, Ittner and Larcker (2012), Chu, Faasse and Rau (2015) used corporate governance control and still found that the pay of these non-‐executive compensation services was often multiples of the pay of compensation consulting services. And CEO and executives can decide or highly influence whether to use these non-‐executive compensation services. Therefore, I think the investigation of the relationship between compensation consultant and executives’ pay is an important issue. If the hiring of executive compensation really enables the executive’s pay higher, then the company should consider how to avoid the misalignment of compensation.
change in having a compensation consultant and the change in number of compensation consultants during years as my key independent variables. Firms may change their compensation consultant for various reasons such as for cost saving or for more comprehensive compensation suggestions from various consultants. They may also remain at the same number. My study result of this will provide evidence to support that change in having a compensation consultant and change in number of compensation consultants in a firm from t-‐1 to t has an influence on CEO and top five executives’ compensation level.
Compared with previous literature, instead of CEO pay, I used the top five executives’ pay to conduct my research. This helped to have a broader result and not just limited on CEO. I included the pay of top five executives also because these executives have a significant impact on corporate value and often viewed and assessed as a team, Fee and Hadlock (2004). In addition, these executives especially top five executives have an important role in firms’ management. What’s more, compensation consultants not only provide suggestion on CEO pay packages, they also give advice on pay of all executives. So my results of this study will contribute to new evidence about the relationship between the number of compensation consultants and the top five executives pay packages. Furthermore, I also compared the differences between the changes effects of compensation consultant number on change in CEO pay and change in other top five executives’ pay except CEO.
ownership stake to measured CEO power and found no evidence to support that CEO power will increase the firm hiring a pay consultant. I followed Bebchuk, Cremers and Peyer (2011) and used the CEO pay slice to measure CEO power to see whether the CEO power make the agency issues stronger. My study of this part makes contribution to providing new evidence on whether CEO power affects firm’s decision on having a compensation consultants and firm’s change behavior in number of compensation consultants.
To conduct this thesis, I used the executive compensation data and compensation consultant data from 2006 to 2013. Furthermore, as previous studies showed that the level of executive compensation was correlated with firm size, operating and stock price performance and investment opportunities (Lambert and Larcker 1987; Smith and Watts 1992; Core and Guary 1999), I included data on firm information, stock price to construct my control variables on firm characteristics. Coles, Daniel and Naveen (2008) suggested that board size had a negative association with managerial and board decision making. Core, Holthausen and Larcker (1999) found that there was association between excess CEO compensation and several board characteristics. Therefore, I included control variables related to corporate governance used data of ISS/Riskmetrics.
My primary sample consists of 3701 US firms that have compensation consultant. I ran four regressions with industry-‐fixed effect and firm-‐fixed effect. Firstly, I estimated the association between change in having a compensation
consultant and change in top five executives’ pay. Then I investigated the relationship between change in number of compensation consultant and change in top five executives’ pay. I also investigated how the change in having a compensation consultant and the change in number of compensation consultants influenced the change in CEO pay and the change in other top five executives’ pay except CEO. Whether having a compensation consultant is denoted by a dummy variable D_CONSULTANT that equal to 1 if the firm has a compensation consultant and 0 if not. D_ΔCONSULTANTit is defined as the change in having a compensation consultant from t-‐1 to t. If the company without hiring a compensation consultant changes to have one, then D_ΔCONSULTANTit will be 1. If the company keeps the same as before in hiring a compensation consultant, then this variable will be 0. If the company with hiring a compensation consultant changes to without one, then this variable will be -‐1. The change in number of compensation consultant was calculated by the difference between consultant number in t-‐1 and t. Similar to most executive compensation research, I used the natural logarithm of executive compensation in my study because of highly (right) skewed distribution of pay. Secondly, I CEO pay slice to measure the effect of CEO power on the decision of having a compensation consultant and change in number of consultants. CEO pay slice is the percentage of CEO compensation in top five executives’ total compensation.
support there is a significant association between change in having a compensation consultant and change in top five executives’ pay. Secondly, I surprised to find that the relationship between change in number of compensation consultant and change in top five executives’ compensation is negative. I attribute this finding to a support of the rent extraction theory as a larger change in number of compensation consultant prevent top management from using their power to influence multiple consultants. The negative relationship is also existed between change in number of compensation consultant and change in CEO compensation. These results are different from what Goh and Gupta (2010) found. They did not found any evidence to support there is a positive relationship between changes in the number of compensation consultant and changes in compensation. So they considered multiple consultants was not necessarily be a vehicle for rent extraction. My results also challenged the finding of Kabir and Minhat (2008). They found that an increase in the number of compensation consultants would lead to an increase in CEO equity-‐based pay. But this positive relationship only exist when the increase happen. However, both of our results support the rent extraction theory. In addition, my results reject the hypothesis that higher CEO power has a negative influence on having a compensation consultant and change in number of compensation consultants. I did not found any evidence can support these expectations. This is consistent with Voulgaris, Stathopoulos and Walker (2009) that showed that different proxies of CEO power were insignificantly or even negatively related to the
probability of hiring a compensation consultant.
This thesis is organized as follows. Section Ⅱ reviews prior literatures. Section Ⅲ explain the methodology of this thesis. Section Ⅳ discuss the data and descriptive statistics. Section Ⅴ provides results. Section VI provides conclusion.
II. Literature review
In terms of idea about executive pay, there are two traditional views. One is the optimal contracting view. This view held that shareholders set executive pay to limit agency problems and executive compensation provided a cost-‐effective way to align incentives and to maximize firm value, Hall and Murphy (2003). It also pointed out that executive pay package had the function of attraction, retention and incentive. So it attributed the differences in pay levels of firms to various economic characteristics. Some empirical studies results are consistent with this view. For example, Murphy (1999) found that executive compensation practices vary with company size, industry and country. His analysis showed that pay levels are higher and pay-‐performance sensitivities are lower in larger firms. And both pay levels and pay-‐performance sensitivities are higher in regulated utilities than in industrial firms. In addition, it also provided result that levels of pay and pay-‐performance sensitivities were higher in the US than in other countries. The other one is the managerial power view (the rent extraction view), this view argued that CEOs have a great power or control over the board of directors and CEOs can use this power to extract “excess” pay levels that are larger than economic characteristics justify. So
this view thought that executive compensation was a part of agency problem, it cannot solve the agency problem. Bebchuk and Fried (2003, 2004) indicated that managerial power plays a significant role in arranging manager’s pay packages. Kabir and Minhat (2008) examined the UK companies from 2003 to 2006 and reached the conclusion that CEO’s equity-‐based compensation increases as the number of compensation consultants increases.
Previous studies have investigated what would influence the executive compensation. For example, Lambert and Larcker (1987) conduct an analysis about the use of accounting and market measures of performance in executive compensation contracts. They found that firms would put more weight on market performance in compensation contracts for three situations. Firstly, when the variance of the accounting measure of performance is higher than the variance of the market measure of performance. Secondly, when the firm is facing high growth rates in assets and sales. Lastly, when the value of the manager’s personal holdings of his firm’s stock is low. Smith and Watts (1992) indicated that measures of the firm’s investment opportunity set like availability of growth options and firm size are related to executive compensation policies. They documented that if firms were with more growth options, the executive compensation would be higher. They also pointed regulated firms would have lower executive compensation and less frequent use of both stock-‐option plans and bonus plans. Furthermore, their results showed there was a positive relationship between firm size and level of executive
compensation. Core and Guay (1999) suggested that the optimal portfolio of incentives from stock and options varies with hypothesized economic determinants like firm size, growth opportunities and proxies for monitoring costs. Core, Holthausen and Larcker (1999) provided evidence that CEOs in firms with weaker governance can receive higher compensation levels as weaker governance lead to greater agency problems. Their results showed that board and ownership structure has a significant relationship with variation in CEO compensation level. Coles, Daniel and Naveen (2008) evaluated the association between firm value and board structure. Although they did not research on the compensation level directly, their results made great contribution when discussing the correlation between board characteristics and compensation level. They results supported the opinion that board size has a negative effect on managerial and board decision-‐making.
In terms of idea about using compensation consultants, a number of previous literatures have investigated the relationship between compensation consultant and CEO’s pay. For example, Waxman (2007) argued that corporate consultants could have a financial conflict of interest if they provide both executive compensation advice and other services to the same company. There is a cross-‐selling incentive because whether to use other consultant services is decided or highly influenced by CEO and executives. Murphy and Sandino (2010) also found that both in the US and Canada, CEO pay is higher in firms with consultants provide multi services and pay is higher in Canadian firms when the fees paid to consultants for other services are
large relative to the fees for executive compensation service. They also tried to examine differences between consultants retained by the board and those retained by the management and their results showed that pay is higher in US firms where the consultant hired by the board rather than hired by management. For consultants’ cross-‐selling incentives, Cadman, Carter and Hillegeist (2010) make a distinction between consultants that provide compensation services alone and consultants that provide multi services, based on this, they found that no evidence indicating pay levels are higher or pay-‐performance sensitivities are lower in firms with consultants provide multi services. Consistent with the rent extraction theory of executive pay, Conyon, Peck and Sadler (2009) estimated based on UK data and found that CEOs had higher pay and the amount of equity used in CEO pay package was higher in firms with compensation consultants. Their study yielded not evidence that using consultants with potential conflicts of interest leads to higher CEO pay and higher equity proportion in pay packages. Goh and Gupta (2010) also found that all components of executive compensation are higher in firms with compensation consultants. In addition, based on their UK data, they provided new evidence that CEOs and executives of firms that switch their main consultant receive higher salary increments in the year of the switch and a less risky compensation package. And also based on UK data, Voulgaris, Stathopoulos and Walker (2009) showed that the use of a compensation consultant has an increasing effect on the level of total CEO compensation.
However, Armstrong, Ittner and Larcker (2012) argued that the ultimate responsibility for setting and approving pay levels rested with the board of directors rather than consultants. Their results pointed out that the differences in CEOs’ pay between firms with consultant and firms without consultant were not driven by the use of consultant firms. Instead, they claimed that CEOs’ higher pay driven by weaker corporate governance. This is because firms with weaker corporate governance are more likely to hire compensation consultants. They used propensity-‐scoring methods to match firms on both economic and governance characteristics to find that no significant pay differences in consultant users and nonusers. But their results do not vary between “specialized” and “nonspecialized” consultant, providing no support for claims that consultants who offer multi-‐services are more likely to facilitate excess pay levels because of greater conflicts of interest.
Chu, Faasse and Rau (2015) followed Armstrong, Ittner and Larcker on the use of corporate governance control and provided empirical evidence for the hiring of compensation consultants as a justification device for higher CEO pay. Based on their longitudinal data from 2006 to 2012, they firstly found that CEOs at firms with consultants earn significantly higher pay levels than their peers at firm-‐, CEO-‐ and governance-‐matched firms. Another important finding in this paper is related to regulation change for firms in hiring a consultant. It is SEC additional disclosure rules that require firms to disclose the fee paid if consultants provide multi services to the firm. Its objective is to ensure the independence of the compensation consultants, in
other words, the essence of the SEC regulations support that multi-‐service consultants are conflicted. This change leads to increase on turnover of compensation consultants and major multi-‐service consultants’ spin-‐off. Based on these changes, Chu, Faasse and Rau (2015) found that CEOs of firms switched to newly spun-‐off consultants were significantly paid more in median total and non-‐incentive compensation than those of matched sample firms remained with multi-‐service consultants. They also pointed out that firms pay their CEOs more upon compensation consultant adoption and firms where CEOs enjoy a greater increase in pay are less likely to turn over consultants the following year.
With the exception of Guh and Gupta (2010) who include all executive directors, all these previous literatures conduct their research between compensation consultants and CEO pay, so we know little about the pay packages of other executives except CEO. In my study, I choose the top five executives ranked at the first 5 by sum of salary and bonus. There are two reasons for why I want to investigate top five executives pay packages. Firstly, these executives have a significant impact on corporate value and often viewed and assessed as a team, Fee and Hadlock (2004). And these executives especially top five executives have an important role in firms’ management. Secondly, compensation consultants not only provide suggestion on CEO pay packages, they also give advice on pay of all executives. So my results of this study will contribute to new evidence about the relationship between change in compensation consultant number and the top five
executives pay packages.
Based on the previous researches, they discussed a lot about the using of compensation consultant and CEO’s compensation. However, although some of them estimated association between the switch of compensation consultant and the compensation level of CEO, they seldom investigate the change in having a compensation consultant and the change in compensation consultant number between years in the same firm and what effect this change would post on CEO compensation and on top five executives’ compensation. Gua and Gupta (2010) had investigated the change in number of compensation consultants, but they only estimated the situation of increasing consultants’ number. My study was focus on this change of compensation consultant number both increase and decrease. Kabir and Minhat (2008) involved in estimations with number of compensation consultants hired by a firm. They separated their sample into three categories of firms representing increase, decrease and remain in the number of consultants used from one year to next year. But my study did not use the separated categories, my key independent variable is the change in number of compensation consultants from one year to next year. My results would provide new evidence to support the idea that change in having a compensation consultant and change in compensation consultant number affects the firms’ compensation level of CEO and top five executives.
power to increase the likelihood of the firm hiring a compensation consultant. Voulgaris, Stathopoulos and Walker (2009) also had proposed this expectation and they used a proxy for controlling CEO power. They chose CEO ownership stake because Bebchuk, Fried and Walker (2002) predicted that the higher the CEO’s shareholdings the higher their power. However, their results showed no evidence to support that CEO power will increase the firm hiring a pay consultant. I have a doubt on their use of proxy for CEO power and their results, so I try to use another way to include CEO power in my study and test whether the CEO power influence decision on having a compensation consultant and change in number of compensation consultants. Bebchuk, Cremers and Peyer (2011) found that the CEO pay slice could reflect the relative importance of the CEO as well as the extent to which the CEO is able to extracts rents. In other words, to the extent that the CEO has power and influence over the company’s decision making, the CEO might use this power and influence to raise CEO pay slice above its optimal level. In this case, the excess of the actual CEO pay slice over the optimal CEO pay slice reflects rents captured by the CEO and can be viewed as a product of agency problems. So I used CEO pay slice to measure the CEO power and CEO pay slice is defined as the percentage of the total compensation to the top five executives that goes to the CEO. The previous literatures only expected that CEO power would influence CEO pay. However, because top five executives are always considered as a group in a company, I think CEO power also has an effect on other top five executives’ compensation. Therefore,
I used CEO power to test whether it affects change in having a compensation consultant and change in number of compensation consultant. In case of this, my study makes contribution to providing new evidence on whether powerful CEOs avoid hiring a compensation consultant or change to hiring more compensation consultants.
III. Methodology
i.
Hypotheses
Based on the previous literatures and theories, there are three main hypotheses in my study.
Previous researches have discussed a lot on the relationship between the using of compensation consultants and CEO compensation, but their results were inconsistent. Some of their results supported the optimal contracting view and considered compensation was a good way to limit agency problems. Some are supporters of the rent extraction theory and their results showed that powerful CEO was able to extract excess compensation. These two theories stimulated my interest to conduct a research about executive compensation. What is more, as discussed in literature review, a great amount of prior literature suggested that the use of compensation enabled the higher CEO pay. But most of their studies only investigate the association between hiring a compensation consultant and CEO pay. As all top five executives are important in company management and they are always considered as a group, I included top five executives ranked by total compensation.
Furthermore, when running the regressions, I also included models with only CEO compensation and other top five executives compensation except CEO as my dependent variable separately. By doing this, I compared different levels of effects that the change in having a compensation consultant post on only CEO compensation and on other top five executives compensation except CEO. To investigate the relationship between compensation consultant and executives’ pay, I have two related hypotheses. I first interested in the association between change in having a compensation consultant and change in executives’ compensation. That is, whether a firm changes from not having a consultant to having one or from having one to not having one influence the change in executives’ compensation. Based on this, I form hypothesis 1 as follows:
H1: There is a positive relationship between a change in having a compensation consultant and change in executive’s compensation.
In addition, firms may do not have only one compensation. Sometimes, number of compensation consultants may change. I consider this change behavior will influence the compensation as well, so I form my hypothesis 2.
H2: There is a positive relationship between change in number of compensation consultant and change in top five executives’ compensation.
If the hypothesis 1 and 2 is rejected, then these rejections would indicate that change in top five executives’ higher compensation is result from firm characteristics, CEO characteristics, executive characteristics or corporate governance. For example,
firm may change their top five executives’ compensation because the firm enlarges their size or the firm gains more return compared with prior years.
Next, I investigate whether CEO power can influence firm behavior about having a compensation consultant and change in numbers of compensation consultants. As CEO has a great power on company’s decision making, this investigation is important to show whether CEO use their power to extract excess compensation through change in having a compensation consultant or change in number of compensation. I expect that a powerful CEO may not want to hire a compensation consultant if those consultants are indeed preventing managerial rent-‐extraction. In addition, top five executives always view as a group, so CEO power may also influence other top five executives. Based on the reasons discussed before, I formulate hypothesis 3 and 4 as:
H3: Higher CEO power has a negative association with having a compensation consultant.
H4: Higher CEO power has a negative association with change in number of compensation consultants.
A rejection of Hypothesis 3 and 4 would show that CEO power does not have a negative effect on firm’s decision about whether to hire a compensation consultant or firm’s change behavior in number of compensation consultant. These will provide no evidence that powerful CEO would use compensation consultant to extract excess compensation.
ii.
Sample Construction
My sample includes 3701 publicly-‐listed US firms and 18822 observations from 2006 to 2013. The data I used in this thesis has five components.
One is data on compensation consultant. To collect this data, I used a script1 that downloaded and automatically searched all 10-‐K filings from publicly listed firms as available in the SEC’s EDGAR database. At last, this data included a sample of 3546 US firms that have compensation consultants in different year and the observations are 18944. Table 3 reports compensation consultant market share based on number of clients for each fiscal year from 2006 to 2013. M stands for multi-‐service firms, S denotes specialist compensation consultants and specialist firms that were spun-‐off by a multi-‐service parent are denoted by *. Firms are ordered by market share in 2006. As in table 3, the consultant with highest frequency is Towers Watson and predecessors with 26.85% of the sample in 2006. Towers Watson and predecessors includes Towers Perrin, Watson Wyatt and Towers Watson. Pay Governance is the specialist compensation consultant that were spun-‐off by it from 2009. The second one is Mercer who took 16.25% of the sample in 2006, followed by Aon Hewitt, associated co’s and predecessors (15.94%). Aon Hewitt, associated co’s and predecessors includes Hewitt&Associates, Aon, Aon Hewitt, Radford and McLagan. Following is Frederic W.Cook (12.74%), Pearl Meyer (7.4%), Hay Group (3.58%), Amalfi Consulting (3.32%), Compensia (2.63%), Semler Brossy (2.07%), Deloitte
(1.76%), Buck Consultants (0.94%) and PricewaterhouseCoopers (0.75%).
The second component is data on executive compensation. I collected this data from ExecuComp of Wharton WRDS. This data includes a sample of 2395 US firms and 88732 observations.
The third component is data on firm information. This data is from COMPUSTAT of Wharton WRDS. Through this data, I constructed my control variable of firm characteristics.
The fourth component is data on stock price. This data is from CRSP of Wharton WRDS. In this data, the time period is from 2004 to 2013, I extended time period from 2006 to 2004 because I need to calculate the prior stock return for 2 years.
The fifth component is data on SIS/RiskMetrics Directors. This data was downloaded from Wharton WRDS. The time period is from 2007 to 2013. I also use CRSP-‐COMPUSTAT linktable from WRDS to link my data with GVKEY to ISS/RiskMetrics data. Using this data, I made my corporate governance control variables.
I then merged all the data and finally obtained my final sample of 3701 US firms and 18822 firm-‐year observations.
iii.
Variable Measurement
The dependent variable is Δlog(PAY)it that means the change in executives compensation. PAY is defined as average compensation of top-‐5 executives. Top-‐5 executives are executives who ranked at the first 5 by sum of salary and bonus. I use
TDC1 variable from ExecuComp. TDC1 is defined as total annual compensation, which consist of salary, bonus, non-‐equity incentive plan compensation, grant-‐date fair value of stock awards, grant-‐date fair value of option awards, deferred compensation, and other compensation. In addition, I also generated another two sub dependent variables. They are Δlog(CEO PAY) and Δlog(Other top-‐5 PAY).
Δlog(CEO PAY) means compensation change of CEO. And Δlog(Other top-‐5 PAY)
means compensation change of top-‐5 executives except CEO, where Other top-‐5 PAY is average compensation of top-‐5 executives except CEO. What’s more, as most research on compensation consultant, I used the natural logarithm of PAY, CEO PAY and Other top-‐5 PAY with the purpose of adjust the highly skewed distribution of pay.
The key independent variable for hypothesis 1 is D_ΔCONSULTANTit that is change in having a compensation consultant from t-‐1 to t. D_CONSULTANT is a dummy variable that equal to 1 if the firm has a compensation consultant and 0 if not. To be clearer, if the company without hiring a compensation consultant changes to have one, then D_ΔCONSULTANTit will be 1. If the company keeps the same as before in hiring a compensation consultant, then this variable will be 0. If the company with hiring a compensation consultant changes to without one, then this variable will be -‐1.
The key independent variable for hypothesis 2 is ΔCONSULTANTit that is defined as the change in number of compensation consultant from t-‐1 to t. CONSULTANT is
the number of compensation consultant that the firm hired.
To test whether the CEO power influence the firm’s decision on having a compensation consultant and change behavior in number of compensation consultant, following Bebchuk, Cremers and Peyer (2011), I used CPS as a measure of CEO dominance. CPS is the CEO Pay Slice that defined as the percentage of the total compensation to the top five executives that goes to the CEO. It is calculated as the total compensation of CEO scaled by the total compensation of top 5 executives.
In the choice of main control variables, I followed Armstrong, Ittner and Larcker (2012).
For firm characteristics, the control variables are firm size, book-‐to-‐market ratio, return on assets, change in return on asset, prior 1 year return and prior 2 year return. This is because previous study indicated that CEO compensation level has a positive relationship with investment opportunities, firm size, operating and stock price performance, Lambert and Larcker (1987); Smith and Watts (1992); Core and Guay (1999). Firmsize is the natural logarithm of market capitalization at the beginning of the fiscal year, where market capitalization equals multiply of company’s outstanding shares (COMPUSTAT item CSHO) and common stock price (COMPUSTAT item PRCC_F). In addition, the Book to market ratio was used to measure the firm’s investment opportunities, which influence the CEO compensation positively. Book-‐to-‐market Ratio is the book value of total assets (COMPUSTAT item AT) scaled by market capitalization. The operating performance
in the previous year is a good way to evaluate expected performance in the current year and this is likely to influence the pay levels of executives. What’s more, the change in operating performance is a determinant of the bonus payouts and other variable pay elements that are based on the period’s operating performance. The operating performance is measured by return on asset and the change in operating performance is change in return on asset. ROA is return on asset, which is calculated by operating income after depreciation (COMPUSATA item OIBDP minus DP) scaled by total assets. ΔROA is change in return on assets between the current fiscal year and the prior fiscal year. For stock price performance, I also used two variables.
PriorReturn(-‐1) is the stock return over the prior fiscal year. It is calculated by the
common stock price of current fiscal year minus the common stock price of prior fiscal year. PriorReturn(-‐2) is the stock return over the preceding year. It is calculated by the common stock price of current fiscal year minus the common stock price of the preceding year.
For CEO characteristics, the control variables are CEO proportion incentive pay, the CEO age and CEO tenure as all of them may influence compensation levels because. CEO age and CEO tenure can capture CEO experience that is often used as a justification CEO pay. And the proportion of CEO’s incentive pay is substitute for annual pay. If the incentive pay proportion is high, there may be little reason to provide additional incentive using annual compensation, and that will lead to lower compensation levels. CEOIncentivepay is the incentive pay proportion of CEO. It is
defined as the long-‐term incentive compensation (COMPUSTAT item TDC1 minus Salary and Bonus) scaled by the total compensation of CEO. CEOAge is age of CEO in that year. CEOTenure is number of years the CEO has held the title of chief executive officer.
In addition, because my research is focus on the top-‐5 executives’ compensation, their age and the proportion incentive pay may also influence their total compensation. I include the average age of top-‐5 executives and the average proportion incentive pay of them as well. Following is the detail descriptions of these variables. AverageIncentivepay is average of the incentive pay proportion of the top-‐5 executives. It is defined as the long-‐term incentive compensation (COMPUSTAT item TDC1 minus Salary and Bonus) scaled by the total compensation of top-‐5 executives. AverageAge is average of the age of the top-‐5 executives.
Furthermore, CEO power’s influence on change in pay may highly influence by the board governance. Following Armstrong, Ittner and Larcker (2012), I chose five control variables. As shown by Coles, Daniel and Naveen (2008), board size can influence managerial and board decision-‐making. I included number of directors on the board to measure board size. Number of Directors is equal to the natural logarithm of the number of directors on the board. As Core, Holthausen and Larcker (1999) proved that excess CEO compensation was correlated with board characteristics, I used the following four control variables to capture board characteristics. Outside Director% is measured by the percentage of board members
classified as outsiders. Old Director% is the percentage of board members who are at least 69 years old. Busy Director% is the percentage of board members who serve on at least two boards of directors. Outside Chairman is a dummy variable that equal to one if the chairman of the board is classified as an outsider and zero otherwise. All these variables are generated using ISS/RiskMetrics Directors.
Additionally, similar to previous researches, I used year-‐fixed effects and industry-‐fixed effect to capture year-‐specific and industry-‐specific differences in executives’ compensation levels. Industry fixed-‐effect indicators are based on four-‐digit SIC codes.
Table 1 shows all description of variable measurement.
iv.
Regression Equation
I ran two panel data regressions to conduct my research. In order to test Hypothesis 1, I ran the equation (1).
Δlog(PAY)it=β0+β1D_ΔCONSULTANTit+β2Firmsizeit+β3Book-‐to-‐marketit+β4ROAit
+β5ΔROAit+β6PriorReturn(-‐1)it+β7PriorReturn(-‐2)it+β8CEOIncentivepayit+β9CEOAgeit
+β10CEOTenureit +β11AverageIncentivepayit +β12AverageAgeit
+β13Number of Directors+β14Outside Director%+β15Board Old%+β16Board Busy%
+β17Outside Chairman +αi+λt+εit (1)
β1 measures how much will the change in having a compensation consultant make contribution to the change of top 5 executives’ pay.
variable is ΔCONSULTANTit. β1 measures how much will the change in number of compensation consultants make contribution to the change of top 5 executives’ pay.
To test Hypothesis 3
D_CONSULTANTit=β0+β1CPSit+αi+λt+εit (3)
β3 measures how the CEO power influences having a compensation consultant.
To test Hypothesis 4, the only change in regression 4 is the dependent variable is ΔCONSULTANT.
In addition, to test whether change in having a compensation consultant and change in number of compensation consultants have a greater influence on change in compensation of CEO or change in compensation of top-‐5 executives except CEO, I also used Δlog(CEO PAY) and Δ(Other top-‐5 PAY) in all regressions.
IV. Data and descriptive statistics
Table 2 presents the descriptive statistics of compensation, consultant, firm-‐level characteristics, CEO characteristics, executives’ characteristics and corporate governance. The table reports number of observations, mean, standard deviation, min and max of all variables on the full sample. Because of outlier, this discussion is focus on mean. In the compensation part, the mean salary of all top five executives in this sample is $679.2 thousand dollars. While the mean CEO salary is $781.3 thousand dollars and the mean other top-‐5 executives’ salary (except CEO) is only $392.6 thousand dollars. This shows that CEO salary is almost twice other top-‐5 executives’ salary and CEO salary is also higher than the mean salary of all top five