CEO Pay Slice and Firm Value: Evidence from UK Panel Data

This study examines a relationship between the CEO Pay Slice (CPS) – the fraction of the top five executive directors’ total compensation that is captured by CEO and firm value in the UK. CPS reflects the relative importance of CEO as well as the extent to which the CEO is able to extract rents. CPS may also alter effectiveness of board performance by influencing cooperation and cohesiveness among its members. Using a large sample of UK-listed companies over the 1997 to 2010 period, we document evidence supporting a negative relationship between CPS and firm value as measured by Tobin’s Q. Our results are consistent with the hypothesis that high CPS is associated with agency problems, and is likely to impact negatively on the executive team’s spirit and motivation. Our results have major implications for the on-going debate on how to reform executive remuneration, and highlight the importance of considering remuneration issues at the board level, supporting the principles of UK Corporate Governance Code (2010). JEL classification: G32, G35, G38, J33, L29


Introduction
Corporate scandals about 'fat cats' compensation packages in Britain[1] are a timely reminder that pay-performance problem in the UK requires further attention. Shortcomings in regulation of compensation-related issues have been addressed by the UK Corporate Governance Code 2010 (The Code), with the particular attention being paid to the importance of establishing definite connections between director's remuneration and firm performance [2]. In September 2013, the UK Government also introduced a mandatory "say on pay" legislation allowing firm's shareholders to vote on the remuneration of executives.
In this study, we aim to shed additional light on the link between executives' compensation and a firm performance in the UK context. In particular, we investigate whether pay inequality between CEO and top executives affects performance of British companies. We hypothesize that fraction of aggregate compensation of top-five managers captured by the CEO personally (the CPS) impacts board effectiveness, which in turn affects firm outcomes. In our analysis, we controls for several corporate governance characteristics (board composition, board size, CEO duality, CEO tenure, and board busyness) and for various firm characteristics (company age, company size, ratio of capital expenditures to total assets, and leverage). We use a large sample of non-financial companies listed on the London Stock Exchange, comprising 1,401 firms and 6,959 firm-year observations over the 1997 to 2010 time period. Our empirical methodology includes estimation of panel data by using various fixed effects models.
i. See The Guardian: "David Cameron to curb 'fat cat' pay with people power"`, 7 th January, 2012, available at http://www.theguardian.com/business/2012/jan/07/david-cameron-fat-cat-pay. "We've got to deal with the merry-go-round where there are too many cases of remuneration committee members sitting on each other's boards, patting each other's backs and handing out each other's pay rises," he said. "We need to redefine the word 'fair'. We need to try to give people a sense that we have a vision at the end of this, of a fairer, better economy, a fairer, better society, where if you work hard and do the right thing you get rewarded" ii.
Section D: Remuneration. Main Principle: "levels of remuneration should be sufficient to attract, retain, and motivate directors of the quality required to run the company successfully, but a company should avoid paying more than is necessary for this purpose. A significant proportion of executive directors' remuneration should be structured so as to link rewards to corporate and individual performance." (The UK Corporate Governance Code, June 2010: p.22).
We find consistent support for the proposition that higher CPS is associated with lower firm value as measured by Tobin's Q. Our results rule out the optimal contracting hypothesis, which suggests that high CPS is determined deliberately by a company as the motivation incentive (to motivate CEO as well as top executive directors) with the view to improve firm outcomes. However, the results strongly support the agency perspective, suggesting that high CPS level could be due to the agency problem in firm with powerful and influential CEO. In  Yermack, 1996;Bebchuk and Cohen, 2005;Gompers et al., 2003;Bebchuk et al., 2009 among others). We contribute to this literature by considering another aspect of governance arrangements, the CPS, and its impact on firm performance. Finally, our work enhances the literature that analyses different CEO qualities and characteristics and their effects on firm outcomes. We highlight CPS as an important feature, which can provide additional insights into understanding of CEO compensationfirm performance link.
The reminder of the paper is organised as follows. We provide a theoretical background iii. See Zalewska (2014a,b) for detailed discussion of the UK board mechanisms and structures. and develop our hypotheses in Section 2. Section 3 contains the sample description and summary statistics. Section 4 outlines the methodology used for the analysis. Section 5 examines the relationship between CPS and firm value. Section 6 concludes.

Remuneration Disparity and Firm Performance
Academic literature suggests that pay difference within top management has important consequences for functional efficiency of the management team and, subsequently, on firm performance (see Bebchuk and Grinstein, 2005;Murphy and Zabojnik, 2007;Lee et al., 2008;Fredrickson et al., 2010;Fridman and Saks, 2010;Bebchuk et al., 2011;Zalewska, 2014a;Forbes et al, 2014 among others). Even though researchers confirm existence of the relationship between pay inequality and firm performance, there is a disagreement regarding the nature of this relationship. Lazer and Roshen (1981), Rosen (1986), Eriksson (1999), Henderson and Fredrickson (2001), DeVaro (2006a, 2006b, Lee et al. (2008), Kale et al. (2009), Rankin and Sayre (2011) find that pay disparity has a positive effect on company performance. On the other side, Hayward and Hambrick (1997), Bloom and Michel (2002), Carpenter and Sanders (2002), Siegel and Hambrick (2005), Fredrickson et al. (2010), Bebchuk et al. (2011), Zalewska (2014a, report that a wide remuneration gap among executives affects firm outcomes in a negative way. Despite the growing body of literature on executives' remuneration disparity, there are only a handful of studies examining its effect on firm performance using UK data. Correa and Lel (2014) investigate the effect of "say on pay" law on executive compensation, CPS and firm value using a large cross-country sample from 39 countries including the UK. They find that CEO pay -firm performance links become stronger when "say on pay" laws are implemented, and that companies that previously had greater CPS, experience significant performance improvements. These findings imply a negative correlation between CPS and firm outcomes, and are consistent with results from previous research (e.g., Bebchuk et al., 2011). Zalewska (2014a) analyses the link between remuneration dispersion at executive board level and firm performance using a large sample of British companies. She also unveils a negative relationship between remuneration dispersion and performance. Forbes et al.
(2014) criticise CPS as a valuable measure to be used in the analyses of pay disparitycorporate performance relationship, and introduces a Gini coefficient as an alternative.

Optimal Contracting
Under the optimal contracting theory, CEO compensation is determined by a complex set of factors and reflects CEO talent, ability, experience, and career concerns. Optimal compensation reflects the extent to which companies are willing to offer 'tournament' incentives to top executives other than a CEO. Optimal contracting arguments (See Edmans and Gabaix, 2011 for review) suggest that high CEO pay -relative to pay of other top executives -is determined deliberately by companies as motivation incentive with the view to improve firm outcomes. In a typical rank order tournament framework, the best performer is promoted to the next level in the managerial hierarchy. The promotion guarantees a higher pay level, so the framework motivates executive directors to exert greater efforts and perform better. Earlier empirical research on labour economics (e.g., Bognanno, 2001), and the most recent research on corporate finance (e.g., Kale et al. , 2009) use the compensation gap between CEO and lower-rung executives as a measure of tournament incentives. Lee et al. (2008) and Kale et al. (2009), among others, find tournament incentives to be an important mechanism in motivating managers.
Based on the above, we hypothesize that companies intentionally decide to set high CPSs to motivate their CEOs and top executives other than CEOs. CEO is motivated to be a good steward and make every effort to ensure successful company performance, because he/she takes care about his/her own reputation (Gibbons and Murphy, 1992); in turn, top executives, other than CEO, are also motivated to perform better while competing for the CEO position.
This leads us to the following hypothesis: Hypothesis 1: CPS is positively associated with firm value.

Social Comparison Perspective
In contrast to the optimal contracting theory, there is a very strong opposing view on the effects that high remuneration disparity has on a firm value. Fong et al. (2010) argue that compensation should reflect a manager's ability, but at the same time should not lay the foundation for strong feelings of inequality/injustice among peers on the labour market.
Milgrom and Roberts (1992) (Brill, 1993;Cook, 1990). Hicks (1963) introduced the notion that large pay differences may have a negative impact on employees through feelings of inequity and it may lead to a weaker employees' dedication increasing a dysfunctional conflict, which, in turn, "diminish the efficiency of the team" (Hicks, 1963: p. 334). Akerlof andYellen (1988, 1990), Milgrom and Roberts (1988) and Levine (1991) build up on the earlier work of Hicks (1963) and argue that low pay differences may have positive effect on employees' diligence and productiveness by creating well-balanced and efficient labour relations leading thereby to better outputs. Levine (1991) also demonstrates that lower level of pay dispersion leads into better employee cohesiveness and productivity [4]. Based on the above, we argue that CEO may be able to entrench, maximise his/her own power, and extract rents by securing higher than optimal compensation, regardless of his/her quality of service. A high CPS level may result in lower cohesion and less cooperation among board members, which negatively affects firm performance. This leads us to the following hypothesis: Hypothesis 2: CPS is negatively associated with firm value.

Sample Selection
iv. This argument is also consistent with research on cooperation in general economic situation (Milgrom and Roberts, 1988;Lazear, 1989;van den Assem et al., 2012). v.
The social comparison theory was introduced by Festinger (1957), and underlines equity theory oriented concerns (Adams, 1965).
For this study, we use a large sample of non-financial companies listed on the London Stock Exchange. We obtain firms' financial and market information from Thompson Datastream, and corporate governance and directors' compensation information from BoardEx. The sample period is from 1997 to 2010, and it includes all firms whose information is available from these two sources. The choice of year 1997 as the start year for our sample window is due to the limited availability of corporate governance data prior to this year.
We merged the data from BoardEx and Thomson Datastream and ended up with British corporate boards are, on average smaller than American boards. Only 16% of our sample companies have five or more executive directors at the board level. We compute CPS as the fraction of the total compensation paid to a group of minimum two and maximum five top executives, that is received by the CEO. We use Tobin's Q as a key measure of corporate performance. We control for other potential determinants of firm value, found to be important in the previous studies (see Bebchuk et al., 2011;McNulty et.al., 2013;Zalewska, 2014a;Forbes et al., 2014) and include firm size, company age, capital expenditures, and leverage in our model. We also collect information about the governance structure of each firm, such as board size, board composition, board busyness, CEO tenure, CEO duality, whether the CEO is insider or outsider, i.e. was/was not an employee of the firm before his/her appointment to the CEO position, and information on the compensation of executive directors other than CEO. All variable definitions are provided in Table 1. Tables 2 and 3 give sample calculation examples for Board Busyness and CPS respectively. *****Insert Tables 1, 2 & 3 here*****

Data description and summary statistics
We report summary statistics in *******Insert Table 4 here ******* In 2002, the UK was the forerunner in mandating that shareholders be allowed a non-binding, or advisory vote on executives' pay ("say on pay").
Saks, 2010). Introduction of advisory "say on pay" law in 2002 has not changed this increasing trend. In particular, mean CPS has been increasing gradually, from 32% in the year 1997 to 50% in year 2010, with an average CPS around 40% before the introduction of "say on pay", i.e. before 2002, and average CPS around 47% upon implementation of this law, i.e. from year 2003 onwards. This is in agreement with results in Ferri and Maber (2013) who find that introduction of "say on pay" has a limited effect on the levels of CEO compensation.

Methodology
In this section, we examine the effect of CPS on company performance. We follow the literature that relates firm performance to various corporate governance characteristics and use Tobin's Q as a measure of firm performance (see Demsetz and Lehn, 1985;Morck, Shleifer, and Vishny, 1988;Lang and Stulz, 1994;Gompers et al., 2003). The CPS definition is adopted from Bebchuk et al. (2011) with a minor modification, which was necessary due to the difference in board sizes in the UK and US[8].
We include control variables that have been considered important in the previous literature (see Bebchuk et al., 2011;McNulty et al., 2013;Zalewska, 2014a). We include Board Composition, which is a proportion of executive directors at the board level.
Considering the nature of data available and difficulties with identifying independent directors, we use Board Composition measure as a proxy for board independence (a lower proportion of executive directors at the board level is associated with higher level of board independence). Previous academic research finds board independence important in designing viii.
See Section 3.1 for the definition of CPS.
a CEO compensation plan (see Mehran, 1995;Ozerturk, 2005). We also control for board size and include natural logarithm of a total number of directors at the board level (Board Size). Academics provide controversial evidence on the relation between board size and company performance, with some documenting positive (Pearce and Zahra, 1992;Dalton et al., 1998;Jackling and Johl, 2009), while others reporting negative association (Yermack, 1996;Van den Berghe and Levrau, 2004 where the ~ (tilde) defines demeaned variables, and Performance it is our performance measure, i.e. Tobin's Q. All other variable definitions are in Table 1.
We ran fixed effects models with robust standard errors and robust standard errors clustered by industry. We also use fixed effect models with Driscoll-Kraay (1998) standard errors, which are robust to general forms of cross-sectional and temporal dependence. We also use industry-adjusted CPS in each firm's industry at the same FTAG3 level in the same year. In addition, we examine whether our results are robust to alternative specification of CPS based on the total compensation of a maximum of three (rather than five) executive directors (CPS 3 directors). ********Insert Table 6 here*******

Multivariate analysis
In this section we discuss our empirical results concerning the association between CPS and firm value measured by Tobin's Q. The regression results are reported in Table 7.
We separately report estimation results using fixed effects models with White (Panel A), robust clustered by industry (Panel B), and Driscoll-Kraay (1988) (Panel C) standard errors correction methods. Our main model is Model 1, with Tobin's Q being regressed against CPS and our selected firm and governance control variables. We find that CPS coefficients are negative and significant (at the 1% level in Panel A and Panel B, and at the 10% in Panel C).
In terms of economic significance, one standard deviation change in CPS is associated with a reduction in Tobin's Q by 11.91%.
We consider industry-adjusted CPS[9] (Model 2) and CPS computed using total compensation of maximum three executive directors (Model 3). The results remain robust to these alternative specifications with negative and statistically significant CPS coefficients.
These results are consistent with the view that high CPS adversely affects firm performance,  (2014a), who argue that a wide remuneration gap among executives affects firm outcomes in a negative way.
Throughout our analysis we were not able to find support for the optimal selection argument (Hypothesis 1). Considering that UK boards are not strongly hierarchical (see Zalewska, 2014aZalewska, , 2014b, our results suggest that "tournament incentives" are ineffective in British companies. Examining control variables in the regressions, we find some interesting results. One of our corporate governance characteristics, Board size, has coefficients that are negative and statistically significant (at the 1% level), supporting the view that small boards are more efficient and perform better than their large counterparts when it comes to managing company performance. Both company size and capital expenditure coefficients are positive and statistically significant (at the 1% level), suggesting that bigger companies and those with higher capital expenditure levels perform better. We also find positive relationship between leverage and Tobin's Q. These findings are in line with findings in previous literature (see Bebchuk et al , 2011;McNulty et al., 2013;Zalewska, 2014a). ********Insert Table 7 here*******

Conclusion
In this paper, we investigate how CPS, the proportion of maximum top-five executive directors' aggregate compensation captured by CEO, affects firm performance. We offer new insights into the pay inequality -performance relationship by evaluating three different arguments that are prevalent in the finance and management literature. One view claims that high CPS level distinguishes a company's CEO and helps to create a good competition spirit within the board room resulting in better corporate performance. However, two other views suggest exactly opposite: high CPS can be an indicator of agency problems in a company in which a powerful CEO extracts unjustified rents (agency argument), and could harm board effectiveness by impairing team cohesiveness and motivation (social comparison argument), in either case resulting in poor corporate performance Our analysis reveals a negative association between CPS and corporate performance measured by Tobin's Q. This evidence supports both the agency and social comparison arguments. The results of our study are robust for controlling for various firm, board and CEO characteristics. Our results are also robust to the different specifications of CPS. Our findings are in line with Bebchuk et al. (2011) andCorrea andLel (2014). However, this is the one of the first studies that we are aware of, which investigates the CPSperformance relationship using the broad sample of UK-based companies [10]. We find that results from the UK sample are similar to those from studies on US companies. However, the underlying reason for the negative relationship between CPS and firm performance could differ between the UK and US contexts. Considering the specificity of UK corporate boards (see Zalewska, 2014a,b), it is natural to put forward the social comparison argument as an important reason for the negative associations between CPS and firm performance [11].

Given the changes in remuneration practices introduced by the UK Corporate
Governance Code (2010) and the "say on pay" legislation (2013), we argue that CPS is an important aspect of firm governance and management that deserves attention of researchers and policy makers. The fact that high CPS negatively impacts on firm performance has strong implications for the on-going debate on how to reform executive remuneration so that it x.
Forbes, Pogue and Hodgkinson (2014) use smaller sample of companies, considering firms from the UK FTSE 100 index in their study of CPS-performance relationship. xi.
UK companies are generally characterised by high corporate governance standards, but agency problems may still exist in some companies. However, considering the attention the business community has given to the issue, and the recommendations provided by the most recent UK Corporate Governance Code (2010), it is natural to assume that agency conflicts would be minimal, and that the social comparison argument is more likely to explain the negative CPSperformance relationship.
provides the right incentives. It highlights the importance of considering remuneration issues at the board rather than at the CEO or at the sectoral or industry levels, and supports the UK Corporate Governance Code (2010) principles [12]. UK corporate governance reforms move towards increasing board's responsibilities for company's performance, and it is important to consider board-wide remuneration issues without narrowing them down simply to the details of CEO compensation.
xii. "The performance-related elements of executive remuneration… should be sensitive to pay and employment conditions elsewhere in the group" (Supporting principle, Section D: Remuneration, The UK Corporate Governance Code, 2010: p. 22).

Table 1 Variable Definitions
The data variables refer to the corresponding compensation and corporate governance variable identifiers in the BoardEx annual database and to the corresponding performance and firm characteristics variables identifiers in the Tomson Datastream database.

Compensation CEO pay slice (CPS)
The fraction of the total compensation to the group of minimum top-two and maximum top-five executives, including CEO that is received by the CEO.

CEO pay slice (CPS), 3 directors
The fraction of the total compensation to the group of minimum top-two and maximum top-three executives, including CEO that is received by the CEO.

Board busyness
The proportion of busy directors at the board level. Busy directors are defined as directors holding three or more directorships, including the "home" company, in the public companies at the same time.

Board composition
The proportion of executive directors on the board. Total number of supervisory directors divided by the total number of all directors on the board.

Board size
The natural logarithm of the total number of all directors on the board.

CEO tenure
The number of years directors have served on the board

Duality
Indicator variable: equals one if CEO and Chairman is the same person CEO outsider CEO Outsider is a dummy equal to one, if CEO was working at the firm for less than one year before becoming CEO.

Performance measure
Tobin's Q (Book value of assetsbook value of common equitybalance sheet differed taxes + market value of equity)/book value of total assets: (WC02999-WC03501 -WC03263 + MV)/ WC02999        ) and t-statistics based on the robust standard errors clustered by industry (we use FTAG3 index as an industry identifier). Panel C shows the regression results obtained by using fixed effects models with t-statistics based on Driscoll-Kraay (1998) robust standard errors. The dependent variable is Tobin's Q. CPS is the ratio of CEO total compensation to the sum of maximum top five executives' total compensation and is expressed as decimals. CPS 3 directors is the ratio of CEO total compensation to the sum of maximum top three executives' total compensation and is expressed as decimals. CPS adjusted is industry-adjusted CPS. The industry adjustment is made by subtracting industry mean CPS (at the same FTAG3 level) from firm CPS in the same year. All other variable definitions are in Appendix 1. Standard errors are in parentheses beneath each coefficient estimate. *, **, and *** indicate statistical significance at the 10%, 5%, and 1%, respectively.