Are Executive Contracts Justifiable?
Tiago Filipe Montes de Jesus
University of Amsterdam
BSc. Economics & Business
Specialization: Economics & Finance
Supervised by:
Rafael Ribas
R.PerezRibas@uva.nl
Statement of Originality
This document is written by Student Tiago Filipe Montes de Jesus 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
Since the beginning of the Euro Area, corporate governance has been a widely discussed topic. While the unification of member states and corporate practices are an ongoing development, transparency in pay is still amid controversy. This paper tests whether executive compensation in central European countries can be justified. In a simple principal-‐agent model, the justifiability of the pay setting is tested. It is observed that executive compensation is influenced more by accounting performance metrics rather than market performance. While few factors prove to be statistically significant, it is found that executive pay reacts immediately to a positive performance shock but becomes stickier when the performance is poor. Overall, the results proved insufficient to consider executive compensation contracts as justifiable.
Table of Contents
I. Introduction………...
4
II. Literature Review……….. 7
III. Methodology………
a. Data...………....
b. Empirical Method………..
10
10
12
IV. Descriptive Analysis………... 13
V. Results………... 17
VI. Discussion……… 19
VII. Conclusion……….. 20
VIII. Bibliography………..
21
IX. Appendix………. 24
I. Introduction
Corporate Governance establishes the relationships between the shareholders and its board. The shortcomings of corporate governance quality in Europe have left the shareholders to question the weak correlation between executive remuneration and company performance (Gordon, 2014 and Owen et al, 2006). One of its main concerns is the separation of control and ownership, where management may find incentives to follow its best interests rather than the shareholders’. The corporate governance scene in Europe is shown to have a weak separation of control and ownership. While in the United States the prevalent model for ownership is dispersed, in Europe, the prevalent model for ownership is that of concentrated ownership (Owen et al, 2006). In a dispersed ownership structure there are many shareholders and a clear separation of ownership and control. In contrast, a concentrated ownership structure is composed of institutional investors, families, or a small group of shareholders who hold a significant stake in a company. Hence, these investors have a certain level of control over its management. The consequential problem that EU regulations face in a concentrated ownership scenario is that it must ensure that the interests of the institutional shareholders are in line with the minority’s interests – that is, the dispersed shareholders. Recent studies have shown that European listed firms have been subject to a misalignment of interests between management and its shareholders. Complex contracts in management have managed to cloud out executive compensation policies and their justifiability.
This paper focuses on whether executive contracts in EU listed companies can be justified. This paper also highlights the contractual problem of executives in a simple principal-‐agent model. It starts by identifying performance measurability. I specify how I will quantify non-‐cash compensation and how I will measure luck1 performance factors. First, I present a case study of performance metrics and executive compensation in the Consumer Discretionary industry. Second, I use percentage changes over my specified timeframe to compare and contrast the directions of performance vs. executive pay. Accounting for fixed effects in my model, I will look at the detailed components of executive pay and test whether certain compensation
1 This term stands for random performance factor that the firm cannot influence. This is explained more in depth
components are significantly affected by various company-‐ and market-‐related factors. My research consists of four clustered compensation components: total compensation, total variable compensation, cash-‐based compensation, and equity-‐based compensation. I hypothesize that higher firm and industry performance will have a positive relation with compensation. On top of performance measurements, I also look at firm characteristics and corporate governance policies. In a European corporate governance scenario, I observe whether large blockholders can significantly influence executive compensation and whether executives holding more than one position are compensated differently. The hypothesis is that an increased number of blockholders will have a negative relationship with the level of compensation. Moreover, I hypothesize that if an executive is a Chairman-‐CEO, the level of compensation is expected to rise. I find that accounting performance measures and accounting luck performance respond significantly to total granted compensation and cash-‐based compensation. Corporate governance aspects show to not be significant, though the directions of the variables are in accordance to the hypotheses. My findings are consistent with the findings of other studies such as those of Bertrand and Mullainathan (2001) and Parthasarathy et al. (2006).
Studies have suggested that blockholders (large shareholders) in Europe don’t act as owners but rather as investors, undermining the interests of the minority (Steen, 2005). In April 2014, there were 10,400 registered companies listed in EU indices totaling a market value of more than €8 trillion (European Commission, 2014, p. 9). Under European law, these companies must comply all regulations imposed and agreed upon by the European Commission, while still operating under the supervision of their respective member state’s securities exchange commission. One of the issues that arose following the creation of the European Monetary Union was the implementation of a single authority over all member states’ financial regulations for markets and corporate governance, which caused some countries to adopt the new laws at different times. The shortcomings of concentrated ownership, being that strong blockholders have incentives to appoint desired managers to further their interests, creates a misalignment between the owners and management.
As a result of the misalignment between owners and management, Internal Market and Services Commissioner, Michel Barnier, stated that short-‐term relationships between
management and shareholders has been damaging European Companies and the Economy (Hughes et al. 2014). A potential cause for the lack of longer-‐term shareholder relationships would be the absence of more significantly dispersed ownership in corporations. In the German large market capitalization index (DAX), 25% of the corporations have institutional shareholders as the predominant shareholder (European Commission, 2014, p. 9). Meanwhile, the European average of dispersed shareholders in the financial year of 2014 was only 11%.
The European Commission has recently released new adopted measures to strengthen shareholder engagement in “say on pay” and to promote more sophisticated corporate governance practices. This would mean that investors in European listed firms would have enhanced transparency on the right as shareholders to vote on executive remuneration. At the European level, there have already been acts put forth that highlight corporate governance aspects. Acts such as Directive 2007/36/EC gives shareholders the right to information and Directive 2004/109/EC gives shareholders the right to transparency on major shareholdings and financial information. Nevertheless, in the period that the European Monetary Union has been placed under operation, many laws have only been passed in recent years. The directive addressing the protection of shareholders in a situation of a capital increase was only ratified in 2012 -‐ Directive 2012/30/EC. This illustrates how the premature foundation of the European Commission lacks focus on shareholders.
The reevaluation of policies and sound corporate governance measures are in constant development. The remuneration standards in the management body of a firm have become more highly supervised and executive contracts have become more complex. The power struggle within a governing body has led to a misalignment of interests between the executives and its shareholders. The owners, being the shareholders, are exposed to a principal-‐agent problem because they do not control the firm’s resources; instead, this is delegated to its management (Parthasarathy et al, 2006). In such case, it is in the principal’s (shareholder’s) interest to incentivize the agent (executive) to maximize the value of the firm in order to avoid agency costs and separation of control. However, one of the inefficiencies with this system is that lack of supervision on the executive’s reward vs. performance.
II. Literature Review
The determination of executive compensation can in reality be very difficult to measure. In this paper, it is assumed that a shareholder is risk-‐neutral and a CEO is risk-‐averse. In a simple principal-‐agent model the principal (shareholders) may find it difficult to observe the pay of the agent (executive). Simplified models can help us to understand the drivers of executive compensation. Previous studies show that companies often find ways to tie executive pay to various factors including performance, value of human capital of the executive, and corporate governance quality. Stemming from these various pay relationships, corporate pay metrics have established a principle tool – pay for performance.
In Holmstöm (1979) and Grossman & Hart (1983), simple agency theory is considered to derive pay for performance. In other words, CEO contracts are designed with the purpose to meet the interests of a risk-‐averse executive and a risk-‐neutral shareholder. Realistically however, pay practices may be not fully transparent, questioning the feasibility of this relationship. Morgenstern (1998) suggests that some components of pay, such as deferred compensation, are not disclosed in a transparent fashion. Cooper et al. (2009) identify that such occurrence may be due to non-‐cash components being hard to value. For example, a principal will not be able to precisely observe the value of payment in kind or the value of long-‐term equity instruments of an agent if (or for which) there are many macro factors involved.
Furthermore, the direct actions of CEOs can be difficult to observe. Assuming that shareholders are trying to maximize their value through a risk-‐averse CEO, the pay sensitivity will depend on a measurable factor: performance (Bertrand and Mullainathan, 2001). The performance of a firm is defined by actions taken by the CEO and random factors that the CEO does not influence. Essentially, one can only observe the firm’s performance itself and observable random factors. Bertrand and Mullainathan (2001) refer to these random factors as “luck.” They find that CEOs react to a general dollar as much as a lucky dollar and their compensation depend on external factors as much as internal factors. The luck measure creates a compensation bias in the sense that compensation increases with good industry performance but does not decrease with poor performance. Bertrand and Mullainathan also point out that shareholders tend to pay less attention to the compensation scheme when a firm is performing well and draw more
scrutiny if the firm performs poorly. I hypothesize that an increase in the measurement of firm and industry performances will have a positive relation with executive compensation.
The complexity in the determination of CEO compensation is evident. For example, if CEOs are in a position that permits the self-‐influence of their pay, they will weaken the pay-‐for-‐ performance relationship (Bebschuk, Fried, and Walker 2003). In existing European corporate governance guidelines, CEOs are permitted to have a voice in key committees of the board of directors, such as the Remuneration Committee and Nomination Committee. In the event that a CEO is a member of such committees, he or she can influence the pay setting and seek support within the board of directors by influencing the choice of director appointment. This can skew executive compensation without the oversight of the shareholders. Current guidelines such as the CEO having a voice in the remuneration committee do not contribute to corporate governance quality. According to Parthasarathy et al. (2006), the principal (shareholders) should have the delegated right to appoint agents and to review their compensation. Parthasarathy et al. (2006) defines that good corporate governance comes essentially from the composition of the board.
There are four criteria in assessing the composition of the board: Independence ratio on the board, separation of power between the Chairman and the CEO, owner-‐managers, and institutional shareholding. Essentially, independent directors in the board are meant to represent the dispersed shareholders’ interests and therefore the company can enhance corporate governance quality by having a higher ratio of independent directors. The European Commission requires all member states to include board independency and to incorporate independent directors in committees.
According to Parthasarathy et al. (2006) and Ghosh (2003), the separation of power between a Chairman and a CEO can contribute for better corporate governance and for board independence. Arguably, a Chairman-‐CEO will receive more for occupying both positions. Parthasarathy et al. (2006) also identify that a Chairman-‐CEO can use this situation to create a compensation bias that inflates the value of his or her human capital. For this reason, I hypothesize that if an executive is in charge of both positions, the level of compensation is expected to rise. A limitation to this hypothesis, however, is that in the two literatures
mentioned, only Ghosh (2003) found significant results despite the fact that both correlated positively to executive compensation levels.
In various studies, firm size is shown to be significant in the setting for executive compensation. The theoretical expectation is that larger firms will require more skilled CEOs to manage the firm and to increase its value. Murphy (1999) tests this theory comparing S&P 500 firms with the median ExecuComp Dataset and concludes that size is a determining factor for compensation. In my sample, the lagged form of firm size is taken. As Nickell (1981) pointed out; lagged variables in a fixed effect regression could lead to a correlation between the lagged variable and the error term and would cause biasness with a factor of 1/T, where T is number of time periods in the sample.
As previously mentioned, one of the biggest differences between corporate governance standards in North America and in Europe is the ratio of blockholders. In Europe, a large shareholder is considered to be significant once the entity owns 5% or more of the target company. By law, companies are required to report institutional ownership. Bertrand and Mullainathan (2001), whose theory was constructed from the literature of Schleifer and Vishny (1986), argue that large shareholder presence in an entity could help to solve the agency problem and improve governance because of the “intuition of having a principal around” (p. 921). In other words, when a single investor holds a significant amount of stock in another firm, the investor has incentives to scrutinize a manager more closely than the dispersed shareholders. This can become less costly and more efficient. Hence, it is hypothesized that an increased number of blockholders will have a negative relationship with the level of compensation. In contrast to having a large shareholder improving corporate governance, I hypothesize that it does just the opposite, even though it can decrease compensation levels. This is because in Bertrand and Mullainathan (2001), it is assumed that the principal-‐agent problem can be solved. I argue that this occurrence brings about another problem where the large shareholder must now ensure that board representation is still in the interest of all shareholders, hence creating a principal-‐ principal problem. Su et al (2008) define that a principal-‐principal problem is the “appropriation of value from minority shareholders by majority shareholders, often by influencing board level decisions such as asset sales and purchases” (17-‐18). This problem can create controversy in
corporate governance. For example, when the majority of a firm’s capital (>50%) is held by dispersed shareholders but at the same time there are significantly large shareholders owning 5%+ of total capital, the large shareholder must make sure that the interests of the dispersed shareholders are met. This problem can increase agency costs in a firm. Having said that, a difference in the literature by Su et al (2008) is that they argue that a principal-‐principal problem is most common in emerging economies while I assume that my sample is constituted of mature firms and developed markets (German, Swiss, Dutch, and Belgian). Nevertheless, this theory applies to the common governance problem in Europe – strong blockholders and weak separation of ownership and control. I will test these two corporate governance factors to measure whether corporate governance quality influences pay.
III. Methodology
a. Data
In this study a sample is constructed from a combination of indices that comply with the regulations of the European Commission. This aggregate index (from here onward referred to as ‘INDX’) combines several major Central European markets totaling a number of 93 component firms. The components in the INDX are the largest firms in their respective countries. These firms are from the DAX (Germany), AEX (Netherlands), BEL-‐20 (Belgium), and SMI (Switzerland). While there were initially 95 companies in the sample, as a result of reported group compensation data and missing data on company aspects, two firms were removed from the sample leaving the sample with a total of 529 observations.
The data were collected from two sources, AMA Partners’ tool, DirectorInsight, and Capital IQ. DirectorInsight contains detailed information on the compensation packages of all members of the board in European listed firms. It also contains detailed pay-‐for-‐performance metrics such as the overview of Total Shareholder Return alignment with executive remuneration. Capital IQ was used to retrieve descriptive data on the company, such as net assets, share prices, cash flows, and net income.
In my sample, the companies listen in the Swiss SMI index are included. It is worth noting that even though Switzerland does not take part in the Euro Area, it still has access to the
European Union’s Single Market due to bilateral agreements that were created in order to facilitate trade. Henceforth, Swiss listed companies must oblige to the same regulations as EU companies. Since firms on the SMI index disclose executive compensation data in Swiss Franc, the values were converted into euros according to the financial year-‐end exchange rate of each company. This is to avoid fluctuations in the exchange rate as different firms have different financial year periods. While most firms in the sample have a regular financial year (January-‐ December), some firms report their annual accounts from April-‐March or from October-‐ September. My data is reported according to each firm’s financial year. To adjust for seasonal economic changes, I include time-‐fixed effects. Furthermore, in order to avoid problems stemming from repeated time values in the regression analysis, all other executive board members that had compensation data for the same year and company were removed from the sample. This allows my analysis to solely focus on the behavior of CEO compensation. In a situation of a replacement of the CEO during the year, the CEO that remained on the board for the majority of the financial year was the one whom was kept in the sample; severance amounts were also observed. In the case of a Co-‐CEO, the CEO with the shortest tenure and the least contractual specifications was removed. This was done in order to improve the pay-‐for-‐ performance relationship between the executive and the company while avoiding repeated time values in the time-‐variant panel data sample2.
Execomp is the total annual granted compensation and total variable compensation of the executive varying at entity i and time t. Within execomp, I will further analyze whether the above factors justify the determination of solely cash compensation and equity compensation. For the sake simplicity, only the granted amounts are considered. This means that any amount in a CEO’s granted compensation that may lapse (when referring to shares or options) or be forfeited (when referring to shares, options, or deferred cash) in a future time period is not taken into account. The event of having shares lapsed or options not exercised is likely in many executive contracts, however, my model does not allow to observe such long-‐term changes since I am only
2 In the INDX there were three companies operating under the legislation of Co-‐CEOs. When setting the panels in
order to construct a panel data regression, the panels are found to be unbalanced. This is realistic due to the fact that in the component indices of the INDX, companies could have entered or exited the index, allowing for the absence of data in a given timeframe.
considering short term salary determination. That is, I observe sensitivity factors in time t-‐1 that may affect compensation in time t.
b. Empirical Method
This paper deals with four broad factors that jointly determine the total granted compensation for CEOs. The model constructed is similar to those of Devers et al (2008) and to Bertrand and Mullainathan (2001). The model is written such that execompit is a function of:
− Shareholder Wealth − Performance
− Company Characteristics − Corporate Governance
In this model, the four factors will be regressed to test how much they affect ‘execomp’. Shareholder wealth is typically measured as how much value the company delivers to its shareholders. Since the INDX is solely composed of large and arguably mature companies, total dividend payout is used as a tool to measure shareholder wealth. It is understood that a firm that issues dividends is assumed to have slowed down growth prospects and investments. However, this is not always the case. For instance, if Volkswagen AG decides not to pay dividends for 2015, one can assume that its cause is related to VW’s efforts to cut expenses to cover the costs of the scandal.
For scaling reasons, all explanatory variables will be measured as a ratio of total net assets of the firm. Therefore, in order to equilibrate the ratio, shareholder wealth will be observed as the ratio of total dividends paid out by the firm divided by the net assets of the respective entity (,%- *''%-'/')*+%'!"#"$%&$'/')*+%).
The description of all explanatory variables can be found on Table A1 in the Appendix section. Previous studies indicate that performance contains both observable and unobservable components. Within the theory established by Bertrand and Mullainathan (2001), I will analyze whether the end performance factor is a justifiable explanatory variable of his or her annual granted compensation. Instrumenting for luck is beyond the scope of this research. Instead, I will
observe the end performance of a firm that Bertrand and Mullainathan categorize as perf and also regress the observable random performance factor as independent explanatory variables. In my model, perf is divided into accounting and market performance measures.
As a proxy for market performance, I will use the price per share, yet, to maintain the same measurement unit among all explanatory variables, the price per share will have to be multiplied by the firm’s number of shares in order to get the value of all shares. Because the value of all shares in a firm is the market capitalization, I will use the ratio of market capitalization over net assets as a proxy for market performance.
To measure accounting performance, I will use the ratio of the firm’s net income over net assets. Moreover, I will also measure luck as an explanatory variable. In order to do so, I will establish peer groups based on industry. I assume that a CEO’s random performance factor is trended by the peer group under which the CEO’s company is in while excluding the performance data of the CEO’s company. This allows me to control for effects that a CEO of a company cannot influence. This is arguably a proxy for measuring luck. Because a firm is likely to compete with other firms in the same industry, observing industry gains may be optimal for the CEO to influence compensation. For instance, if the energy industry is doing exceptionally well overall, CEOs may want to ‘free-‐ride’ on the opportunity to increase their pay.
Having stated the explanatory variables that will be measured as ratios over net assets, to capture the sensitivity of company aspects to executive pay, I will use net assets to proxy for firm size. Net assets will be used as a lagged variable. That is, we seek to find a relationship in the size of a firm at time t-‐1 affecting compensation levels at time t. To proxy for governance quality, I will review the present problem in Europe, which is having strong blockholders and a weak separation of power. Therefore, using a with a dummy variable, I will analyze whether a CEO also holds a position as chairman in the board of directors. My other variable used to describe governance quality is the number of blockholders present on the board that have a 5% or more stake in the company.
In order to maintain an observable range, the natural logarithm will be taken from the total granted compensation. As a result, I am able to construct a model that is more suitable for time-‐
and company fixed effects3. The model for determining CEO remuneration is as follows:
𝑙𝑛(𝑒𝑥𝑒𝑐𝑜𝑚𝑝)",- = 𝛽<+ 𝛽>𝐷𝐼𝑉𝑆",-+ 𝛽CSP",-+ 𝛽F𝑖𝑛𝑐𝑜𝑚𝑒",- + 𝛽H𝑙𝑜𝑔𝑙_𝑛𝑎",-+ 𝛽Lind_SP ",-+ 𝛽Pind_income",- + 𝛽Uchairceo",-+ 𝛽Yblockholders",-+ 𝛾"+ 𝜒-+ 𝜀",-
Where 𝛾" is the firm-‐fixed effect varying on i, 𝜒- is the time-‐fixed effect, and 𝜀",- is the residual.
In the equation above, our dependent variables are the total granted compensation, total variable compensation, cash, and equity compensation. Since this is a panel data study, regressions will be used to best describe the model empirically in first order differences. Firm and time fixed effects are taken into consideration. For the sake of robustness, multiple extensions to the model are analyzed. After performing sensitivity checks, I am able to control for unobserved attributes that are time-‐dependent but remain entity-‐invariant. In resemblance to the business cycle, it can be assumed that companies were systematically affected by the crises, recessions, and recovery.
IV. Descriptive Analysis
The executive contract components that establish the total granted compensation are described in Table A1 of the Appendix. The components for total variable compensation are the cash bonus, value of the short-‐term incentive (STI) shares, deferred cash, value of long-‐term incentive (LTI) shares, and the value of options calculated using the Black-‐Scholes model. Cash-‐ based compensation is taken from the aforementioned components as is equity-‐based compensation.
A first look at the sample shows that the average total remuneration for CEOs of listed companies in Germany, Netherlands, Switzerland, and Belgium is €4.91 Million per year. Compensation will be expressed in logarithmic form. As observed in Figure 1, the distribution of executive remuneration in logarithmic form approximates normal distribution. Nevertheless, it seems to still be in the presence positive skewedness. The positive skewedness of the total granted compensation indicates that a few number of executives receive exceptionally large compensation packages.
Figure 1. Distribution of Total Granted Compensation
Variable Mean Std. Dev. Min Max
ln(total) 15.1128 0.7861 13.0693 17.6817 total compensation € 4,909,853 € 4,272,741 € 474,144 € 48,237,297 To better understand the tails of the distribution of executive remuneration, we observe remuneration clustered by type of industry. I take the Consumer Discretionary industry as a brief case study. The industry is composed of companies that sell secondary goods in the sense that they are dependent on the state of the economy. In my sample, companies such as Adidas AG, The Swatch Group, and automakers fall into this industry. Before I perform a regression analysis, I observe the pay for performance trends in this particular industry. As it can be seen in Figure 2, compensation levels seem to be in line with performance levels.
Figure 2. Total Compensation and Cash Flow in Consumer Discretionary Industry
-‐€ 2,000,000 € 0 € 2,000,000 € 4,000,000 € 6,000,000 € 8,000,000 € 10,000,000 2008 2009 2010 2011 2012 2013 2014
Consumer Discretionary
Figure 3. Percentage Changes in Consumer Discretionary Industry
An unusual trend in Figure 2 is the slowing down of cash flows from 2008-‐2010 while total compensation is speeding up during that timeframe. In Figure 3 it becomes more clear that compensation trends immediately upward when performance is positive but compensation changes become stickier when performance changes become negative. In Figure 3 there are 4 years when the performance change is negative. In 2012 and 2014 compensation changes are also negative while 2009 and 2011, compensation changes are slightly positive. In other words, compensation and performance only have opposite signs when performance is negative. The fact that this industry’s performance trends more downward than upward may be due to the crisis which stretches for most of this time frame. Despite that, compensation levels are persistently above the index average (€ 4.91 Million). To see similar trends in other industries, refer to Figures A3-‐A5 in the Appendix.
From the figures above, we can induce that firm performance does not solely justify the composition of CEO pay. We can also intuitively predict that CEO compensation will contain complex factors that are immeasurable, such as the value of human capital of a CEO, the bargaining power of a CEO, and the costs of replacement and search for new CEOs. We can, however, examine which components of pay have been changing over time. As total granted compensation is assumed to change over time, we summarize in the table below the constituents of the remuneration package. The remuneration package has been broken down into its various
-‐200.0% -‐150.0% -‐100.0% -‐50.0% 0.0% 50.0% 100.0% 150.0% 2009 2010 2011 2012 2013 2014
% Changes Consumer Discretionary
components. For instance, if we see an increase in equity-‐based compensation over time, it could signal that the firm wants to incentivize the CEO to increase the firm’s value.
Table 1. Ratios of Remuneration Components as a Percentage of Total Granted Compensation
Year Base
Salary Bonus Cash Shares STI Shares LTI Options Cash/Severance/ Other Pension/ Deferred 2008 20.93% 19.38% 1.17% 18.26% 29.64% 10.62% 2009 21.37% 20.39% 1.76% 23.48% 23.05% 9.95% 2010 24.09% 24.21% 2.42% 21.66% 13.97% 13.65% 2011 24.53% 24.90% 2.27% 20.60% 11.80% 15.89% 2012 24.53% 23.37% 3.15% 20.53% 12.55% 15.87% 2013 22.57% 20.97% 3.03% 25.00% 10.66% 17.77% 2014 22.08% 20.87% 2.41% 23.56% 11.46% 19.62%
Throughout the early years, cash-‐ and equity-‐ based pay have shown to be fairly balanced near the 50-‐50 threshold but over time, equity-‐based pay has been gradually decreasing as a percentage of total granted compensation. This is mainly due to the increasing absence of options is executive contracts. In 2008, options made up for 29.64% of total remuneration whereas in 2014 it only made up for 11.46%. The slight increase in Long-‐Term Incentive share plans has not been able to keep to cash-‐ and equity-‐based pay at a balanced level. What this may suggest is that CEOs have gradually becoming more disconnected with the firm they manage. Balanced equity-‐cash compensation would assume that a CEO’s pay package is in both the interests of the firm and CEO.
V. Results
Table 2 regresses the model against many versions of executive pay for the sake of robustness. As discussed earlier, the total dividends paid by the firm over net assets represents the sensitivity of shareholder wealth on executive pay. To measure the performance of the firm, the market performance measure is written as SP and the accounting performance measure is income. To measure the industry performance, ind_SP is the market performance measurement and ind_income the accounting performance measurement. The lagged net assets represent the size of a firm in logarithmic form.
Table 2. Panel Data Regression on Executive Compensation (Firm and time fixed effects)
Dependent Variable
log(total
compensation) log(total variable compensation) log(cash) log(equity)
DIVS 1.693 (2.687) (17.648) 22.648 (1.652) 2.822* (19.309) -‐2.316 SP 0.142* (0.099) (0.617) -‐0.961 (0.0515) -‐0.06 (0.663) 0.445 income 2.091*** (0.55) (2.809) 5.021* 1.749*** (0.572) (3.794) 4.677 logl_na 0.176 (0.131) (0.926) -‐0.381 0.256*** (0.095) (1.425) -‐1.499 ind_SP 0.003 (0.031) -‐0.174 (0.194) -‐0.002 (0.028) 0.293 (0.355) ind_income 1.087 (2.025) 1.766 (1.662) 4.424*** (1.583) 1.505 (1.163) chairceo 0.105 (0.398) (1.79) 0.659 (0.187) 0.015 (3.828) -‐0.973 blockholders -‐0.016 (0.109) (0.631) -‐0.672 (0.074) 0.008 (1.473) -‐1.160 constant 10.514*** (3.277) (21.922) 23.498 8.015*** (2.331) (35.131) 50.283 Firm Fixed
Effects Yes Yes Yes Yes
Time Fixed
Effects Yes Yes Yes Yes
Sample Size 529 529 529 529
Adjusted R2 0.302 0.007 0.289 0.011
Coefficients are reported above and robust standard errors are reported in parentheses. *, **, *** -‐ Significant at a 10%, 5%, and 1% level, respectively.
Table 3. Test Outcomes of Regression Coefficients
Dependent
Variables compensation) log(total log(total variable compensation) log(cash) log(equity)
Prob > F 0.0052 0.4256 0.0061 0.7413
From the regression analysis in Table 2, we observe interesting results in the justification of executive pay. The ratio of a firm’s net income over the firm’s lagged assets seems to be the most significant explanatory variable. When tested against total granted compensation, an increase in the accounting performance ratio by 1% increases total compensation by 2.09% and is significant at the 1% significance level. When tested against cash compensation it is also significant at 1% affecting cash compensation 1.749% as the ratio increases by 1%. An evident
observation in this model is that cash components are easier to justify than equity. In fact, no explanatory variables showed significance when regressed against equity-‐based compensation. The regression coefficients are tested against the null hypothesis (X1=X2=…=Xn) in table 3.
It is found that the regression coefficients under total granted compensation and cash-‐based compensation are significantly different. However, when testing for total variable compensation and equity compensation, the null hypothesis is not rejected. We also fail to ignore biasness in our lagged variable, firm size. As mentioned in the Literature Review, the lagged net assets will be biased with a factor of 1/T periods. In my sample the bias factor 1/7. Because this factor is quite large and because firm size did not show consistent significance in the determination of pay, a bias in the lagged variable is not ignored.
The overall regression outcome seems to replicate that of Bertrand and Mullainathan (2001) where the accounting performance measure seems to influence executive compensation more significantly than market performance measurements. Moreover, the noise levels in the regression analysis may indicate that immeasurable factors can greatly influence pay. In determining the total granted compensation and cash-‐based compensation, the error terms are significant at 1%. The table below observes the time fixed effects in the regression.
VI. Discussion
The results indicate that compensation levels for CEOs of large companies in Europe are not completely justified. While accounting performance seems to be a consistent factor, market performance and corporate governance measures are not statistically significant even though in most cases its coefficient contains the hypothesized sign. When a CEO is also a Chairman, compensation levels are said to rise and when the number of blockholders increases, compensation levels decrease, despite the insignificance.
My results support the conclusion from Davila and Penalva (2004), which state that weaker corporate governance is associated with lower variable compensation levels and higher cash-‐ based compensation levels putting more weight on accounting returns rather than market measurements. What may seem to be counterintuitive is that shareholder wealth positively affects pay and is significant when regressed against cash-‐based compensation. From simple agency theory, an increase in the wealth of shareholders would imply a decrease in the wealth
of the manager. This is true when regressed against equity-‐based pay, but the results show to be insignificant.
Interestingly, pay for luck seems to significantly influence cash-‐based pay. That is, an increase in the wealth of the firms in the same peer group will lead to an increase in the CEOs pay who cannot influence performance. A valuable explanation for this could be that firms do not want to underpay their CEOs in order to still attract quality CEOs. In other words, firms compete against each other in the job market for the best CEOs. Hayes and Schäfer (2009) support the theory that firms compete in the job market. They conclude that no firm wants to admit to having a CEO who is paid below their peer group average resulting in firms not lagging in market expectations. Hence, if the market is performing well, firm x will use this information to increase the pay of the CEO – in this case cash-‐based pay.
It is observed that luck isn’t a significant factor for executive compensation over all types of pay except for cash-‐based. While good performance increases pay, compensation becomes stickier when the performance is poor. Bertrand and Mullainathan (2001) relate this to CEOs’ outside option. Taking the automotive industry for instance, if the industry is doing well overall but a particular company performed poorly, the firm must approximate the market standard in order to avoid the CEO to look for better outside options.
Referring back to equity-‐based compensation being insignificant across all explanatory variables could support the theory mentioned in the literature review that equity compensation is difficult to measure. In fact, the Black-‐Scholes option valuation method contains some shortcomings that may consider this model to be unrealistic. For example, the model assumes no dividends are paid to the option holder while in most cases of my sample firms pay dividends. The method also assumes constant volatility, which may be questionable in the real world.
It is unclear why firm size is only significant when regressed against cash-‐based compensation. A 1% increase in the size of the firm in time t-‐1 will increase cash-‐based pay by 0.256% in time t. Regardless, it is previously mentioned that the short time span of the sample could have caused a measurement bias. Though I did not test for this, I based my reasoning on previous studies. It would be crucial for future studies to range the time span in the sample. Not only would the researcher gather more data, he or she would also diminish measurement bias
when utilizing lagged variables.
VII. Conclusion
This thesis intends to find potential relationships that justify CEO salaries. In order to get a better understanding of how different parts of compensation are awarded, the independent variables were tested against four types of pay: total compensation, total variable compensation, cash-‐based compensation, and equity-‐based compensation. Panel regressions with fixed effects are adopted. While performance is a factor of pay, accounting performance seems to be what justifies executive pay more. It is found that accounting performance, whether from the firm or from a peer group increases cash-‐based pay by 1.749% and 4.424%, respectively. As for market performance, it is only found to be significant when testing against the total granted compensation.
There was not enough evidence to conclude that corporate governance quality is significant in all types of pay. It should be noted that the model was highly simplified, for instance when measuring equity and valuing shares. Moreover, other unobservable factors that are theorized to be relevant to the model were not taken into account. Such factors are the value of human capital of a CEO and unobservable measures of performance.
This suggests that improvements in the model can be made by replicating this research with a more extensive model and a larger number of observations. If tested at European level with 300+ companies, one could observe whether accounting performance is a robust determinant of executive pay in the whole Euro Area.
Furthermore, by instrumenting unobservable factors, one could reduce noise levels and uncertainty in the pay setting process. This model can be further developed and extended to the trade-‐off theory, where it is tested the CEOs incentives to seek outside options. Ultimately, it is suggested to measure risk-‐taking decisions from CEOs in order to observe whether CEOs get
VIII. Bibliography
Bebchuk, L. A., Fried, J. M., and Walker, D. I. (2002). Managerial power and rent
extraction in the design of executive compensation. University of Chicago Law Review 69, 751-‐ 846.
Bertrand, M., Mullainathan, S. (2001). Are CEOS Rewarded For Luck? The Ones Without Principals Are. The Quarterly Journal of Economics. 901-‐930.
Cooper, M. J., Gulen, H., and Rau, P. R. (2009). Performance for pay? The relationship between CEO incentive compensation and future stock price performance, 1-‐22.
Davila, A., and Penalva, F. (2004). Corporate Governance and the Weighting of Performance Measures in CEO Compensation. Working Paper No. 556, IESE Business School, University of Navarra.
Devers, C., McNamara, G., Wiseman, R., and Arrfelt, M. (2008). Moving Closer to the Action: Examining Compensation Design Effects on Firm Risk. Organizational Science, 19(4), 548-‐563.
European Banking Authority. The application of Directive 2013/36/EU (Capital Requirements Directive) regarding the principles on remuneration policies of credit institutions and investment firms and the use of allowances (pp. 2-‐10). (2014). European Banking Authority.
European Commission (2014, April 9). Directive of The European Parliament and of the Council On Amending Directive 2007/36/EC as Regards the Encouragement of Long-‐Term Shareholder Engagement and Directive 2013/34/EU as Regards Certain Elements of the Corporate Governance Statement. Brussels: European Commission. Print.
Finkelstein, S and R D'Aveni (1994): ‘CEO Duality as Double-‐Edged Sword: How Boards of Directors Balance Entrenchment Avoidance and Unity of Command', Academy Journal of Management, 37(5), 1079-‐1108.
Ghosh, Arijit (2003). Board Structure, Executive Compensation and Firm Performance in Emerging Economies: Evidence from India, IGIDR Working Paper Series, IGIDR,Mumbai Gordon, S. (2014, December 28). Top Managers’ Pay Reveals Weak Link to Value. Financial
