Subjectivity as a Contributor to Excessive
CEO Compensation
Master thesis Accountancy (EBM869B20)
Rijksuniversiteit Groningen
Faculty of Economics and Business
S. G. ter Horst
Student number: S2199173
Nieuwe Ebbingestraat 31a
9712ND GRONINGEN
S.G.ter.Horst@student.rug.nl
Thesis supervisor: Dr. R. C. Trapp
Thesis co-assessor: N. J. B. Mangin
2
Table of Content
ABSTRACT 3
1. INTRODUCTION 4
2. LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT 8
2.1 CEO compensation 8
2.2 Performance measures 11
2.3 Downsides to subjective performance measures 12
2.4 Agency Problems 14
2.5 Effective Compensation Committees 15 2.6 CEO influence 18 3. METHOD SECTION 20 3.1 Independent variable 21 3.2 Dependent variable 21 3.3 Moderating variables 22 3.4 Control variables 24 4. RESULTS 25 4.1 Descriptive statistics 25 4.2 Correlation analysis 27 4.3 First-stage regression 27
4.4 Second-stage regression and hypothesis testing 29 5. DISCUSSION 30 6. CONCLUSION 32 6.1 Contribution 32 6.2 Implications 32 6.3 Limitations 33 6.4 Future research 34
APPENDIX A. Calculation of Discretionary Bonus 35 APPENDIX B. Variable Definitions and Descriptions 36 7. REFERENCES 38
3
Abstract
Purpose – The aim of this paper is to study whether discretionary bonuses contribute to excess CEO compensation. This prediction is based on the unexplained existence of excess CEO compensation and downsides to the use of subjective performance measures like inflated ratings and the development of favoritism among evaluators. In addition to this, this paper attempts to find an answer whether this relationship is mitigated by effective compensation committees and strengthened by CEO influence.
Methodology – Data of 125 randomly selected firms listed on the S&P 500 is collected for the year 2015. A first-stage OLS regression analysis is performed to estimate levels of CEO total compensation based on firm performance and economic determinants used in prior research. This estimate is then subtracted from actual CEO total compensation to result in excess CEO compensation. Excess CEO compensation is then used in a second-stage OLS regression analysis. Discretionary bonuses are obtained from proxy statements provided by the firms and proxies are used for effective compensation committees and CEO influence. Findings – The findings provide empirical evidence of the contribution of discretionary bonuses to excess CEO compensation. In addition to this, this study also provides evidence of the mitigating effect of effective compensation committees and the strengthening effect of CEO influence.
Originality – This is the first study to investigate whether the identified downsides of the use of subjective performance measures in previous studies result in excess CEO compensation, which previous research is still working on to fully explain. This study is also the first in investigating the role of effective compensation committees and CEO influence in the relationship between discretionary bonuses and excess CEO compensation. The findings of this study can potentially be important in the further awarding of discretionary bonuses and the use of subjective performance measures.
Keywords – Discretionary Bonus, Subjective Performance Measure, Excess CEO Compensation, Compensation Committee, CEO, Incentives, Shareholder
4
1. Introduction
This research paper investigates the role of subjectivity in CEO compensation contracts. More specifically, this paper investigates if subjective performance measures contribute to excess CEO compensation and if this relationship is different when the compensation committee can function effectively or when the CEO has influence on the setting of compensation.
The appropriate level of CEO compensation has been the subject of a large body of academic research and many others like shareholders, politicians, regulators and the business media participated in the debate (Bogle, 2008; Core et al., 2003; Hill et al., 2016). Prior research describes the appropriate level of CEO compensation as compensation that is linked to firm performance (e.g. return on assets) or economic factors (e.g. firm size). When CEOs receive more compensation than explained by these determinants they receive excess compensation, which is detrimental to shareholder value (Brick et al., 2006; Core et al., 1999 & Hill et al., 2016). The compensation of a CEO is set up in a contract, which consists of a fixed salary and an incentive component (Murphy, 1999). Annual incentives are compensated in cash, while long-term incentives are compensated in equity (Carter et al., 2016). The long-term equity incentives received most of the attention in prior research, possibly because they account for the largest part of CEO compensation and are for that reason the main driver of excess compensation (Carter et al., 2016). Several causes of excess compensation identified in prior research are for example, mutual back scratching or cronyism, unsolved agency problems and powerful CEOs (Bebchuk et al., 2002; Brick et al., 2006; Carter et al., 2016; Hill et al., 2016). However, research on excess compensation has proven to be difficult, as it failed to identify all causes of excess compensation (Hill et al., 2016).
One of the recent changes to CEO compensation contracts is the introduction of subjectivity to annual incentives (Golman & Bhatia, 2012). Subjectivity is about judgments based on personal impressions, feelings and opinions rather than on external facts like accounting numbers, which are often used to objectively measure performance (Bol, 2008). When
compensation is based on annual incentives and subjective performance measures it is called a discretionary bonus. Discretionary bonuses are payments that are not determined by a strict formulaic approach and reflect the subjective assessments of performance of the board (Ederhof, 2010; Höppe & Moers, 2011). While the responsibility of assessing contributions
5
and compensating lies with the board, this task is usually delegated to the compensation committee (Sun & Cahan, 2012).
Prior research identifies two ways in which subjectivity is used in CEO compensation
contracts. The most common way of introducing subjectivity to compensation contracts is by allowing the complete performance evaluation to be subjective (Gibbs et al., 2004; Murphy & Oyer, 2003; Prendergast, 1999). The other use of subjectivity is as a subjective adjustment to objective performance measures (Ederhof, 2010). This can be beneficial because events with a negative influence can emerge, while not being accounted for when designing the contract. The focus of this study is on discretionary bonuses that purely rely on subjectivity, as they contain the most information about the subjective performance evaluations of the
compensation committee.
Subjective performance measures are used in CEO compensation contracts because they offer several benefits identified in prior research. The optimal performance measure would reflect all contributions to firm performance and should use all information that is available, weigh this properly so that incentives are balanced appropriately across all dimensions of the job (Baker et al., 1994; Bol, 2008, Gibbs et al., 2004). However, for most employees and specifically for CEOs, it is very difficult to objectively measure all contributions to firm performance. While objective measures like accounting returns can be a reasonable representation of current CEO performance, they do not reflect the benefits of the growth opportunities identified or current strategic planning. Because the compensation committee is closer to the CEO and has more information about the actions of the CEO, they can
subjectively evaluate and reward for actions that would otherwise go unnoticed, or would not even have been undertaken, while being directed to increase firm value (Gibbs et al., 2004).
But, there are downsides to the use of subjective performance measures which are described in prior research. Using subjectivity in performance measurement can lead to significantly biased evaluations. Evaluations can be impaired because of a perceptual biases, opportunism such as effort aversion, or favoritism towards the evaluated person. Several reasons for this are the evaluator’s own perceptual biases and this is increased by a systematic basis due to interpersonal relationships, likeability and affect (Angelovski et al., 2016; Baker et al., 1988; Dai et al., 2018; Prendergast & Topel, 1996). As bad evaluations can result in confrontations, it is generally in the evaluators’ best interest to avoid providing ratings that may be perceived
6
as too low. Confrontations are costly for several reasons as illustrated by Bol et al. (2016). Confrontations are time consuming as evaluators need to make time to listen and respond to the reaction and potentially need to collect additional information to justify their subjective assessments. Bol et al. (2016) also show that evaluators experience psychological discomfort when asked to justify ratings that are considered to be too low, confrontations are thus psychologically painful. Evaluators are also concerned that confrontations could potentially cause personal relationships with the CEO to deteriorate, which could lead to increased tension in future interactions. Few individuals consider themselves to perform below average and this causes evaluators to be tempted to restrict their ratings to the upper part of the rating scale, which results in rating compression as the ratings of weaker performance will be inflated (Bol et al., 2016). The use of subjective measures may be causing compressed and more lenient evaluations and these tendencies are more pronounced when evaluators have strong relationships with the evaluated persons (Bol, 2011; Moers, 2005).
However, prior research did not investigate the relationship between the downsides of
subjective performance measures and the existence of unexplained excess CEO compensation before. Prior research has mainly focused on long-term equity compensation and has only focused on objective performance measures in relation with excess compensation. A potential reason for this is identified by Gibbs et al. (2004) and Höppe & Moers (2011) who mention that many interesting issues arising with the use of subjectivity in compensation contracts are not easily studied using traditional datasets. Because they raise behavioral issues, like trust, and conflicts about perceptions between principal and agent. Since discretionary bonuses are becoming more common in use, more firms are reporting them in their proxy statements together with the rationale for awarding them. This makes it possible to identify discretionary bonuses that are completely based on subjective evaluations and shed light on a potential driver of excess compensation.
The cornerstone of the CEO compensation literature is the agency theory. Based on the agency theory, CEOs are considered to be self-serving and concerned with maximizing their own wealth, so they will aim to maximize their compensation (Jensen & Mackling, 1976; Fama & Jensen, 1983). In a traditional agency-principal framework it is important to align interests with efficient contracts. This is a contract that maximizes shareholder value and CEO compensation, or in other words minimizes agency costs (Core et al., 2003).
7
This paper posits that it is important to investigate the relationship between discretionary bonuses and excess CEO compensation as they both have been identified in prior research, but their association has not been investigated before. Although discretionary bonuses do not account for the biggest fraction of the total compensation of CEOs, they reflect the subjective performance evaluations of the compensation committee. However, it is unclear if the
identified downsides of subjective performance measures are contributing to excess
compensation which in turn is detrimental to shareholder value. This leads this paper to study the potential of discretionary bonuses being a contributor to excess CEO compensation. The research question that will be guiding this research is:
Do discretionary bonuses contribute to excessive CEO compensation?
Because there are many variables at play in the relationship between discretionary bonuses and compensation, this paper will also consider two of the most influential moderating effects. This paper will attempts to find an answer whether this relationship is mitigated by effective compensation committees and strengthened by CEO influence. Effectiveness of compensation committees is important because they play an important role in the relationship between shareholders and CEOs. Influence of the CEO is important because CEOs are considered to act in their self-interest and they his position in the firm offers him opportunities to influence his own compensation.
To test these predictions this paper develops a first-stage OLS regression analysis to make a prediction of CEO total compensation which is later subtracted from actual CEO
compensation to result in excess CEO compensation. Excess CEO compensation is then used in a second-stage OLS regression analysis with discretionary bonuses and in a moderated OLS regression analysis with proxies for effective compensation committees and CEO influence.
The findings provide empirical evidence of the contribution of discretionary bonuses to excess CEO compensation. In addition to this this study also provides evidence of the mitigating effect of effective compensation committees and strengthening effect of CEO influence.
8
In order to provide an answer to the research question of this paper, it is essential to start with a review of the relevant literature on subjective performance measures and its relation to excess CEO compensation, together with the influence of the compensation committee and the CEO.
2. Literature Review and Hypothesis Development
2.1 CEO compensation
Cash compensation of S&P 500 CEOs has more than doubled over the three decades leading up to 1999 (Murphy, 1999). When equity compensation is also considered, total CEO
compensation has nearly quadrupled over this period (Murphy, 1999). Carter et al. (2016) illustrate that this trend not only increased, but increased at an increasing rate in the following decade. Prior research identified several explanations for the increase of CEO compensation. CEOs will receive more compensation as firms become more complex, for example because of increased globalization (Brick et al., 2006). As firms become more complex they require more talented CEOs who in turn demand greater compensation to manage increasing numbers of sales, employees, board meetings and business segments (Core et al., 2003; Dah & Frye, 2017).
Criticism of high levels of CEO compensation emerged and became a major political issue in the 1990s (Murphy, 1999). Presumably because CEO compensation started to exceed the level of around 25 times that of the average employee. However, the mere increase of CEO compensation over the last decades is no reason for concern per se. As mentioned before, compensation is only considered to be excessive when it cannot be explained by firm performance or economic factors. Some critics will claim that the problems with CEO compensation are so pervasive that most CEOs receive excess compensation (Hill et al., 2016).
The excessive part of CEO compensation received attention in the academic literature. Several factors explaining inappropriate levels of CEO compensation have been identified, like for example: type of ownership, firm size, governance, reporting practices, external influences, luck and benchmarking.
9
For firms with dispersed ownership, it is shown that there are greater possibilities of the CEO deriving non-pecuniary interests, differences in shareholder and management risk profiles and conflicts about decision-making time horizons (Chalmers et al., 2006). This jeopardizes the fundamental obligation of the CEO to act in the interest of shareholders.
While it is no question that firm size and the resulting complexity is a significant economic determinant of CEO compensation, it remains uncertain whether large firms appropriately compensate for this (Hill et al., 2016). A possible explanation is the existence of
compensation practices that reward CEOs for increasing firm size, while it is not in the best interests of shareholders (Bebchuk & Fried, 2006).
Inappropriate levels of CEO compensation are also identified for firms with poor
performance. Brick et al. (2006) found this to be caused by the mutual back scratching or cronyism between the board and CEO. Core et al. (1999) illustrate that firms with less effective governance have higher unexplained compensation levels.
Governance problems are also illustrated by concerns of the market caused by excess CEO compensation and are found to blur fair investment judgments. Investment decisions such as mergers and acquisitions are found to be driven by CEOs’ personal objectives, such as maximizing personal wealth and benefits, at the expense of shareholder value (Feito-Ruiz & Renneboog, 2017). While equity-based compensation is set up to lead to value maximizing decisions, markets negatively react to mergers and acquisitions when CEOs are found to receive excess compensation (Feito-Ruiz & Renneboog, 2017).
Hooghiemstra et al. (2017) illustrate that less readable remuneration reports are causing a reduction in say-on-pay voting dissent and are associated with excess CEO compensation. Say-on-pay votes are important to shareholders as they offer an opportunity to express their support for, or opposition against CEO compensation. High levels of voting dissent are found to frequently accompany considerable negative publicity. Ignoring it is potentially costly to the compensation committee as it is likely to subject the quality of supervision to public criticism and it is directly undermining the reputation of the committee members in the labor market. Potential reputational damage explains why compensation committees are responsive to the mere threat of voting dissent. But as it is found, this can be limited by obfuscating the reporting on CEO compensation.
10
Bertrand and Mullainathan (2001) illustrate how compensation is not linked to performance and CEOs are getting paid for luck, influenced by external factors. In a similar fashion, Bebchuk et al. (2008) illustrate how powerful CEOs are more likely to be rewarded for luck due to industry wide shocks and happen to receive “lucky” options grants at the lowest price of the month.
The decisions of the compensation committee are particularly important in setting CEO compensation. But their role is influenced as they take advice from various sources (Conyon, 2013). Employees from the HR department are supplying inside information, which might be partial to the incumbent CEO. Advice is also provided from outside the firm as compensation committees take advice from compensation consultants. Their advice might be compromised as their incentives are depending on it. Compensation consultants that recommend low CEO compensation may find their advising contracts not getting renewed. Compensation
consultants often also perform other activities for the firm that might be put at risk if they recommend low levels of CEO compensation.
Firms have a tendency to benchmark CEO compensation to that of other CEOs. While some use of benchmarking is consistent with competitive compensation, it is also a way of CEOs increasing their compensation as they benchmark themselves to highly paid CEOs (Cremers & Grinstein, 2014). When a compensation committee finds that its CEO compensation reposes in the fourth quartile compared to other CEOs, it is argued that CEO compensation currently is too low and it will be raised to, for example, the second quartile. This way compensation for their own CEO is increased, but another CEO is dropped into the fourth quartile and so the cycle repeats. This is allowed by compensation committees that typically tend to fall in love with their CEOs, until something big goes wrong (Bogle, 2008). CEO compensation should be related to performance that they can influence and not to how much they earn compared to other CEOs (Cremers & Grinstein, 2014).
Firms paying excess amounts of CEO compensation commonly do this through compensation contract design by setting less challenging targets which is detrimental to shareholder value (Abernethy et al., 2014). Less challenging targets are considered to be overly concerned with short-term performance, making CEOs shortsighted. The choice of performance targets is also contributing to this by setting easy obtainable earnings per share targets or stock market prices (Abernethy et al., 2014). Another suggestion of a weak relationship between performance and
11
compensation is provided by Bebchuk and Fried (2004). They suggest that the relationship ceases to exist when objective accounting performance is poor or declining. CEOs are found to receive increased compensation for good earnings performance, but they lack penalties for poor earnings performance.
These examples of inappropriate levels of CEO compensation have one thing in common, they all relate to objective performance measures or external factors and the way of compensating is with equity and in particular with stock options (Conyon, 2013). But the findings do not explain all the deviations from the appropriate level of CEO compensation. Not all determinants of excess CEO compensation have been identified since a complete performance measure also includes subjective performance measures. The importance of subjective performance measures is illustrated by their common use in practice and similar to any type of compensation they presumably have incentivizing benefits, but also have a contribution to excess compensation. While the incentivizing benefits of subjective
performance measures have received attention in academic research (e.g. Bol, 2008), their relationship with excess compensation has received a lack thereof.
This offers an interesting research opportunity and leads this study to expect that subjective performance measures also have a potential influence on excess CEO compensation. The next section will further elaborate what subjective performance measures are, why they are used and why they might contribute to excess CEO compensation.
2.2 Performance measures
Within a certain framework the compensation committee is responsible for setting performance targets, choosing performance measures, either objective or subjective, and determining the weights of the measures. In practice however, compensation plans are never perfect. This is the result of the fact that not all effort is objectively measureable.
Imperfections of objective performance measures are caused by noise. Noise is an
uncontrollable random event which increases the uncertainty the agent faces (Gibbs et al., 2004). Controllable events also impact the value of the firm, but actions of the CEO do have an effect. They can even be random, but as long as the CEO can react to them by mitigating the negative effects, while exploiting the positive effects, they will still have a positive effect on firm value. Baker et al. (1994) show in their analytical model that compensation contracts which contain noisy objective measures can be improved if the principal has knowledge and
12
information about actual noise realizations and if he uses this to subjectively determine compensation. The beneficial risk reduction is coming from the possibility to ex post make adjustments according to the received noise signal. The noise signal can be used by the principal to make discretionary adjustments in compensation to filter out uncontrollable events (Höppe & Moers, 2011).
Gibbs et al. (2004) illustrate a positive effect of subjectivity in performance measurement. When discretionary bonuses are awarded, they increase pay satisfaction, productivity and profitability. The use of subjectivity in compensation contracts will only have positive effects on the alignment of incentives when there is adequate trust between the principal and agent. Because it involves discretion it only works if the principal is able to make fair, unbiased judgments and the agent needs to accept these judgments and is not trying to influence the principal in any way.
When compensation is based on incomplete performance measures it makes rational CEOs ignore the dimensions that are not measured. Incomplete measures lead to distorted incentives (Bol, 2008). Objective performance measures are usually based on accounting measures, which are backward looking by nature. This ignores the long-term effect of actions which should be included in a complete performance measure. This could lead to CEOs preferring to take actions that favor the short term that could potentially destroy shareholder value in the long term. This highlights where the introduction of subjectivity to performance measures can be beneficial because it can account for good effort that is not easily quantifiable (Bol, 2008; Murphy & Oyer, 2003). This prevents the agent from purely focusing on objectively
measured tasks, which makes incentives distorted.
2.3 Downsides to subjective performance measures
As mentioned before, compensation plans are not perfect. While subjective performance measures contribute to a more complete performance measure by coping with the noise in objective performance measures, they also have their own shortcomings. Downsides of subjectivity are illustrated by for example, Golman and Bhatia (2012) and Moers (2005). Subjective performance measures suffer from severe leniency effects. These surface when compensation committees are mild in their assessments and in turn cause inflated ratings.
13
Inflated ratings occur in many different situations, but this effect is worse when used to determine CEO compensation or when there is a strong relationship between the CEO and the compensation committee. The strong relationship is assumed because the committee works with the CEO on a regular basis in either a monitoring or advising function. An assumption is that it can simply be unpleasant to give negative feedback (Bol, 2008). While negative
feedback can be accurate it can damage personal relationships and this can lead to discussion and criticism (Bol, 2008). Research by Varma et al. (1996) shows that this causes
compensation committees to show defensive behavior. This is of negative influence on the accuracy of the evaluations and is detrimental to appropriate compensation and promotion decisions.
Compensation committees do not necessarily desire inflated ratings, but when noisy signals are inevitable, they are still introduced to counteract the inherent imprecision of the
performance signal. When performance signals are noisy and there is a strong aversion to unfair low ratings compared to high ratings this can result in compensation committees inflating their performance ratings (Golman & Bhatia, 2012). Asymmetric fairness
considerations inflate performance evaluations as illustrated by Bol and Smith (2011). The strong relationships between the compensation committee and CEO increases the
performance evaluation bias.
Subjective performance measures also show a centrality bias with evaluators compressing ratings so that they differ little from the norm. Altogether this can cause situations where everyone is rated above average and this results in excess CEO compensation.
Another downside is that it is possible for subjective evaluators to develop favoritism (Prendergast & Topel, 1996). In almost every case the compensation committee that holds information necessary to make subjective evaluations of performance is not the residual claimant of the benefits of an accurate evaluation. In other words, since the evaluator, the compensation committee, cannot “pocket” the savings, the temptation to renege on promised compensation is not the cost savings of the compensation. In this context, favoritism is compensating preferred, rather than deserved actions of CEOs. Shirking of evaluation efforts is promising random or equal compensation, rather than in proportion to assessed
performance. In their model, Prendergast and Topel (1996) analyzed favoritism in subjective performance evaluations. They increased the importance of subjectivity and this lead to a
14
reduction in the accuracy of the evaluations of supervisors. Supervisors will indulge in more favoritism when the stakes to the CEO are higher and this results in CEOs “charging extra” for the increased arbitrariness of compensation by demanding higher expected compensation. These factors are limiting the use and informativeness of subjective performance measures in compensation contracts.
2.4 Agency Problems
Most research on executive compensation has been rooted in the agency theory. In short it is based on a moral hazard dilemma that exists when one person is able to make decisions on behalf of another person and an adverse selection dilemma because the availability of
information is different for both parties (Jensen & Mackling; 1976). In the economic field this dilemma occurs when a CEO (agent) makes decisions on behalf of a shareholder (principal). CEO compensation plans are designed to align the interests of risk-averse self-interested CEOs with those of shareholders (Murphy, 1999). In a typical situation CEOs are assumed to take actions when they produce shareholder value and result in compensation. In this situation shareholders precisely know what actions CEOs need to take, but they cannot directly observe them. The best compensation contract is assumed to maximize the risk-neutral shareholders’ objective of maximizing shareholder value while offering enough incentives for the CEO to choose preferred shareholders’ actions over maximizing his self-interests. In other words, the fundamental problem is not to get the CEO to work harder, but rather to get him to choose actions that increase shareholder value rather than decrease it (Murphy, 1999). Shareholder value is increased when investments are made in positive net present value projects or when resources are diverted from negative net present value projects.
As the tasks of a CEO become more complex, the set of potential actions that he can perform to increase shareholder value is also expanded. When CEOs can choose from an unlimited amount of actions, incremental informativeness can distort incentives. Some examples include taking actions that decrease performance of reference groups, sandbagging target setting processes to easier achieve targets, shifting accounting returns by accelerating or delaying revenues and costs across periods, artificially inflating or deflated reported earnings by
making accounting choices or making cuts in investment choices to increase short term profits at the expense or long term profits (Murphy, 1999).
15
In sum, excess CEO compensation is caused by agency problems. To mitigate this,
compensation committees set up compensation contracts in order to align their interests with those of the CEO. Alignment is achieved by linking performance to compensation. The level of performance needs to be monitored and this is most commonly done with objective performance measures. However, as these are imperfect by the noise they contain, they are supplemented by subjective performance measures that act as discretionary adjustments in compensation to filter out uncontrollable events. Compensating this way is done with discretionary bonuses. While the supplementation of subjective performance measures has benefits, it is evident that these are not perfect and can result in inflated ratings and
favoritism. This leads this paper to predict that the use of subjective performance measures is possibly contributing to excess CEO compensation. More formally, this results in the first hypothesis:
H1: Discretionary bonuses are positively associated with excess CEO compensation
As mentioned before, the existence of excess CEO compensation is caused by agency
problems. These problems are derived from the relationship between principals and agents. In the ideal relationship compensation committees can balance different interests while
remaining impartial and CEOs do not act in their self-interest. In other words, compensation committees can operate effectively and CEOs do not use their power to benefit their self-interest. In the next section, these two concepts will be further discussed.
2.5 Effective Compensation Committees
The ideal role of the compensation committee is to set pay levels and programs, define and enforce the compensation strategy and to monitor the process (Murphy, 1999). When designing compensation contracts, the compensation committee has several goals in mind. Highly skilled CEOs need to be attracted, retained, provided with incentives to exert sufficient effort, while keeping the shareholders’ interests in mind and their behavior needs to be
monitored (Bebchuk et al., 2002; Murphy, 1999). The compensation committee must also be prepared to counter clear violations of shareholder interests, which in most cases means “pushing back” on seemingly excessive pay recommendations as CEOs, like any other individual, prefer more to less (Murphy, 1999). Based on the agency theory, these tasks are only effectively performed when they are in the best interest of the shareholder. Collusion
16
with the CEO is dependent on whether the interests of the compensation committee are more tightly related with those of shareholders or the CEO.
This study expects compensation committees to operate at the right trade-off between encouraging the agent to exert good effort while minimizing the risk that good effort is not observed (Bol, 2008). To find the right trade-off the principal is encouraged to capture the efforts of the CEO as good as possible. Effective compensation committees can balance these different interests while maintaining their impartial position. The following concepts of effectiveness will be discussed below: dependence, diversity and teamwork.
In previous literature it is argued that directors can be dependent on the CEO and can form close alignments with the interests of the CEO and form coalitions that entrench themselves at the cost of shareholder interests (Conyon & Peck, 1998). The result of this is that they may have a general propensity to offer the CEO generous compensation terms. Effective
compensation committees are considered to consist of directors that are considered to be free from personal conflicts of interest and should be able to exercise their independent judgment when there is a disagreement with the CEO. Dependency is detrimental to effectiveness because it limits the compensation committee to remain impartial and balance different interests between shareholders and the CEO.
As compensation committees consist of multiple members, researchers also studied the interpersonal interactions among group members. They found that these may induce changes in their individual preference and behavior, which in turn influence the group’s decision making (McGarth, 1984). More important, they found that group decisions improve in quality when they are drawn from diverse perspectives and a wide range of experiences (Kosnik, 1990). In her study, she illustrates that diversity in members is a critical factor in the
formation of a well-balanced and active compensation committee by reducing the probability of complacency. Similarly, Baker et al. (1994) illustrate in their model that it is the private information the CEO holds that affects the subjective performance assessment. When there is only one compensation committee member who has a distortion in its subjective evaluation, the CEO can try to exploit this and adjust performance solely so that his actions are perceived favorably by this specific member. When the CEO tries to exploit this distortion by altering his actions, he is serving his self-interest and the alignment of goals between the agents and principals is distorted. CEOs who are concerned with their self-interest will participate in
17
influencing activities as long as their influencing efforts lead to more personal benefits than efforts devoted to value enhancing activities (Bol, 2008). This situation can be avoided by combining the subjective performance evaluations of multiple members, assuming they do not share the same distortion in their subjective evaluation. Diverse compensation committees are less likely to be co-opted by the CEO, have greater bargaining power to limit the CEO in his influence on compensation decisions and are less likely to offer the CEO a compensation plan with higher compensation and lower incentives. Based on this, diversity is beneficial to compensation committee effectiveness because it allows the compensation committee to remain impartial and balance different interests between shareholders and the CEO.
For compensation committees to make effective decisions, they need to devote a significant amount of time, resources and effort (e.g. committee members and their knowledge base) (Laksmana, 2008; Murphy & Oyer, 2003). Previous studies show that the effectiveness is achieved by the ability to distribute workloads and assignments. Committees achieve this by planning regularly and well-organized meetings (Conger et al. 1998). In other words,
compensation committee effectiveness benefits from teamwork. At these meetings the
compensation committee engages in substantive discussions and sets up follow-up reviews for after the meetings (Laksmana, 2008). Based on this, teamwork is beneficial to compensation committee effectiveness because it allows the compensation committee to better balance different interests between shareholders and the CEO by the ability to distribute workloads and assignments.
In sum, as the responsibility of designing compensation contracts is delegated to the compensation committee, this leads me to believe that the effective functioning of the compensation committee plays an important role in the relationship between discretionary bonuses and excess CEO compensation. Effective compensation committees can balance different interests of shareholders and the CEO, while maintaining an impartial position. This is achieved by avoiding dependency of the CEO, being diverse and working in a team. The result is a better monitoring function and as compensation committees balance the different interests of shareholders and the CEO, they are not influenced by the self-interests of the CEO. Based on these examples and the agency problems, this leads this study to predict that the compensation committee has a mitigating effect on the relationship between discretionary bonuses and excess compensation. More formally, this results in the second hypothesis:
18
H2: The effective functioning of the compensation committee mitigates the positive relationship between discretionary bonuses and excessive CEO compensation
2.6 CEO influence
The CEO acts as the central decision maker of the firm, he represents the center of many communication lines (Barkema & Pennings, 1998). This gives him an informational
advantage with respect to the compensation committee. CEOs also play a central role in the recruitment, selection and compensation of board members and they can influence the agenda of board meetings. These directors are essentially “hired” by the CEO (Ittner et al., 1997). This makes the functioning of the compensation committee less effective as these directors may be unwilling to take adversarial positions to the CEO (Ittner et al., 1997). This reinforces his central position in the communication network and makes the board and compensation committee less independent so they lose sight of the shareholder values. All of this is possible because the board functions on the basis of information provided by the CEO.
In contrast to the role of the compensation committee it is assumed that without incentives to act in the creation of shareholder value, CEOs will act in their own self-interest. As CEOs form the center of the firm they are in a special position where they can influence their compensation and extract rents. Rent extraction is similar to excess compensation and is caused by the tendency of compensation committees for passivity, their dependence on the CEO for information and their lack of exposure to share returns (Bebchuk et al., 2002). This causes sub-optimal incentives and is detrimental to shareholder value.
These are examples of the adverse selection dilemma and they need to be resolved or they can be exploited. For example, it can lead to lower setting of targets, which the CEO knows are easily obtainable and therefore not incentivizing additional effort. He can cut investments to increase short term profits when certain targets might not be met on time, sacrificing long term profits (Murphy, 1999). This all results in the preference of the CEOs self-interest over that of shareholder value, which causes an imbalance in the principal-agent relationship. This eventually leads to both the principal and agent being worse off when compared to a situation where interests are balanced (Jensen & Mackling; 1976).
Because there is no contract that perfectly aligns the interests of shareholders and CEOs, the best contract from a shareholder’s perspective is one that minimizes agency costs (Bebchuk et
19
al., 2002). Deviations from the best contract can be the result of several causes. The compensation committee can be captured, subject to influence or sympathetic to the CEO (Bebchuk et al., 2002). As predicted by the agency theory, CEOs are assumed to want more total pay, but less variance in their compensation because they are risk averse. When CEOs gain the upper hand over the compensation committee, the CEO will use this influence to gain a greater level of compensation. The following concepts of CEO influence will be discussed below: time, share holdings and fraction of top executive compensation.
An indicator of the concept of CEO influence is time. With the CEO playing a central role in recruitment and compensation, the CEO and compensation committee members are expected to develop social exchange relations over time. CEOs who longer operate in their current function are expected to have accumulated more influential power and to be more successful in influencing their compensation (Barkema & Pennings, 1998). This influence is also illustrated by Lippert and Porter (1997), who find evidence of CEOs using their influence to circumvent monitoring and incentive alignment mechanisms. Combs et al. (2007) argue that time is the key ingredient in the process of building influential power. Goals mentioned are influence in the design of their compensation and golden parachutes, all obtained by
leveraging their position to pursue self-interests irrespective to those of shareholders (Combs et al., 2007).
Two other indicators of the concept of CEO influence are found in actual CEO compensation and CEO shareholdings. Previous research by Ittner et al. (1997) and Murphy (1999)
illustrates that CEOs with large shareholdings compared to their salary are an indicator of how much influence a CEO has on compensation practices. In addition to this, prior research of Bebchuk et al. (2011) and Correa and Lel (2016) finds that the fraction of total
compensation of the top five highest paid executives captured by the CEO is a reflection of the relative importance and influence of the CEO. It also reflects the extent to which he is able to extract rents and higher odds of the CEO receiving lucky option grants at the lowest price of the month. Taken together it is an indicator of the aforementioned agency problems and thus an indicator of excess CEO compensation.
In sum, the central position of the CEO in the firm gives him an informational advantage. This leads this study to believe that the influence of the CEO is playing an important role in the relationship between discretionary bonuses and excess CEO compensation. His
self-20
interest can make him exploit this to influence his compensation and extract rents. His influence makes the CEO the first among equals as he enjoys a privileged position (Barkema & Pennings, 1998). Influencing subjective performance measures are an excellent vehicle for increasing compensation and rent extraction as they are not audited and reliant on
interpretation (Bol, 2008). This results in sub-optimal incentives and is detrimental to shareholder value. Indicators to identify influential CEOs are present in prior research and illustrate how CEO influence is increasing over time as compensation committees and the CEO develop social exchange relations. Influence is also indicated by large shareholdings compared to salary and as the CEOs fraction of total top five executive compensation. These examples and agency problems lead this study to predict that CEO influence strengthens the relationship between discretionary bonuses and excess CEO compensation. More formally, this results in the third hypothesis:
H3: The influence of the CEO strengthens the relationship between discretionary bonuses and excessive CEO compensation
3. Method Section
In the next section, this paper will discuss all variables used, how they are measured, from where the data is collected and what kind of analysis is used.
The sample is constructed from firms listed on the S&P 500 and the data is obtained from several sources. Data will be hand collected from proxy statements provided by S&P 500 firms and from several databases like Compustat, CRSP, BoardEx and GMI. All the data will be for the most recent year available, 2015. The preference for the most recent year is because firms are more commonly using discretionary bonuses in their annual compensation and using the most recent year assures the availability of more data. Data collection started with
obtaining the required data from databases. This resulted in a total sample of 434
observations. As firms are not required to report all the information needed for this study in their proxy statements, this is limiting the sample size. Time restrictions required this study to make a random selection of the S&P 500 firms for which to hand collect data. The random selection eliminated 287 observations and resulted in a sample of 147 observations for which the data required for this study was hand collected. However, as 2 firms did not report the
21
number of members on the compensation committee they were excluded from the sample. Similarly, 3 firms did not report how many times the compensation committee met, so they were excluded from the sample. The proxy statements of 17 firms did not provide information about the awarding of discretionary bonuses so they were excluded from the sample. This results in a final sample of 125 observations.
3.1 Independent variable
Data on the awarding of discretionary bonuses (DISCB), is obtained from the proxy
statements provided by firms. Discretionary bonuses are paid in cash and are included in the non-equity annual incentives of CEOs. The focus is on discretionary bonuses that purely rely on subjectivity, as they contain the most information about the subjective performance evaluations of the compensation committee. To be included in the sample the discretionary bonuses needs to be based on a subjective performance objective and subjectively weighted. This is in contrast with a subjective adjustment to the weighting of performance of an
objective performance measure. Discretionary bonuses included in the sample are reported as a multiplier of salary or a target and are set at the discretion of the compensation committee. An illustration of how a discretionary bonus is reported in proxy statements and how it is calculated to be included in the sample of this study is shown in Appendix A. As some firms did not report a discretionary bonus or reported a discretionary bonus as a subjective
adjustment to an objective performance measures, they were excluded from the sample. Discretionary bonuses are winsorized to transform the statistics and reduce the possibility of outliers. After the variable is transformed the natural logarithm is used in the regression analysis. A summary of all variable definitions and descriptions is shown in Appendix B.
3.2 Dependent variable
This paper assumes, like prior literature, that the level of compensation of a CEO is related to firm complexity and risk (Brick et al., 2006). Therefore, proxies for complexity such as size, investment opportunities, stock return and return on assets should play an important role in determining the level of CEO compensation. Prior studies found firm size, investment opportunities and stock return to be determinants of CEO compensation (Chalmers et al., 2006; Core et al., 1999). Firms who are either larger, have the availability of more investment opportunities or have large stock returns are found to demand higher quality CEOs and are willing to pay for such quality. Firm size (SIZE) is measured as the natural logarithm of total assets. Investment opportunities (INVOP) are measured as the market-to-book ratio of equity.
22
Stock return (RET) is measured as the annual stock market return. Prior literature found firm performance to be associated with CEO compensation, suggesting a positive
pay-for-performance link and is often measured by return on assets (Chalmers et al., 2006). Return on assets (ROA15) is measured as EBITDA divided by total assets. CEOs are also found to be receiving more compensation when they are operating in riskier environments (Core et al., 1999). Firm risk (RISK) is measured as the standard deviation of daily stock market return excluding dividends. All data on determinants of CEO compensation is obtained from the Compustat and CRSP databases.
As the dependent variable, excess CEO compensation, is not directly observable, a two-step procedure used in prior literature is followed (e.g. Brick et al., 2006; Chalmers et al., 2006; Core et al., 1999) to define excess compensation (EXCESS) as the difference between observed compensation levels (CEOTOT) and expected compensation levels (EXPCOM) after controlling for firm complexity and risk. To model expected CEO compensation, a first-stage OLS regression is developed where the dependent variable is actual total CEO
compensation, while the independent variables include proxies of firm complexity and risk. Data about actual total CEO compensation is obtained from compensation summaries from the proxy statements and is winsorized to transform the statistics and reduce the possibility of outliers. After the variable is transformed the natural logarithm is used in an OLS regression analysis. Excess compensation is then calculated as the difference between actual total CEO compensation and expected total CEO compensation. Discretionary bonuses are then used in a second-stage OLS regression with excess CEO compensation. The formulas below illustrate the first-stage regression and excess CEO compensation calculation.
Expected CEO total compensation = α0 + β1 SIZE + β2 INVOP + β3 RET + β4 RISK + β5 ROA15 +
ɛ
Excess CEO compensation = CEOTOT – EXPCOM
3.3 Moderating variables
The second prediction of this paper is that the effective functioning of compensation committees mitigates the relationship between discretionary bonuses and excess CEO compensation. This prediction is based on the idea that effective compensation committees
23
can balance different interests of shareholders and the CEO, while maintaining an impartial position. This is achieved by avoiding dependency of the CEO, being diverse and working in a team. The result is a better monitoring function and as compensation committees balance the different interests of shareholders and the CEO, they are not influenced by the self-interests of the CEO.
This paper uses several proxies for the effective functioning of the compensation committee. The proxy used for independence is the fraction of outsiders (COMOUT) and is measured as the amount of outside directors divided by the total amount of directors (Conyon & Peck, 1998). Effective compensation committees will prefer outside directors on their committee to maintain independence and function effectively. Data on the fraction of outsiders is obtained from the GMI database. The proxy used for diversity is the size of the compensation
committee (COMSIZE) and is measured as the total amount of directors sitting on the compensation committee (Bol, 2008; Kosnik, 1990). Prior research suggests that
compensation committees with more members are more diversified and thus more effective (Bol, 2008). Data on the size of the compensation committee is obtained from the BoardEx database. The third and last proxy of the effective functioning compensation committees is the amount of meetings (COMMEE) of the compensation committee and is used as a proxy for teamwork (Conger et al., 1998; Laksmana, 2008). A few studies suggest that the frequency of committee meetings is a relatively good proxy for board diligence (Laksmana, 2008). As compensation committees operate in a group and have the ability to set meetings, this is achieved by setting regular meetings. A sufficient number of well-organized meetings will increase committee effectiveness (Conger et al. 1998). This is measured as the total amount of times the compensation committee met and data is obtained from the proxy statements.
The third prediction of his paper is that influence of the CEO strengthens the relationship between discretionary bonuses and excess CEO compensation. This prediction is based on his privileged position in the firm giving him an informational advantage. His self-interest can make him exploit this to influence his compensation and to extract rents. Influencing
subjective performance measures are an excellent vehicle for this as they are not audited and reliant on interpretation. Indicators of CEO influence that are identified are the increase over time, large shareholdings compared to salary and the CEOs fraction of top five executive compensation.
24
Several proxies are used for the influence of the CEO. The proxy for the increase of influence over time is tenure of the CEO (TENURE) and is measured as number of years the CEO is in place as the CEO of the firm (Barkema & Pennings, 1998; Combs et al., 2007; Lippert & Porter, 1997). Data on CEO tenure is obtained from BoardEx database. The proxy for large shareholdings compared to salary is the CEO shares multiplier (MULTI) and is measured as the multiplier of the value of CEO shareholdings and base salary (Ittner et al., 1997; Murphy, 1999). Data on CEO shares multiplier is obtained from the GMI database. The proxy for the fraction of CEO compensation of top five executives is the CEO pay slice (PSLICE) and is measured as the share of total CEO compensation of total compensation of the top five
highest paid executives (Bebchuk et al., 2011; Correa & Lel, 2016) Data on the CEO pay slice is obtained from the proxy statements. The second and third hypotheses are analyzed with a moderated OLS regression.
The intention of this study is not focused on how these individual proxies relate to the
relationship between discretionary bonuses and excess compensation, but it is focused on how their combination is used as a proxy for effective compensation committees and CEO
influence. For that reason the individual proxies will be combined into one variable. To analyze the moderating effect of an effective compensation committee, the three individual proxies (COMSIZE, COMMEE and COMOUT) are standardized by subtracting the mean from each individual observation and then by dividing it by the standard deviation. In the next step the values are added and combined into a new variable: COMEFF. Similarly, the
individual variables of CEO influence (TENURE, MULTI and PSLICE) are standardized and combined into a new variable: CEOINFL.
3.4 Control variables
The implications of agency problems and the resulting excess CEO compensation are studied before. For example, Carter et al. (2016) assumes that excess CEO compensation is a
reflection of agency problems and thus negatively related to future firm performance. This study follows the assumption that worse future performance is a result of current year excess compensation and includes worse future firm performance as a control variable for excess CEO compensation. A proxy used to measure worse future firm performance is return on assets for year t+1 and is measured as EBITDA divided by total assets for year 2016. Similar to the ROA for 2015, data obtained for the ROA for 2016 is obtained from the Compustat database. However, as it is an indicator of worse future performance it will be included as a
25
dummy variable and is equal to 1 if the ROA for 2016 is less than the ROA for 2015 and is equal to 0 otherwise.
This paper includes industry dummies to control for differences between industries. A distinction is made between the following industries: (1) mining, (2) manufacturing, (3) transportation and public utilities, (4) wholesale and retail trade, (5) finance, insurance, real estate and (6) services. SIC codes are used to differentiate between industries. The distinction between six industries is sufficient because it represents an appropriate division of industries, preventing groups with only a few observations because of the relatively small amount of total observations.
The following models are used to test the hypotheses:
Hypothesis 1:
Excess CEO compensation = α0 + β1 DISCB + β2 ROA16 + β3 ∑IND +
ɛ
Hypothesis 2:
Excess CEO compensation = α0 + β1 DISCB + β2 DISCB * COMEFF + β3 ROA16 + β4 ∑IND +
ɛ
Hypothesis 3:
Excess CEO compensation = α0 + β1 DISCB + β2 DISCB * CEOINFL + β3 ROA16 + β4 ∑IND +
ɛ
4. Results
4.1 Descriptive statistics
Descriptive statistics for the first-stage regression, independent and control variables are presented in Table 1. The following observations can be made. The average CEO in the sample of this study received a discretionary bonus of $891.696 and his total compensation amounted to $13.428.465. From his total compensation, $1.004.968 is considered to be excessive on average. The smallest discretionary bonus awarded was worth $51.000 while the
26
largest was worth $4.511.875. The CEO who was compensated the least total compensation received $1.338.110 while the most compensated CEO received $47.462.526. Regarding excess CEO compensation, not all CEOs seem to be compensated excessively.
Descriptive statistics indicate that the average firm in the sample had a compensation committee that consisted for 88% out of independent outside members, had at least 4
members considering diversity and held at least 5 meetings to work as a team. For CEOs the statistics illustrate that the average CEO has been CEO of the firm and has been developing relationships for 4 years and 7 months. His total compensation accounted for 43% of the total compensation paid to the top 5 executives and his shareholdings have a value of at least 19 base salaries. The return on assets for 2016 is on average for 54% of the firms lower compared to the return on assets for 2015.
Table 1. Descriptive statistics
Variable name Mean Std dev Min Max
DISCB ($) 891.696,89 934.899,86 51.000,00 4.511.875,00 CEOTOT ($) 13.428.465,53 7.429.450,38 1.338.110,00 47.462.526,00 EXCESS ($) 1.004.968,31 6.893.715,62 -7.731.120,19 26.656.346,69 SIZE 16,94 1,32 14,57 21,49 INVOP 4,99 5,89 ,46 39,62 RET -,0214 ,2256 -,5549 ,5179 RISK ,0161 ,0044 ,0096 ,0322 ROA15 ,1324 ,0863 -,1158 ,4486 COMSIZE 4,88 1,38 3 9 COMMEE 5,90 1,94 2 16 COMOUT ,8826 ,0507 ,6667 1 TENURE 4,63 4,03 ,2 21,9 MULTI 19,91 25,89 1,07 150,02 PSLICE ,4329 ,0879 ,1953 ,6781 ROA16 ,54 ,50 0 1
27
4.2 Correlation analysis
Pearson correlations between variables are presented in Table 2. The highest level of correlation found is -0,473 between ROA15 and SIZE and this indicates a moderate
correlation. As a strong correlation is indicated by a value larger than 0.5 or smaller than -0.5, multicollinearity should not cause any difficulties when the variables are used in the
regression analysis.
Table 2. Correlation matrix
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) 1. DISCB 1 2. SIZE ,350** 1 3. INVOP ,064 -,189* 1 4. RET ,051 -,121 ,160 1 5. RISK -,179* -,188* -,084 -,324** 1 6. ROA15 ,042 -,473** ,402** -,185* -,027 1 7. COMSIZE ,031 ,171 ,112 -,002 -,070 -,130 1 8. COMMEE ,325** ,271** -,059 ,027 ,029 -,129 -,016 1 9. COMOUT -,021 ,154 ,040 ,028 ,019 -,030 ,287** -,007 1 10. TENURE ,199* -,091 -,008 ,068 -,118 ,060 -,065 ,118 -,259** 1 11. MULTI ,240** -,090 ,110 ,222* ,030 ,236** -,082 ,002 -,109 ,401** 1 12. PSLICE ,137 -,303** ,007 ,021 -,042 ,195* ,012 ,029 ,105 ,131 ,102 1 13. ROA16 -,001 ,020 -,014 -,151 ,039 ,176* -,057 -,029 ,030 ,041 ,178* -,002 1 **. Correlation is significant at the 0.01 level (2-tailed).
*. Correlation is significant at the 0.05 level (2-tailed).
Variable definitions and descriptions are provided in appendix B
4.3 First-stage regression
This study is focused on the relationship between discretionary bonuses and excess CEO compensation. The dependent variable is excess CEO compensation and as it is not directly observable a two-step procedure is used. The first step involves developing an estimate of CEO compensation based on the determinants: firm size, investment opportunities, stock return, risk and return on assets. This makes it possible to estimate individual CEO
28
compensation is then compared to the actual reported CEO compensation and the difference is considered to be excess compensation, which is used as the dependent variable in the second-stage OLS regression. The results of the first-second-stage OLS regression are presented in Table 3. The model explains 28.8% of the variance in total CEO compensation and is line with previous studies (Brick et al., 2006; Core et al., 1999).
In the next step, the intercept and coefficients are used to create a formula that predicts CEO total compensation for the firm individual determinants. The resulting formula is shown below.
Expected CEO total compensation = 5.210 + 0.103 SIZE + 0.001 INVOP – 0.022 RET – 1.247 RISK + 1.017 ROA15
For each firm individually, expected CEO total compensation is then subtracted from actual CEO total compensation to determine excess CEO compensation, the dependent variable for the second-stage OLS regression.
Table 3. First-stage regression analysis TOTCOMP
Variable name Coefficient estimate Standard error
Constant 5.210*** 0.329 SIZE 0.103*** 0.017 INVOP 0.001 0.004 RET -0.022 0.091 RISK -1.247 4.721 ROA15 1.017*** 0.267 n 125 F-statistics F = 8.560; p = .000 0.234 Adjusted R²
*** Significance at the 0.1% level
The intercept is presented on a logarithmic scale
29
4.4 Second-stage regression and hypothesis testing
The first-stage regression is performed to result in an excess CEO compensation variable. In the second-stage regression excess CEO compensation is used as the dependent variable. The second-stage regression is performed for each hypothesis individually. The findings related to the three hypotheses are presented in Table 4.
The first hypothesis states that discretionary bonuses are positively associated with excess CEO compensation. Panel A of Table 4 presents the findings from the first regression. The discretionary bonus coefficient indicates that discretionary bonuses are significantly and positively associated with excess CEO compensation (β = 0.334 and p < 0.01). The positive and significant discretionary bonus coefficient supports the hypothesis and based on that there is evidence to conclude that discretionary bonuses are positively associated with excess CEO compensation.
Panel B of Table 4 presents the findings regarding the second hypothesis: the mitigating effect of the effective functioning of the compensation committee on the relationship between discretionary bonuses and excess CEO compensation. The coefficient indicates that effective functioning of the compensation committee is significantly and negatively associated with the relationship between discretionary bonuses and excess CEO compensation (β = -0.040 and p = 0.049). These findings support the hypothesis and indicate that the effective functioning of the compensation committee has a mitigating effect on the relationship between discretionary bonuses and excess CEO compensation. This finding suggests that the more effective the compensation committee can function, the less discretionary bonuses lead to excess CEO compensation.
Panel C of Table 4 presents the findings regarding the third hypothesis: influence of the CEO strengthening the relationship between discretionary bonuses and excess CEO compensation. The coefficient indicates that influence of the CEO is significantly and positively associated with the relationship between discretionary bonuses and excess CEO compensation (β = 0.026 and p = 0.089). These findings support the hypothesis and indicate that influence of the CEO on compensation has a strengthening effect on the relationship between discretionary bonuses and excess CEO compensation. This finding suggests that the more influence a CEO has, the more discretionary bonuses lead to excess CEO compensation.
30
Table 4. OLS regression analysis for hypotheses 1, 2 and 3
Variable definitions and descriptions are provided in appendix B * Significance at the 10% level
** Significance at the 1% level *** Significance at the 0.1% level
5. Discussion
The purpose of this study was to investigate the contribution subjective performance measures to excess CEO compensation. The research question that guided this study was: Do
discretionary bonuses contribute to excessive CEO compensation?
To balance the different interests of shareholders and CEOs, performance is linked to compensation in compensation contracts (Bol, 2008). When CEOs receive more
compensation than can be explained by firm performance or economic factors it is considered to be excessive (Brick et al., 2006; Core et al., 1999 & Hill et al., 2016). Performance is then commonly measured with objective performance measures. However, uncontrollable random
EXCESS
(A) (B) (C)
Variable name Coefficient estimate Standard error Coefficient estimate Standard error Coefficient estimate Standard error Constant -1.885*** 0.193 -1.883*** 0.191 -1.880*** 0.192 IND1 0.103* 0.048 0.098* 0.047 0.101* 0.047 IND3 -0.022 0.041 -0.020 0.041 -0.027 0.041 IND4 -0.110 0.069 -0.085 0.069 -0.107 0.069 IND5 -0.155*** 0.035 -0.148*** 0.035 -0.153*** 0.035 IND6 -0.031 0.046 -0.025 0.045 -0.038 0.046 ROA16 -0.025 0.027 -0.032 0.026 -0.029 0.026 DISCB 0.334*** 0.033 0.334*** 0.033 0.332*** 0.033 DISCB*COMEFF -0.040* 0.020 DISCB*CEOINFL 0.026* 0.017 n 125 125 125 F-statistics F = 18.014; p = .000 F = 16.659; p = .000 F = 16.233; p = .000 Adjusted R² 0.490 0.503 0.496
31
events called noise are increasing the uncertainty the CEO faces and cause imperfect objective performance measures (Gibbs et al., 2004). The beneficial effect of including subjective performance measures comes from the possibility to make discretionary adjustments in compensation to filter out noise (Höppe & Moers, 2011). However, subjective performance measures are not perfect either. Subjective performance measures suffer from leniency effects causing inflated performance ratings (Golman & Bhatia, 2012; Moers, 2005) and it is possible for evaluators to develop favoritism (Prendergast & Topel, 1996). As the manifestation of these downsides of subjective performance measures in excess CEO compensation have not been investigated before, it appears to be important to investigate this relationship.
The study of this paper is set up in the development of three hypotheses. The first hypothesis states that discretionary bonuses are positively associated with excess compensation. This prediction comes from the downsides of subjective performance measures and the remaining unexplained existence of excess CEO compensation. The analysis results in evidence that the hypothesis may be accepted. Therefore this paper concludes that discretionary bonuses are associated with excess compensation.
The second hypothesis stated that the effective functioning of the compensation committee mitigates the relationship between discretionary bonuses and excess CEO compensation. The analysis related to this hypothesis found evidence for this statement. Therefore this paper concludes that effective compensation committees mitigate the relationship between
discretionary bonuses and excess CEO compensation. As compensation committees function in a more effective way they seem to reduce the excessive levels of CEO compensation. This effective functioning may be improved by avoiding dependency of the CEO, being more diverse and working as a team.
The third and last hypothesis stated that CEO influence strengthens the relationship between discretionary bonuses and excess CEO compensation. The analysis that followed found evidence for this statement. This leads this paper to conclude that CEO influence strengthens the relationship between discretionary bonuses and excess CEO compensation. As CEOs become more influential they seem to increase the excessive levels of CEO compensation. Influential CEOs may be indicated by the development of social exchange relationships with the compensation committee that increase over time. Or they may be indicated by large CEO
