The managerial influence in relative performance evaluation to
peer firm selection among performance awards
July 1, 2018
Using a sample of S&P 1500 firms between the years 2006 and 2015, the practice in which peer groups are selected in relative performance evaluation is examined using obtained data from the ISS Incentive Lab on proxy statements (DEF14A filings). The results of the multiple logistic regression analyses show that RPE firms select peer groups to filter out common shocks. However, little evidence is provided for the opportunistic use of RPE in
compensation contracts which are based on the total shareholder return (TSR) performance metric. By studying the manner in which RPE firms select peer groups based on price-based and/or accounting performance metrics, contributes to the existing literature on the incentives behind these awards.
Master Thesis
Name Lotte de Wit
Student nr. 10642056
Supervisor Dr. T. Jochem
Specialization Corporate Finance University University of Amsterdam
1
Statement of originality
This document is written by Student Lotte de Wit who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
2
Table of contents
1. Introduction 3
2. Literature review
2.1 Relative Performance Evaluation………...5 2.2 Performance metrics in RPE awards.……….7
3. Data 3.1 Institutional background……… 9 3.2 Sample selection……… 9 3.3 Descriptive statistics………..10 4. Methodology 4.1 Hypotheses development………... 12 4.2 Models………... 13 4.3 Explanatory variables……… 16 5. Results 18 5.1 Results models………... 18 6. Limitations 24 7. Conclusion 25 References 27
3
1. Introduction
Over the last decade, executive compensation has risen considerably and continues to attract policymakers, shareholders and the public at large. Anecdotal evidence suggests that firms tend to cherry-pick peer group members to enlarge pay packages. For instance, in a New York Times article written in 2003, the problems of using peer groups in determining executive compensation caught the public’s attention. The New York Stock Exchange disclosed that it has paid his chairman, Richard A. Grasso, $140 million in compensation. One reason was that the peer group included highly profitable investment banks and financial institutions which were much larger and complex compared to the Exchange. Through the inadequate peer group selection, an unreasonable high compensation could be set. 1
Although setting the level of executive compensation relative to a predetermined group of firms is a common practice, it involves discretion. The main rationale for the use of compensation benchmarking can be found in economic theory. To shield risk-averse agents against common exogenous shocks which influence firm performance, relative performance evaluation (hereafter, RPE) theory provide a solution were the agent should not only be reward solely on the agents own firm performance but also on the performance of others (e.g. Hölmstrom, 1979, 1980; Jensen & Murphy, 1990; Aggarwal & Samwick, 1999a). But, the potential of managerial influence and conflict of interest among parties, allow for the possibility of firms choosing peers opportunistically and increase compensation (e.g. Bebchuk & Fried, 2003; Gibbons & Murphy, 1990; Murphy, 2001). The board of directors determine the custom peer group used in RPE, and is therefore subjected to managerial influence and questioned to be biased. Prior literature tried to attempt possible explanations for RPE usage in contracts but find mixed evidence which results in an unsolved puzzle.
While there is considerable literature on compensation benchmarking, there is less research on peer group selection in RPE award settings. Understanding the practice of how peer groups are designed for the purpose of RPE is decisive to understand the risk-sharing and incentives of these awards. This study contributes to the ongoing discussion if peer group selection in RPE contracts are used to filter out common shocks or to inflate executive
compensation. Using a framework similar to Gong et al. (2011) and Bizjak et al. (2017), the aim of the thesis is to provide more intuition in the practice how peer groups are selected based on different types of performance metrics among RPE firms. This study is different in
1 In 2006, the New York Times wrote an article about peer pressure and how it inflates executive pay. Retrieved
from: https://www.nytimes.com/2006/11/26/business/yourmoney/peer-pressure-inflating-executive-pay.html
4
contrast to the authors based on the research to opportunism in the performance metrics total shareholder return (TSR) and earnings-per-share (EPS). Gong et al. (2011) use a sample of explicit RPE awards granted in 2006 and examine the characteristics of peer firms selected in the RPE peer group. They find that for added (dropped) peers the industry adjusted expected performance is significant (insignificant) negative. The study of Bizjak et al. (2017) is closely related to Gong et al. (2011), but use comparative static analysis to identify the characteristics of peer firms that are the drivers of award payout. This analysis provides the ability to present justification behind the included variables in the regression and at the same time to identify potential managerial influence in selecting strategic peers. Although Bizjak et al. (2017) perform the analysis on RPE firms with the performance metric TSR, this study will also provide insight on the peer selection for RPE firms based on the performance metric EPS.
For this study, two hypotheses questions are constructed. The first hypothesis
provides insight if whether RPE firms select peers that are consistent with economic theory to filter out common shocks or to increase award payout. The second hypothesis is constructed as a continuation on the results of hypothesis 1. The second hypothesis is developed to shed light on the potential managerial influence to select a custom peer group which they expect to outperform in the established performance metric. To examine these questions the
Institutional Shareholder Services (ISS) Incentive Lab (IL) is used for the sample period 2006 till 2015. Prior to 2006, firms were not mandatory to disclose detailed information on RPE contracts in their proxy statements. To increase transparency of executive pay packages, the SEC announced a new disclosure rule in 2006 which requires companies to provide
mandatory disclosure concerning compensation benchmarking. By means of this rule, detailed information can be collected on RPE contracts for a large sample of U.S. firms. Further, stock and financial data is supplemented from CRSP and Compustat.
The results of this research provide evidence that peer groups are selected to filter out common shocks from performance and improve the measurement of managerial ability and effort. This is shown by the findings that RPE firms select peers that are more likely to come from the same industry. In addition, small evidence for opportunism is found for dropped peers by RPE firms which awards are based on the performance metric TSR. However, the evidence is weak.
5
The thesis is structured as follows. In Section 2, the existing literature on relative performance evaluation and performance metrics used in RPE awards are covered. The description of the data and summary statistics are described in Section 3. Section 4, covers the development of the hypothesis and methodology. The empirical results are discussed in Section 5. The limitations of this study are presented in Section 6, followed by the conclusion in Section 7.
2 Literature Review
Section 2 will elaborate on the existing literature on relative performance evaluation and performance metrics used in RPE awards. Literature on relative performance evaluation is studied to provide more insight into the use and justification behind RPE. In addition, the literature on the design of RPE awards is studied to produce more intuition behind the performance metrics setting in RPE. Subsection 2.1 elaborates on the use of relative compensation evaluation, followed by Subsection 2.2 which covers the literature on performance metrics in RPE awards.
2.1. Relative Performance Evaluation
When the agent’s effort is non-contractible and unobservable, the second-best contracting is to provide an incentive contract were the agent’s compensation is linked to observable measures of firm performance (Hölmstrom, 1979). This prediction is consistent with prior literature that chief executive officers (CEOs) are rewarded for the increase of their own firm stock return (Jensen & Murphy, 1990; Aggarwal & Samwick, 1999a). However, this
incentive contract imposes risk for the risk-averse agent to the extent that firm performance is influenced by common shocks that are outside the agent’s control. These uncontrollable shocks potentially decrease the utility of the agent, thereby reducing contracting efficiency. Hölmstrom (1982) proposed a solution whereby the agent should not only be rewarded solely for their own firm performance, but also depend on the performance relative to that of her peers. This is the idea behind the Relative Performance Evaluation Theory.
Prior studies have attempted to test this efficient contracting in CEO compensation contracts (e.g. Antle & Smith, 1986; Jensen & Murphy, 1990; Aggarwal & Samwick, 1999a), but find mixed evidence which in turn resulted in studies that seek to explain this ‘’RPE puzzle’’. Bertrand and Mullainathan (2001) dispute that firms with weak governance are less likely to use relative performance evaluation because CEOs have the potential to affect their pay-setting process and are paid for positive external shocks but not penalized for negative
6
ones (i.e., pay-for-luck). Other streams of studies argue that RPE contracts allow
management to extract rent from shareholders (e.g. Gibbons & Murphy, 1990; Murphy, 2001; Bebchuk & Fried, 2003). In the study of Bebchuk and Fried (2003) the optimal contracting view is explained as a possible solution to the agency problem. The study uses executive compensation as an instrument to address the agency problem between shareholders and managers, but find that is also the part of the agency problem itself. They argue that boards of publicly traded companies with dispersed ownership can’t be expected to negotiate arm’s length with managers. This managerial power results in managers influence their own pay and interests of reducing their amount of pay to the extent to which pay is linked to firm performance. Gibbons and Murphy (1990) question that RPE contracts may create incentives to executives to choose improper peers and influence peer performance. This inefficient selection of peers can result in deliberately choosing underperforming RPE peers to inflate firm performance and allow executives to increase compensation. Westphal and Zajac (1994, 1995) propose that RPE firms prefer to select well established and visible peers to reflect symbolic considerations. This suggests that RPE firms justify their choice of peers to external distinctions.
Early literature found only weak evidence on the implicit approach of RPE in compensation contracts (Antle & Smith, 1986; Gibbons & Murphy, 1990; Murphy, 1999; Garvey & Milbourn, 2003). One potential reason for these earlier findings were the wrong description of the peer groups used in the tests. Prior to 2006 firms were not required to disclose RPE use in executive contracts, so the main challenge for most of these studies was data availability. In the absence of disclosure, studies with implicit tests had to rely on assumptions about RPE contracts which lead to measurement errors. However, recent literature pursues to test relative performance evaluation by providing more refined
specifications about RPE peers groups and found supportive evidence for the use of implicit RPE (e.g., Albuquerque, 2009; Lewellen, 2013; De Angelis & Grinstein, 2015; Jayaraman et al., 2015). Albuquerque (2009) finds problems for RPE firms using a SIC group as peers because this group may not face the common external shocks and the firm’s ability to respond to common shocks is likely to vary within the same industry. She finds that firms of similar size and which are active in the same industry are more likely to face similar shocks and have the same ability to respond to those shocks. Likewise, Jayaraman et al. (2015) investigate the role of RPE in CEO pay and turnover and find that firms can completely filter out common shocks in the presence of a large number of peers. These papers show the importance of understanding how peers are selected in explicit RPE awards can guidance for
7
peer group selection of implicit RPE.
The use of relative performance evaluation in executive contracts results in an ongoing debate whether the use of compensation peer groups is efficient (e.g. Bizjak et al., 2008, 2011; Albuquerque et al., 2013; Faulkender & Yang, 2010, 2013). Are RPE firms construct peer groups efficiently to filter out common shocks and improve signals about managerial ability or are peers strategically selected in order to increase award payout. This study contributes to the implications if firms select peer groups confirm the efficient
contracting perspective or the rent extraction perspective.
2.2. Performance metrics in RPE awards
Most relative performance evaluation awards use a rank order tournament to determine award payout. Under this design, the firm grants an RPE award to the executive whereby
performance is measured for the target firm and a group of peers over a pre-specified period of time (Bizjak et al. 2011). When the tournament ends, which is often a three-year period, the target firm is pooled against its peers and ranked by performance, based on either price-based or accounting metrics, to get a percentile or performance ranking. After this, the performance or percentile ranking is graphed by a payout function to regulate the definite award payout of shares, option or cash to the executive.
As explained above, the award payout in a rank order tournament is a function of the target firm performance relative to a group of peers. As a benchmark, three different types of peer groups can be used in the design of RPE awards. Namely, the broad-based market index, the industry-specific index or a custom selected group of peer firms. The broad-based market index (e.g. S&P500 and the Dow Jones Industry Average) is designed to reflect the
movement of the entire market and the Industry-specific Index includes firms that belong to a specific industry or market. When a custom group of peers is used, both the type and size of the peers are determined by the target firm. The board of directors ultimately decide what type of peer group to use (and in the case of a custom peer group, which firms are included). Both the type and composition of the peer group have important values to the extent if peer groups are constructed for the economic justification to filter out common shocks. Since executives are often involved and have input in the selection of peers included in the custom peer group, there is a potential bias in peer selection that could benefit executives.
Next to the composition of the peer group, the award structure has to be established at the beginning of the performance period. The firm can contract on combinations of price-based and/or accounting performance metrics when determining RPE awards in executive
8
compensation. The most common measure of firm performance is total shareholder return (TSR). Gong et al. (2011) report that 70% of RPE firms in their 2006 sample use stock returns as the performance metric followed by 14% and 12% using ROE and sales, respectively. Also, it is possible that firm use more than one performance metric to
benchmark relative performance (e.g. a combination of stock return and EPS). At the end of the performance period, payout levels for the RPE award are typically conditional on
achieving specified rankings or percentiles within the peer group or reaching a certain value, such as the average TSR for the S&P 500 over the performance period.
As an example, one of the firms employing RPE in the sample is Hess Corporation. Hess Corporation is an energy company engaged in the exploration and production of crude oil and natural gas. In 2012, the committee changed the long-term incentive award program so that the annual award shares are divided into 50% performance share units and 50% restricted share units, with stock options being eliminated. The payout of performance share awards is based on the relative performance of the TSR over a three-year performance period compared to fifteen peer companies. Payouts of the 2012 performance share awards will range from 0 to 200% of the firm award based on the companies TSR ranking within the peer group.
To date, the literature lacks a clear understanding of the sources of variation in firm decision to use either price-based and/or accounting performance metrics in RPE contracts. Ideally, a firm should select the RPE performance metric that motivates managers the best to maximize firm value. However, preceding literature suggest that managers receive the
incentive to improve the selected performance measure and take actions consistent with those incentives (e.g. Wallace, 1997; Marquardt & Wiedman, 2007; Young & Yang, 2011). For instance, Wallace (1997) shows that firms using compensation plans based on residual income measures, report higher residual income but have lower levels of investment, increased share price purchases, and increased disposition of assets than other firms. Marquardt and Wiedman (2007) argue that managers are more pronounced to employ convertible bonds transactions to increase diluted EPS with regard to market measures. Finally, Young and Yang (2011) report when EPS is used as the performance metric in bonus plans, firms increase stock repurchases. Consequently, these firms report higher performance defined by the specific measures, but not by other performance metrics. These papers show the importance of the manner how peer groups are constructed and the incentives of these awards.
9
3 Data
3.1 Institutional background
Due to the increased public and shareholder scrutiny of executive compensation practices, the SEC announced a new disclosure rule in 2006 to provide more clarity on executive
compensation contracts. This disclosure rule called ’Executive Compensation and Related Party Disclosure’, requires companies to specify a ‘Compensation Discussion and Analysis’ (CD&A) in their proxy statements. Prior to 2006, firms could voluntarily choose to disclose detailed information on RPE contracts in their proxy statements (Carter et al., 2009; Byrd et al., 1998). Through the mandatory disclosure concerning compensation benchmarking, a firm is obligated to provide a detailed description on the process which is used to select
performance peers and how the evaluation of the firm’s performance translates into the objective compensation determination. By means of this rule, detailed data can be collected on RPE contracts for a large sample of U.S. firms.
3.2 Sample Selection
Data is obtained from the Institutional Shareholder Services (ISS) Incentive Lab (IL) from proxy statements (DEF 14A) on the various aspects of relative performance evaluation awards including all the necessary features to value awards. The IL covers the largest 750 U.S. firms measured by market capitalization. Due to backward- and forward-filling, the Incentive Lab data compass the full S&P 500, most of the S&P Midcap 400 and a portion of the S&P Small-Cap 600. This result in an annual cross-section that compasses the largest 1,000 firms listed on the U.S. stock market in terms of market capitalization. By means of the mandatory disclosure of executive compensation details by the SEC in 2006, a coverage of firms is more comprehensive after that year. Therefore, the analysis is focused on the sample selection of firms between the years 2006 till 2015.
The ISS Incentive Lab collects information for each grant on the form of payout (stock options, stock units, or cash). A firm is defined as RPE firm if the firm has one or more grants which are determined based on relative performance towards a selected group of peers. Otherwise, the firm is classified as a non-RPE firm. Additional, ISS IL provides information on the conditions for payout (the fulfillment of performance criteria [ABS] or relative performance criteria [Rel] or a combination of both [Abs/Rel]) and specifics on accounting- or stock-based performance metrics that are associated with performance-based grants. Finally, ISS IL compile information on the specific peer firms or indexes selected for
10
awarding the grants based on relative performance. The dataset is supplemented with stock data from CRSP and accounting data from Compustat. Annual financial data is collected from Compustat daily updates to compute the firm-level control variables. Further, the CRSP daily stock file is used to compute annual stock returns, stock volatility and the compounded average growth rate of stock returns. This result in a final dataset of 2,224 RPE firm-year awards granted in the sample period.
3.3 Descriptive Statistics
Table I present the summary statistics on the frequency of RPE use on the main features of RPE contracts. Table I, Panel A offers insight on the relative percentage of RPE use in the sample selection. According to the data, the frequency use of RPE has grown limited over time. Around 34.34% of the firms in the incentive lab use some type of RPE award by 2015. In addition, the different types of peer groups used for benchmarking among RPE firms are presented in Panel A. The mostly used peer group used by RPE firms is the custom selected peer group followed by the S&P 500 (broad-market index) and other indexes (industry-specific index), respectively. The summary statistics presented in Table I Panel A are higher than the findings of Gong et al. (2011). The variation in the frequency of RPE use is most likely caused by the difference in sample selection as the authors focus only on the year 2006. The outcomes of the statistic description are more comparable to the results of Bizjak et al. (2017). Although there are small differences, findings show similarities in the increasing usage of relative evaluation performance over time and the custom peer group is mostly used among other benchmark peer groups. Notice that the row values of Panel A in Table I do not add to 100% because firms might use more than one type of peer group (e.g. an index and a custom set of peer group).
In Table I, Panel B the performance metrics used in RPE award contracts are reported. Looking at the summary statistics of Panel B in Table I, the majority of RPE firms in 2015 (84.61%) rely solely on price-based performance metrics (include stock returns) when implementing RPE. Also, accounting performance metrics (for instance return on equity, earnings per share and return on assets) are common in RPE contracts. In addition to the RPE firms that use a single RPE metric, some firms use multiple metrics in their RPE award. These firms rely both on price-based and accounting performance metrics in RPE contracts. The frequency of price-based metrics and accounting performance metrics are comparable with the findings of Bizjak et al. (2017), were most firms rely on stock return followed by accounting metrics, respectively.
11
Table 1: RPE usage
Table 1 provide the descriptive statistics of RPE usage and details of RPE awards in the sample. Panel A reports the frequency of firms which use relative performance evaluation from 2006 to 2015. In addition, Panel A presents the different types of peer group used for benchmarking. Panel B, provide the distribution of performance metrics from 2006 to 2015. Note that rows and columns do not add up to 100% because firms can use more than one RPE award with different characteristics.
Panel A: RPE usage and Peer Group Type
Year N RPE Selected S&P500 Other Index Peers 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 1840 1817 1800 1802 1804 1856 1888 1895 1862 1811 32.50% 32.85% 34.44% 35.15% 35.53% 34.48% 32.99% 32.72% 33.08% 34.34% 97.18% 97.98% 97.58% 97.47% 97.66% 97.66% 98.50% 98.71% 98.54% 98.07% 1.84% 1.00% 1.29% 1.58% 1.25% 1.41% 1.93% 1.94% 1.94% 1.93% 1.34% 1.34% 1.45% 1.26% 1.40% 1.25% 1.61% 1.94% 2.11% 0.32%
Panel B: Price-based and accounting performance metrics
Year Price-based metrics
Stock returns Accounting metrics 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 94.4% 88.23% 93.33% 72.72% 83.33% 77.77% 76.19% 79.16% 83.33% 84.61% 11.11% 17.64% 13.33% 31.81% 22.22% 27.22% 28.57% 25.00% 25.00% 23.08%
12
Not tabulated, a small percentage of RPE firms use other non-financial information performance metrics (for instance, customer satisfaction, debt related or employment engagement) for RPE purpose. In this research, the emphasis is based on the study of price-based and accounting performance metrics used in RPE contracts. Therefore, not much insight will be given to the use of non-financial performance metrics. Also, not presented in the table are the average number of peers within a custom peer group selected by a RPE firm. The average number of peers within a peer group is around 14 peer firms, but there is some variation. In the 25th percentile (75th percentile) the number of set peers is 5 (16).
4 Methodology
The aim of the research question is to provide more intuition in the practice how peer groups are selected based on different types of performance metrics among RPE firms. In other words, if RPE firms select peers to filter out common shocks or to opportunistically increase award payout regarding performance metrics. In this section, the methodology is explained. Subsection 4.1 describes the development of the hypotheses. Subsection 4.2 presents the models. In subsection 4.3 a review is given on the used explanatory variables.
4.1 Hypotheses development.
In Subsection 2.2 the theoretic justification for the use of relative performance evaluation is discussed. However, there exists an ongoing discussion whether RPE firms choose peer groups which have a higher ability to remove common risk or whether firms use RPE as a mechanism to choose improper and underperforming peer groups to increase award payout (Gibbons & Murphy, 1990; Murphy, 2001; Bebchuk & Fried, 2003; Faulkender & Yang 2010, 2013). While there is considerable literature on compensation benchmarking, there is less literature on the practice in which practice peer groups are selected in RPE awards. Understanding the way in which peer groups are constructed for the purpose of RPE is critical to understand the risk-sharing and incentives of these awards.
As described in Subsection 2.3, the award payout in a rank order tournament is a function of the RPE firm performance relative to a group of peers. When the tournament ends, percentiles are mapped to order the define award payout of shares, option or cash to the executive. The higher the firm is pooled against its peer the higher the award payout. From
13
the rent extraction view, peer selection is subjected to managerial influence and they select peers which they expect to outperform (Gibbons & Murphy, 1990; Murphy, 2001). By choosing underperforming firms in the peer group selection, the chance of the RPE firm to outperform its peers (after performance is ranked in percentiles after the tournament ends) increases and compensation inflate. To provide more insight for the purpose of peer groups, the first hypothesis is developed:
Hypothesis 1: RPE Firms select underperforming peer firms in the peer group.
Firms using RPE can contract awards based on performance metrics, including price-based metrics (stock returns) and/or accounting metrics (e.g. EPS, ROE, Cash flow or ROA). Ideally, firms should select the performance metric that motivates managers the best to maximize firm value. However, prior literature suggests that managers usually focus to improve the selected performance metric and take actions consistent with those incentives (e.g. Wallace, 1997; Marquardt & Wiedman, 2007; Young & Yang, 2011). The second hypothesis is developed to detect strategic peer group selecting regarding pre-determined performance metrics by RPE firms. Hereby Hypothesis 2 is developed:
Hypothesis 2: Firms are more likely to select peers into the peer group if they expect to outperform the peer in the determined performance award metric.
The results of this study will contribute to the ongoing debate on the practice in which practice peer groups are selected in RPE awards.
4.2 The models
The objective of this subsection is to gain a better understanding on the implications of peer group selection in RPE awards. The first model is used as a stepping stone to the second model. Using panel data for the sample period of 2006 till 2015, multiple logistic regressions are used in the analyses. The purpose of the logistic regressions is to identify factors that drive peer firm selection. In the analysis, the characteristics of firms added and dropped over time from the RPE peer group are tested. Each RPE firm-year is matched with potential peer firms to create a universe of candidate peers. The way in which peers are added and dropped over time provide insight in which manner peers are selected. This means if peer selection conforms to efficient contracting were the observed added peers are expected to exhibit
14
greater ability to reduce common shocks and have greater similarity than peers that has been dropped. At the same time, if RPE firms are more likely to select poor performing peer firms when modeling peer groups, added peers are expected to perform worse compared to dropped peer firms. The strategical ability of RPE firms to add and drop peer firms from the peer group to select new peer firms and increase award payout may support the rent-extraction view. To estimate the selection of RPE peer firms in the peer group, the following logistic regressions are established:
Equation (1) !"#$%&'(()*+,- = /0+ /23(&3(#+,-+ /43(&'%*#3(#+,-+ /53(&6'7+,-+ /83(&'%*#6'7+,-+ /83(&79:(+,-+ /;3(&<=&+,- + />3(&?@A+,-+ /B7CD2+,-+ /F7CD3+,-+ H
+,-Equation (2) !II(I'(()*+,-J2 = /0+ /23(&3(#+,-+ /43(&'%*#3(#+,-+ /53(&6'7+,-+ /83(&'%*#6'7+,-+ /83(&79:(+,-+ /;3(&<=&+,-+ 3(&?@A+,-+ /B7CD2+,-+ /F7CD3+,-+ H +,-Equation (3) L)=MM(I'(()*+,-J2 = /0+ /23(&3(#+,-+ /43(&'%*#3(#+,-+ /53(&6'7+,-+
/83(&'%*#6'7+,-+ /83(&79:(+,-+ /;3(&<=&+,- + />3(&?@A+,-+ /B7CD2+,-+ /F7CD3+,-+ H+,- The specifications shown in Equation 1 are tested on the universe of potential peer firms that are an actual member of the RPE peer group formed by RPE firm i in year t. Equation 2 is tested on the universe of potential peers that could be added to the actual peer group by RPE in the following year. Here, the non-added peers dominate the sample in this universe.
Equation 3 is tested on the actual peer firms in the peer group and are compared with the peer group the year afterward to see which peers are dropped. Therefore, the logistic regressions of Equations 2 and 3 tests if the composition of the peer group selected by RPE firm i in year t+1 has changed by peer firms that are added and dropped within the peer group.
In Equation 1, the dependent variable ActualPeers is a binary variable that equals one if the candidate peer firm is an actual member of the RPE peer group by RPE firm i in year t, and zero otherwise. In Equation 2, the dependent variable AddedPeer is an indicator variable that equals one if the potential peer firm is added as a member to the RPE peer group by RPE firm i in year t+1, and zero otherwise. Lastly, in Equation 3 the indicator variable
DroppedPeer equals one if the actual peer firm is dropped as a member of the RPE peer
group by RPE firm i in year t+1, and zero otherwise.
Testing these first regressions, it is expected that RPE firms are more likely to select outperforming peer firms for efficient contracting, whereas underperforming peer firms are chosen for rent-seeking behavior (Gong et al. 2011). Besides Gong et al. (2011), it is expected that RPE firms select peers to filter out common shocks from performance to
15
improve efficient contracting (Bizjak et al. 2017). According to Bizjak et al. (2017), peer firms which are added (dropped) are more (less) likely to be in the same S&P 500 index, added (dropped) peer firms tend to be larger (smaller) in terms of assets, and added (dropped) peer firms have lower (higher) stock volatility compared to the RPE firm.
The next logistic tests build further on the first model. The specifications will test if RPE firms choose peer firm characteristics that could suggest opportunism, while using the same explanatory variables used in the first model. Again, each RPE firm-year is matched with all possible candidate firms to join the peer group. However, the logistic regressions are performed for RPE firms that use total shareholder return (TSR) or earnings-per-share (EPS) as a performance metric to grant awards. The purpose of the logistic regressions is to explore if RPE firms include poor performing peers in the peer group which measure the RPE firm is expected to outperform regarding the pre-established award metric. The following
specifications are used:
Equation (4) !"#$%&'(()*NOP,+,- = /0+ /23(&3(#+,-+ /43(&'%*#3(#+,-+ /53(&6'7+,-+
/83(&'%*#6'7+,-+ /83(&79:(+,-+ /;3(&<=&+,- + />3(&?@A+,-+ /B7CD2+,-+ /F7CD3+,-+ H
+,-Equation (5) !II(I'(()*NOP,+,-J2 = /0+ /23(&3(#+,-+ /43(&'%*#3(#+,-+ /53(&6'7+,-+ /83(&'%*#6'7+,-+ /83(&79:(+,-+ /;3(&<=&+,-+ 3(&?@A+,-+ /B7CD2+,-+ /F7CD3+,-+ H +,-Equation (6) L)=MM(I'(()*NOP,+,-J2= /0+ /23(&3(#+,- + /43(&'%*#3(#+,-+ /53(&6'7+,-+
/83(&'%*#6'7+,-+ /83(&79:(+,-+ /;3(&<=&+,- + />3(&?@A+,-+ /B7CD2+,-+ /F7CD3+,-+ H+,-
Equation (7) !"#$%&'(()*QPR+,- = /0+ /23(&3(#+,-+ /43(&'%*#3(#+,-+ /53(&6'7+,-+
/83(&'%*#6'7+,-+ /83(&79:(+,-+ /;3(&<=&+,- + />3(&?@A+,-+ /B7CD2+,-+ /F7CD3+,-+ H
+,-Equation (8) !II(I'(()*QPR,+,-J2= /0+ /23(&3(#+,-+ /43(&'%*#3(#+,-+ /53(&6'7+,-+
/83(&'%*#6'7+,-+ /83(&79:(+,-+ /;3(&<=&+,-+ 3(&?@A+,-+ /B7CD2+,-+ /F7CD3+,-+ H +,-Equation (9) L)=MM(I'(()*QPR,+,-J2 = /0+ /23(&3(#+,-+ /43(&'%*#3(#+,-+ /53(&6'7+,-+ /83(&'%*#6'7+,-+ /83(&79:(+,-+ /;3(&<=&+,- + />3(&?@A+,-+ /B7CD2+,-+ /F7CD3+,-+ H+,- Keep in mind that all the tests above are performed for RPE firms that use total shareholder return (TSR) or earnings-per-share (EPS) as a performance metric. In short, as described in subsection 2.2, this means that the RPE firm grants an award to the executive whereby after the performance period firm performance is ranked and compared to a group of peers. Here, the award is either based on the performance of the RPE firms TSR or EPS relative to the
16
peer group.
In Equation 4 and 7, the dependent variables !"#$%&'(()*NOP (!"#$%&'(()*QPR)
equal one if the potential peer firm is an actual peer firm in the RPE peer group by RPE firm i in year t and the performance metric to grant the award is EPS (TSR) of the RPE firm, and zero otherwise. In Equation 5 and 8, the dependent variables !II(I'(()*NOP
(!II(I'(()*QPR) equal one when the potential peer firm is not a member of the RPE peer
group in observed year t but is included as a member to the RPE peer group by RPE firm i in year t+1 and the performance metric to grant the award is EPS (TSR) of the RPE firm, and zero otherwise. In Equation 6 and 9, the dependent variables L)=MM(I'(()*NOP
(L)=MM(I'(()*QPR) equal one if the peer firm is an actual member of the RPE peer group in
observed year t but it excluded from the peer group by RPE firm i in year t+1 and the performance metric to grant the award is EPS (TSR) of the RPE firm, and zero otherwise. Following a framework similar to Bizjak et al. (2017), opportunism could be suggested if it is expected that RPE firms add (drop) potential peer firms which tend to have poor (better) stock return performance and higher (lower) stock volatility. Also, Gong et al. (2011) find evidence that RPE firms add (drop) peer firms that have closer (larger) firm size compared to the RPE. This is consistent with more (less) similar firms being added (dropped) to (from) the RPE peer group.
Following Bizjak et al. (2017) all continuous variables in the equations above are Winsorized at the 5th and 95th percentiles to mitigate skewness in the data. Using the specifications of the equationsHypothesis 1 and 2 can be either rejected or verified.
4.3 Explanatory variables
The equations described in Subsection 4.2 include the same set of explanatory variables. The explanatory variables that are used in the model are meant to capture firm similarities. Potential determinants for relative performance evaluation in the peer group selection are in accordance with the papers of Gong et al. (2011) and Bizjak et al. (2017). All variables with
Rel are measured as the difference between the RPE firm characteristics minus the same
characteristics for the potential peer firm. Wherefore, a negative (positive) sign indicates that the peer characteristic is larger (smaller) compared to the RPE firm.
17
Performance. To examine differences in past performance between the RPE firm and a
potential peer firm, which could indicate whether peers are selected to increase award payout, the difference in stock volatility (RelVol), the difference in compounded annual growth rate (CAGR) of stock returns for the prior three years (RelPastRet), the difference in stock returns (RelRet), the difference between the average annual growth rate (AAGR) of EPS for the prior three years (RelPastEPS) and the difference in EPS (RelEPS) are included.2 These variables are chosen in the manner that they provide insight into how RPE firms can influence peer groups. According to the literature, if firms want to increase award payout they should pick peers that the RPE firm expects to outperform (Gong et al., 2011; Bizjak et al., 2017). This strategy appears to work best when peers have a higher stock price volatility. The other relative variables indicate if the selected peers relatively under- or outperform the RPE firm.
Common risk. To better protect executives from common shocks which are outside the
agent’s control, RPE firms select peer firms that bear greater common risk with the firms. To determine the exposure of potential peers to common risk, industry indicator variables are included. The Standard Industrial Classification (SIC) code is a four-digit numerical code assigned by the U.S. government to business organizations to identify the primary business of the companies.3 The dummy variable (SIC2) equals one if the potential peer firm starts with the same two-digit SIC code as the RPE firm, and zero otherwise. A similar method is used for the dummy variable (SIC3) which equals one if the potential peer starts with the same three-digit SIC code as the RPE firm, and zero otherwise.
Growth opportunity. The difference in market-to-book value of assets between the RPE
firm and a potential peer (RelMTB) is used as a proxy measurement for growth opportunities. According to Murphy (2001), firms with high growth opportunities are more likely to adopt external standards (such as RPE) in constructing executive compensation. Wherein theory is silent on the effect of firm size on the use of relative performance evaluation, RPE firms tend to select peer firms with comparable sizes (Gong et al., 2011). The variable (RelSize) is measured as the difference in the natural logarithm of total assets between the RPE firm and a potential peer. To mitigate skewness of this variable, a natural logarithm is applied.
2 The average performance period for RPE awards is typically three years. Hereafter the target firm is pooled
against its peers and ranked by performance.
3 Retrieved from SIC Code website: https://siccode.com/en/pages/what-is-a-sic-code
18
5. Results
In this section, empirical evidence on peer selection in relative performance evaluation awards is provided using a framework similar to Gong et al. (2011) and Bizjak et al. (2017). First, the characteristics of firms added and dropped over time from the RPE peer group are tested. This, to explore if peer selection conforms to efficient contracting or the ability to select peers strategically. Second, the subsequent logistic tests build further on the first tests. This time, the purpose of the regressions is to detect if RPE firms include poor performing peers in the peer group which metric the RPE firm is expected to outperform. Subsection 5.1 provides the results of the logistic regression analyses.
5.1 Results models
This subsection will present the empirical results of peer group selection in relative
performance evaluation between the years 2006 and 2015. As described in Subsection 4.2, all dependent variables that are used in the equations are binary variables indicating a one or a zero. Also, all regressions are controlled for year and industry fixed effects. These fixed effects are included to account for potential spurious effects between the variables. The emphasis of the analysis is to observe implications of the peer group selection process by RPE firms. Keep in mind that all variables with Rel are measured as the difference between the RPE firm characteristics minus the same characteristics for the potential peer firm. Whereby, a negative (positive) sign indicates that the peer characteristic is larger (smaller) compared to the RPE firm.
In the first set of logistic regressions, the dependent variables are Actual Peers, Added Peers, and Dropped Peers, respectively. The explanatory variables used in the models are to capture firm similarities. The results of equations 1, 2, and 3 are shown in Table II. Looking at Table II column 1, the actual peers that are included in the RPE peer group have lower CAGR in stock return, have higher earnings-per-share, are larger in terms of size, have lower stock volatility, and have larger growth opportunities compared to the RPE firm. In addition, actual peers are more likely to operate in the same industry (either two-digit and three-digit SIC code) as the RPE firm. All results are significant at a 1% and 5% level. The results of column 1 provide evidence that firms select peers into the peer group to filter out common risk. These findings are consistent with Bizjak et al. (2017) were they find that peer firms are more likely to be included if they have the same industry classification code. If peer firms bear greater common risk within the firms, the peer group can better shield executives from
19
exogenous common shocks. Likewise, Bizjak et al. (2017) find that actual peers are larger in terms of size and in market-to-book value of assets. These results are consistent with larger peers being alternatives for capital investment and are comparable in size (Gong et al., 2011). Now, looking at peer characteristics that could suggest opportunism in the actual peer group, results find that actual peers tend to have lower CAGR in stock return and have lower stock volatility. The result that actual peers have lower CAGR in stock return is consistent with Bizjak et al. (2017). However, the outcome that actual peers have lower stock volatility is inconsistent with Bizjak et al. (2017) and Gong et al. (2011). In Table II column 2 and 3, the results are shown for added and dropped peers over time. Focusing on the results that RPE firms add and drop peer firms to filter out common shocks, both column 2 and 3 indicate a negative sign for the two digit and three-digit SIC code variable. However, due to
insignificance, only an interpretation can be given to the SIC3 variable for the dropped peers, which is significant at a 1% level. The SIC3 variable indicates that actual peer firms are more likely to be dropped as a member of the RPE peer group if the peers are less likely to come from the same industry classification as the RPE firm. Other results show that added
(dropped) peers are larger (smaller) in terms of growth opportunities and smaller (smaller) in terms of firm size. According to Bizjak et. al (2017), this is consistent with the larger firms being alternatives for capital investment. While the literature is accordingly silent about firm size, Gong et al. (2011) find that selected peers with similar ability as the RPE firm are closer in size than unselected peers. This is consistent with the findings on RelSize in column 2 and 3. Focusing now on peer characteristic that could suggest potential opportunism by adding and dropping peers over time, results find that added and dropped peers tend to have lower EPS and lower AAGR of EPS compared to the RPE firm. Moreover, dropped peers are more likely to have lower stock returns and higher stock volatility. Here, all explanatory variables are significant at a 1% level, except for the RelEPS variable which is significant at a 10% level for added peers. The positive and significant sign of the variables RelEPS and
RelPastEPS for added peers could suggest opportunism. At the same time, peers that are
dropped from the peer group have higher volatility compared to the RPE firm. If RPE firms want to increase award payout this would mean that the peers dropped from the peer group would have lower volatility. This contradicts for opportunistic in peer selection. Taken together, the results presented in Table II verify prior studies that peers are selected to filter out common shocks. However, the results for potential opportunism are ambiguous.
According to the literature, if firms use RPE for rent-seeking behavior the selected peers are expected to underperform compared to the RPE firm to increase award payout.
20
Table II
Table II presents the maximum likelihood estimates from a logistic regression for the various aspects related with the tendency for an RPE firm to select other peer firms as a member of its RPE peer group. Each RPE firm-year is matched with all possible candidate firms in the created universe. RelRet is the difference in stock returns. RelPastRet is the compounded annual growth rate of stock returns for the prior three years. RelEPS is the difference in earnings-per-share. RelPastRet is the annual average growth rate in EPS for the prior three years. RelSize is the difference in firm size. Firm size is measured as the natural logarithm of total assets. RelVol is the difference in stock volatility. RelMTB is the difference in
market-to-book ratio of assets. SIC2 and SIC3 are dummy variables indicating a one if the peer firm starts with the same two-digit or three-digit SIC code, respectively. All variables starting with “Rel” are measured as the difference between the RPE firm characteristics minus the same characteristics for the potential peer firm All variables are performed on RPE firms. The dependent variable in column (1) is one if the potential peer firm is an actual member of the RPE peer group by the RPE firm and zero otherwise. The dependent variable in column (2) equals one if the added peer firms are a member of the peer group but was not a member of the peer group in the previous year and zero otherwise. In column (3) the dependent variable equals one if the members of the peer group from the previous year are not a member of the peer group for the observation year and zero otherwise. All continuous variables are Winsorized at the 5th and 95th
percentile level. Standard errors are calculated after adjusting for firm-level clustering. Significance is presented by ***, **, and * at 1%, 5%, and 10%, respectively.
(1) (2) (3)
Actual Peers Added Peers Dropped Peers
RelRet 0.322 0.0900 0.813*** (0.485) (0.0700) (0.177) RelPastRet 0.0134*** -0.000645 -0.000652 (0.00383) (0.000431) (0.00120) RelEPS -0.00103** 0.00006* 0.00137*** (0.000486) (0.00003) (0.000112) RelPastEPS -0.000356 0.000141*** 0.00144*** (0.000567) (0.00004) (0.000134) RelSize -0.00942*** 0.000122*** 0.00520*** (0.000529) (0.00003) (0.000128) RelVol 0.151** -0.00629 -0.578*** (0.0762) (0.00797) (0.0306) RelMTB -0.00562*** -0.000350*** 0.00301*** (0.000780) (0.00006) (0.000222) SIC2 0.342*** -0.000192 -0.00105 (0.0130) (0.000120) (0.000757) SIC3 0.252*** -0.000655 -0.0101*** (0.0145) (0.000479) (0.000979) Observations 143.696 70.170 118.853 R-squared 0.371 0.025 0.017
Firm FE Yes Yes Yes
Year FE Yes Yes Yes
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
21
All three columns provide little evidence of strategic selecting poor performing peers into the RPE peer group. A potential explanation can be given. Table II presents a sample which consists of all RPE firms which performance metrics are based on stock returns and/or accounting measures. So, it is possible that a considerable part of the sample uses relative performance evaluation to improve information about managerial effort and ability and a small fraction may use it strategically.
The findings of Table II, together with results from earlier studies (Gibbons & Murphy, 1990; Murphy, 2001; Bebchuk & Fried, 2003) provide sufficient evidence to motivate research on the composition of peer groups that could suggest opportunism.
According to the literature, firms should select RPE performance metrics which motivates the executive the best to maximize firm value (e.g. Wallace, 1997; Marquardt & Wiedman, 2007; Young & Yang, 2011). For the following table, regressions are tested if peer selection
conforms to efficient contracting and/or may suggest opportunism in the compositions of peer groups related to different performance metrics used among RPE firms. Table III presents the results of RPE firms which use earnings-per-share (EPS) or total shareholder return (TSR) as the performance metric. Column 1 of Table III present the results of the actual peers in the peer group by the RPE firm which use EPS as a performance metric. The variables of interest that might suggest opportunism are the relative EPS (RelEPS) and relative AAGR of EPS (RelPastEPS). According to the literature, RPE firms tend to select peers which they expect to outperform. In case of opportunism, added (dropped) peers should have lower (higher) EPS or lower (higher) AAGR in EPS, if EPS is the performance metric. Looking at the results of column 1, the actual peers in the peer group operate in the same industry and are larger in firm size. The significant results of the variables SIC2 and SIC3 provide evidence that peer firms are selected to capture common exogenous shocks. All other variables are insignificant and can’t be interpreted. Looking at Table III column 2, actual peers that are dropped from the peer group by the RPE firm tend to have higher stock volatility. The results of column 1 and 2 contradict the potential that RPE firms, based on EPS performance
metrics, strategical select peers opportunistically but provide evidence for efficient
contracting. The subsample used to analyze the composition of potential peer selection based on the performance metric EPS consist of a small number of observations. As a consequence, the regression for the dependent variable AddedPeerEPS isn’t reported in Table III. Reason for this is that no potential peers were added to the actual peer group over time. Due to these small number of observations, findings indicate that the composition of peer groups based on the performance metric EPS are relatively stable over time and peer turnover is low. A
22
potential reason for the small number of observations is that the majority of RPE firm rely solely on price-based performance metrics when implementing RPE.4 In addition, EPS is one of many accounting measures a RPE firm can select as the performance metric. Bizjak et al. (2017) found little evidence that RPE firms select peers opportunistically when accounting is the performance metric. These results were based on all measures of accounting performance metrics and not solely on EPS.
Turning to RPE firms which use total shareholder return as a grant over the award performance period. Here, the variables of interest that might suggest opportunism are relative stock return and relative CAGR stock return. In case of opportunism, added (dropped) peers should have lower (higher) stock returns or lower (higher) CAGR of stock returns, if TSR is the performance metric. For this analysis, the focus is on columns 3, 4 and 5. Table II column 3, find that RPE firms include peers which have lower stock return, lower CAGR in stock return, higher AAGR in EPS, are larger in terms of size, lower stock
volatility, and higher market-to-book ratio. In addition, column 3 results show that RPE firms select peers that operate in the same industry classification code. All variables are significant at a 1% level. The significant results of the variables SIC2 and SIC3 provide evidence that peers are chosen that have the ability to remove common risk. These results are consistent with Bizjak et al. (2017). According to column 4 and 5, RPE firms add potential peers to the peer group if AAGR in EPS are smaller and the potential peer firms operate in the same industry. RPE firms tend to drop peers if the peer firms have lower CAGR in stock return, higher EPS, larger in terms of size, and lower growth opportunities. All variables are significant at a 1%, 5% or 10% level. Column 3, provide evidence that peers are selected to filter out common shocks. More interesting, focusing on the dropped peers, column 5 provides little evidence that peers are dropped strategically. The variable of interest
RelPastRet might suggest that peers are dropped if the CAGR of stock return is larger
compared to the RPE firm. This weak evidence provides support that RPE firms drop peer firms for strategic purposes when the performance metric is TSR. Consistent with Gong et al. (2010) this weak evidence could reflect the increased public scrutiny which restricts firms from rent-seeking behavior in peer selection.
To summarize, the results provide evidence that RPE firms, based on the performance metric EPS, select peers to filter out common shocks. However, there is no empirical
4 Summary statistics presented in Table I report that the majority of RPE firms rely on price-based performance
23
evidence for opportunism. Equally, RPE firms with the performance metric TSR show empirical evidence that peers are selected to filter out common shocks. But in addition, show little evidence for opportunism that peers are dropped from the peer group if the CAGR of stock return is larger compared to the RPE firm.
24
6 Limitations
Section 6 discusses the limitations of the study. Possible shortcomings are selection-bias and survivorship bias.
The first potential shortcoming is selection bias. Faulkender and Yang (2010) and Bizjak et al. (2017) found evidence of peer selection bias in compensation benchmarking peer groups. Selection bias arises when groups are selected for analysis in a manner that
randomization is not achieved. Since the board of directors of a firm determine the selection of peers in the peer group, there could be potential for self-serving bias that could benefit executives. If selection bias isn’t taken into account, conclusions of the study may not be accurate.
The second shortcoming is the potential for survivorship bias. Although provisions are taken to mitigate survivorship bias in the index, the sample might be susceptible to survivorship bias if a significant number of firms drop out in a certain industry wherein only the survivors remain. This could result in an overestimation of past performance as poor performing firms are dropped from the database. A potential cause could be the financial crisis in 2007-2008, wherein many firms went bankrupt. If the sample selection is subjected to survivorship bias, the sample in the study may not provide a good representative of the S&P 1500 and might potentially limit the external validity.
The closest related papers to this thesis are Gong et al. (2011) and Bizjak et al. (2017). Gong et al. (2011) provide some evidence of selection bias in basic logit regressions, but were uncertain about the ultimate impact of the self-serving bias on executive compensation. Through the methodology of Bizjak et al. (2017) the selection bias can be directly examined. Bizjak et al. (2017) use three alternative sets of peer groups that are constructed using
industry-size matching and propensity scores. They find that RPE firms that use a custom peer group and TSR as the performance metric, indicate no evidence of self-serving selection bias.
25
7 Conclusion
Over the last years, a common practice in compensation benchmarking is to set the level of executive compensation to a predetermined group of peers. The main rationale firms use relative performance evaluation to benchmark compensation, is that these contracts help to filter out common shocks from firm performance that are outside the executives control and to improve the measurement of managerial ability and effort (Hölmstrom, 1979, 1982). Despite the theoretical appeal, prior literature finds mixed evidence that RPE contracts allow management to strategically select peers in order to increase award payout (e.g. Gibbons & Murphy, 1990; Murphy, 2001; Bebchuk & Fried, 2003). The results of the study contribute to the ongoing discussion if the usage of RPE conforms to efficient contracting or selecting peers opportunistically to extract rents from shareholders in compensation practices. As little research is conducted on the peer group selection in RPE awards, Bizjak et al. (2017) come with more insight. They find that peer selection is consistent with economic motivation to filter out common shocks. Furthermore, they uncover some evidence of opportunism. Results show that RPE firms using a custom peer group and if the performance metric is based on total shareholder return, that peers are selected with the expectation of underperformance relative to the RPE firm (Bizjak et al. 2017). This thesis contributes to the existing executive compensation literature by providing more intuition in the practice how peer groups are selected based on TSR and EPS performance metrics among RPE firms. Using multiple logistic regressions, the peer group composition can be studied by adding and dropping peer firms over time. The way in which peers are added and dropped provide insight if peer
selection supports the efficient contracting view or uncover rent-seeking behavior. The results indicate that the selection of peer groups is consistent with the economic motivation for relative performance evaluation. This means, RPE firms that select peer groups are more likely to operate in the same industry. This suggests that there is supportive evidence that RPE firms select peers to filter out common shocks. At the same time, the potential for opportunism is tested. RPE firms tend to drop peers if the CAGR of stock returns are larger compared to the RPE firm. This weak evidence provides support that RPE firms drop peer firms for strategic purposes when the performance metric is TSR.
The results of the thesis are most in line with the outcomes of Bizjak et al. (2017) and Gong et al. (2011) as these authors explain that peer groups are selected to filter out common shocks. However, the results in this study know limitations. Therefore, this study could be influenced by selection bias and survivorship bias.
26
Although, there is considerable literature on compensation benchmarking, there is still less research on peer group selection in RPE awards. In addition, the existing literature lacks a clear understanding of the sources of variation in firm decisions in RPE contracts based on either price-based and/or accounting performance metrics. As this study touches upon the performance metrics used in RPE awards, hopefully more research will be done to the incentives of these awards entail.
27
References
Aggarwal, R. K., & Samwick, A. A. (1999a). The other side of the trade-off: The impact of risk on executive compensation. Journal of Political Economy, 107(1), 65-105.
Albuquerque, A. (2009). Peer firms in relative performance evaluation. Journal of Accounting
and Economics, 48(1), 69-89.
Albuquerque, A. M. (2013). Do growth-option firms use less relative performance evaluation?.
The Accounting Review, 89(1), 27-60.
Antle, R., & Smith, A. (1986). An empirical investigation of the relative performance evaluation of corporate executives. Journal of Accounting Research, 1-39.
Bebchuk, L. A., Fried, J. M., & Walker, D. I. (2002). Managerial power and rent extraction in
the design of executive compensation (No. w9068). National bureau of economic research.
Bebchuk, L. A., & Fried, J. M. (2003). Executive compensation as an agency problem. Journal of
economic perspectives, 17(3), 71-92.
Bertrand, M., & Mullainathan, S. (2001). Are CEOs rewarded for luck? The ones without principals are. The Quarterly Journal of Economics, 116(3), 901-932.
Bizjak, J. M., Lemmon, M. L., & Naveen, L. (2008). Does the use of peer groups contribute to higher pay and less efficient compensation?. Journal of Financial Economics, 90(2), 152-168.
Bizjak, J., Lemmon, M., & Nguyen, T. (2011). Are all CEOs above average? An empirical analysis of compensation peer groups and pay design. Journal of Financial Economics,
100(3), 538-555.
Bizjak, J. M., Kalpathy, S. L., Li, Z. F., & Young, B. (2017). The role of peer firm selection in explicit relative performance awards.
Byrd, J. W., Johnson, M. F., & Porter, S. L. (1998). Discretion in financial reporting: The voluntary disclosure of compensation peer groups in proxy statement performance graphs.
Contemporary Accounting Research, 15(1), 25-52.
Carter, M. E., Ittner, C. D., & Zechman, S. L. (2009). Explicit relative performance evaluation in performance-vested equity grants. Review of Accounting studies, 14(2-3), 269-306.
28
De Angelis, D., & Grinstein, Y. (2014). Performance terms in CEO compensation contracts.
Review of Finance, 19(2), 619-651.
Faulkender, M., & Yang, J. (2010). Inside the black box: The role and composition of compensation peer groups. Journal of Financial Economics, 96(2), 257-270.
Garvey, G., & Milbourn, T. (2003). Incentive compensation when executives can hedge the market: Evidence of relative performance evaluation in the cross section. The Journal of
Finance, 58(4), 1557-1582.
Gibbons, R., & Murphy, K. J. (1990). Relative performance evaluation for chief executive officers. ILR Review, 43(3), 30-S.
Gong, G., Li, L. Y., & Shin, J. Y. (2011). Relative performance evaluation and related peer groups in executive compensation contracts. The Accounting Review, 86(3), 1007-1043. Hölmstrom, B. (1979). Moral hazard and observability. The Bell journal of economics, 74-91. Hölmstrom, B. (1982). Moral hazard in teams. The Bell Journal of Economics, 324-340. Hölmstrom, B., & Milgrom, P. (1987). Aggregation and linearity in the provision of
intertemporal incentives. Econometrica: Journal of the Econometric Society, 303-328.
Jayaraman, S., Milbourn, T. T., Peters, F. S., & Seo, H. (2017). Product market peers and relative performance evaluation.
Jensen, M. C., & Murphy, K. J. (1990). Performance pay and top-management incentives.
Journal of political economy, 98(2), 225-264.
Lewellen, S. (2013). Executive compensation and peer effects. Unpublished working paper. Yale
University.
Lazear, E. P., & Rosen, S. (1981). Rank-order tournaments as optimum labor contracts. Journal
of political Economy, 89(5), 841-864.
Murphy, K. J. (2000). Performance standards in incentive contracts. Journal of Accounting and
Economics, 30(3), 245-278.
Marquardt, C. A., & Wiedman, C. I. (2007). Disclosure, incentives, and contingently convertible securities. Accounting Horizons, 21(3), 281-294.
29
Wallace, J. S. (1997). Adopting residual income-based compensation plans: Do you get what you pay for?. Journal of Accounting and Economics, 24(3), 275-300.
Westphal, J. D., & Zajac, E. J. (1995). Who shall govern? CEO/board power, demographic similarity, and new director selection. Administrative science quarterly, 60-83.
Young, S., & Yang, J. (2011). Stock repurchases and executive compensation contract design: The role of earnings per share performance conditions. The Accounting Review, 86(2), 703-733.
Zajac, E. J., & Westphal, J. D. (1994). The costs and benefits of managerial incentives and monitoring in large US corporations: When is more not better?. Strategic management
