The use of incentive-compensation practices based on relative performance has grown dramatically in recent years (see Figure 1). As of 2019, more than 50% of large U.S. firms base executive compensation in part on their firms’ performance relative to peer firms, compared to just 20% in 2006. Further, relative total shareholder return (“TSR”) is now the single most prevalent long-term performance metric used by large U.S. firms in their executive pay plans (Equilar, 2020 and Meridian Compensation Partners, 2019). However, despite the roughly 150% increase in the use of relative performance evaluation (“RPE”) since 2006, almost half of firms still opt not to use relative performance metrics in their executives’ incentive-compensation plans. This is noteworthy, particularly in light of RPE’s ease of employ, attractive theoretical properties vis-à-vis risk-shielding, and status as a highly recommended governance tool by proxy advisors. This observation motivates our recent study, in which we examine firms’ RPE choices, with a particular emphasis on factors that lead firms to forgo using RPE. We find that a key factor discouraging firms’ reliance on RPE is the lack of an effective RPE peer group; for some firms, RPE simply does not confer much benefit due to the unavailability of comparable peers. Beyond this finding, we also document how strategic competition and agency conflicts influence the decisions to employ RPE and to choose effective peer groups.
Figure 1. This figure illustrates the use of relative performance evaluation in CEO compensation plans at large U.S. firms over time (data are from Incentive Lab).
Relative Performance Evaluation
The key feature of a relative performance plan is that executives’ pay is determined based on the performance of their firm compared to the contemporaneous performance of other firms, rather than a pre-established absolute performance target. For example, a restricted stock award that vests if the firm’s three-year TSR exceeds 10% provides an absoluteperformance incentive. In contrast, a restricted stock award that vests if three-year TSR exceeds the contemporaneous return of a selected group of industry peers provides a relative performance incentive. In the RPE case, the exact TSR needed for vesting is not known in advance, but rather depends on the peer group’s performance.
The primary benefit of RPE is that it shields an executive’s pay from random and uncontrollable factors that simultaneously affect performance at the executive’s firm, as well as at other firms. An example of such a random and uncontrollable factor is the COVID-19 pandemic, which had a substantial simultaneous negative impact on performance at many firms (e.g., airline stocks). Judged on an absolute scale, these firms would appear to be performing poorly. But, evaluated in the context of peer firms, a moderate decline in stock price might actually constitute strong performance (if peers’ stock prices decline by a greater degree).
RPE helps boards better monitor the executive’s job performance by filtering out the effects of uncontrollable external factors. By measuring performance relative to a set of firms that are similarly affected by the shock, performance trends that are common across firms are stripped away, shielding the executive from this source of risk and making it easier to ascertain how effectively the executive is running the firm.
However, not all firms have ready access to a peer group with similar risk exposures. For example, a firm that occupies an unusual product market niche, or otherwise has an atypical business model, will likely struggle to find peers with similar risk exposures, thus reducing the benefits of RPE. Such firms may decide to use a generic peer group (e.g., the S&P 500 index) in an RPE plan or may choose to forgo the use of RPE altogether in their executive compensation plan due to the lack of an effective peer group. We evaluate firms’ peer group opportunity sets and dub the availability of an effective peer group “peer availability.”
We examine three main questions related to peer availability: First, does peer availability influence the firm’s choice of whether to include RPE in the CEO’s pay plan? Second, conditional on having an effective peer group in their opportunity set, what frictions or incentives might explain a firm’s decision not to use RPE in their CEO’s pay plan? Third, conditional on using RPE in the CEO’s pay plan, do firms construct the most effective peer group available, and if not, what frictions or incentives might explain peer selection?
To measure peer availability, we develop a novel peer selection algorithm that constructs peer groups that are as similar as possible to the focal firm, in terms of their risk exposures. This approach has strong theoretical backing (see, for example, Holmström, 1982) and is based on best practices put forward by compensation consultants (see, for example, Meridian Compensation Partners LLC., 2016). Simply put, our algorithm searches for the combination of potential peer firms that has the highest stock return correlation with the focal firm. We call this optimal combination of peer firms the “artificial peer group.” We then evaluate how well these artificial peer groups explain the focal firm’s performance over the subsequent 36-month period by calculating the stock return correlation over the next 36 months, out-of-sample. This out-of-sample correlation is our primary measure of peer availability; a higher value indicates the availability of a more effective peer group. We find that the algorithmically-constructed artificial peer groups are highly similar to the RPE peer groups firms actually choose to use, suggesting that our algorithm is effective at mimicking the peer selection process.
We find that firms are significantly more likely to use peer-based RPE when an effective peer group is readily available. Peer availability is approximately 40% greater among firms that choose to construct RPE groups, compared to firms that choose not to. This finding suggests that lack of access to a viable peer group may be a driving factor underlying many firms’ non-reliance on RPE. Moreover, we find that in most cases, when firms do choose to use RPE, they tend to construct a peer group that is about as effective as possible at shielding executives from risk. These findings are consistent with the notion that, by and large, RPE is used as a risk-shielding tool and firms construct peer groups with the aim of shielding managers from systematic sources of risk/uncertainty.
With regard to the second question, we find that our sample also contains a fairly large subsample of firms for which a highly effective peer group is available, yet the firm chooses to forgo using peer-based RPE. We posit that non-reliance on RPE in these cases is related to strategic product market considerations. Such concerns arise when firms are competing in concentrated industries, because giving executives explicit incentives to perform better than a group of competitors can incentivize costly sabotage, whereby the executives seek to harm peers, rather than focusing on maximizing their own firm’s value (also see, for example, Bloomfield, Marvão, Spagnolo, 2021 and Feichter, Moers, Timmermans, 2021). Consistent with this supposition, we find that non-reliance on RPE when an effective peer group exists occurs primarily in highly concentrated product markets, where competitive sabotage is more likely to be a concern. Thus, it seems that some firm choose to forgo the potential risk-shielding benefits associated with effective RPE to avoid adverse competitive consequences associated with using RPE.
With regard to the third question, we also identify a group of firms that choose to use RPE, but construct peer groups that are substantially less effective than their opportunity set allows. That is, they appear to construct peer groups that are demonstrably inferior to an available alternative from the standpoint of effective governance. While such cases could be written off as mere cases of incompetence, we posit that reliance on RPE in these cases might be related to opportunism and rent-seeking. Specifically, we consider the possibility that firms construct ineffective, but easy-to-beat peer groups as a subtle method of providing excess compensation to the executive, while maintaining the guise of “pay for performance.” RPE is likely a particularly effective approach to this, because RPE is viewed very favorably by investors and proxy advisors as an indicator of good governance practices (see, for example, Glass Lewis & Co., 2020 and Institutional Shareholder Services Inc., 2020).
Consistent with a rent-extraction explanation, we find that firms which benchmark against low quality peer groups tend to outperform their actual peer group more frequently than they would have outperformed their artificial peer group. We further document that this pattern is stronger for firms with more powerful/entrenched managers and weaker governance mechanisms in place. These additional findings are consistent with the notion that firms purposely decide to benchmark against an ineffective but easy-to-beat RPE peer group, in order to award excess compensation to their executives.
Collectively, our results indicate that most firms construct RPE peer groups that are consistent with a risk-shielding objective, but not all firms appear to do so. Firms tend to choose whether or not to use RPE based on the availability of a viable peer group, and conditional on choosing to use RPE, most firms construct the most effective peer group they can. However, this logic cannot explain the full gamut of observed RPE choices. There are instances in which firms do not rely on RPE for strategic reasons, and there are instances in which firms use RPE, but do so in an ineffective manner, for opportunistic reasons.
Our study adds to an existing body of research that examines firms’ RPE practices, a topic of growing importance given the rising prevalence of this incentive-compensation practice in executive pay packages. Despite the ubiquity of these plans, much about their use cases remains unknown. We depart from prior work by considering the role of the peer group opportunity set, which has not been examined in the existing literature. We develop an algorithm that mimics the RPE peer selection process, and thereby allows us to observe and measure each firm’s opportunity set which we use as a counterfactual to assess firms’ RPE choices. Notably, our algorithm can be used regardless of whether or not the firm actually uses RPE, allowing us to analyze the peer group opportunity sets for RPE users, and non-users, alike. In this way, our algorithm allows us to delve into the black box of RPE and shed new light on firms’ actual RPE choices.
Matthew Bloomfield is an Assistant Professor of Accounting at the Wharton School of the University of Pennsylvania.
Wayne Guay is the Yageo Professor of Accounting at the Wharton School of the University of Pennsylvania.
Oscar Timmermans is a doctoral student at Maastricht University.
This post is adapted from their paper, “Relative Performance Evaluation and the Peer Group Opportunity Set,” available on SSRN.