An Algorithmic Approach to Understanding Firms’ Use of Relative Performance Evaluation

Abstract

We explore firms’ choices to include or exclude relative performance evaluation (“RPE”) from their CEOs’ pay plans and, conditional on using it, examine their peer selections. Specifically, we assess the degree to which RPE choices are related to risk-sharing objectives versus other considerations. To do so, we develop an algorithm that constructs artificial RPE peer groups, designed to maximize risk-sharing effectiveness, and use them as counterfactuals against which to benchmark firms’ choices. We find that RPE choices are largely, but not entirely, consistent with risk-sharing objectives: firms use RPE when doing so is more helpful for risk-sharing, and typically construct peer groups that close align with pure risk-sharing considerations. However, we also document three systematic departures from effective risk-sharing: (1) firms in more concentrated industries—where costly competitive sabotage is more likely a concern—forgo RPE altogether or use peer groups comprised of peers from other industries; (2) some RPE-using firms—especially those with weaker governance—sacrifice risk-sharing by padding their peer groups with easy-to-beat peers, facilitating excess compensation; and (3) some RPE-using firms appear to choose peers using simplistic heuristics, leading to less effective peer groups.

Keywords: relative performance evaluation, peer groups, executive compensation, product market competition, rent extraction

JEL Classification: G30, J33, L1, M12, M52