Relative Performance Benchmarks: Do Boards Get it Right?

• In Event: Relative Performance Evaluation

Abstract

Standard principal-agent models predict that the board will design incentive contracts that filter out common shocks in performance in order to motivate costly effort from the CEO -- a process that entails the judicious selection of benchmarks for relative performance evaluation (RPE). This paper i) evaluates the efficacy of firms' choice of RPE benchmarks in removing common shocks in stock returns against a normative benchmark, ii) assesses the determinants of benchmarking efficacy relative to optimal contracting, and iii) quantifies performance implications using both structural and reduced-form approaches. We find that firms which choose index-based RPE benchmarks perform 14% worse relative to a normative state-of-the-art performance peer set in their ability to explain the time-series variation in returns, and at least 16% greater variance in the measurement error in the common component of performance. Firms which choose specific peers as benchmarks only modestly under-perform. Rejecting determinants from optimal contracting, index-based benchmarking instead is related to governance characteristics such as busy boards. Based on estimates from a principal-agent model, across all firms, these poorer benchmarks imply an average performance penalty in returns of 60 to 153 basis points annually through their effect on the manager's effort. In the cross-section, index-based benchmarking firms also have lower realized ROA (80 bps) and stock returns (320 bps). Collectively, our results provide new evidence on boards' selections of RPE benchmarks in CEO incentive contracts and their implications for firm performance.

Authors

• Paul Ma, University of Minnesota-Minneapolis

• Jee-Eun Shin, Harvard University Business School

• Charles CY Wang, Harvard University Business School