rTSR: Properties, Determinants, and Consequences of Benchmark Choice

Abstract

We examine the properties, determinants, and consequences of peer benchmarks chosen by firms to evaluate relative total shareholder returns (rTSR) in executive relative performance contracts. Among firms that explicitly use rTSR, 60% select specific peer firms while 40% select a stock index as benchmarks. Consistent with the intent to filter out the systematic component of TSR, firms' chosen benchmarks exhibit a return-regression slope coefficient of 1 and remove a significant amount of systematic noise in TSR. However, index-based benchmarks are considerably noisier compared to those based on specific peers. Inconsistent with standard contracting models, firms using index-based benchmarks do not exhibit relatively lower pay-to-performance sensitivities, nor do they face lower gains from filtering precision. Instead, index-benchmark use is associated with weaker corporate governance and compensation consultants' preferences, which are uncorrelated with observable firm attributes. The use of index-based benchmarks is also associated with lower ROA, even after controlling for benchmarks' noisiness and companies' governance attributes. Our analyses suggest the salience of peer comparisons as an important attribute of relative performance benchmarks.

Keywords: Relative TSR; Measurement error; Systematic risk; Compensation consultants; Style effects; Benchmark salience

JEL Classification: G30, J33, M12, M52