Estimating Pay-for-Performance Sensitivity in the Presence of Relative Performance Evaluation
43 Pages Posted: 17 Aug 2011 Last revised: 16 Apr 2020
Date Written: April 14, 2020
We study how a compensation researcher’s empirical use of a net performance measure (i.e., after deducting compensation expense) instead of a gross performance measure (i.e., before deducting compensation, consistent with principal-agent theory) affects inferences regarding managerial effort incentives. We analytically demonstrate that as long as boards use relative performance evaluation, it is irrelevant whether managerial compensation is based on gross or net performance because a researcher can construct an unbiased estimate of managerial incentives from interpreting compensation regressions. However, we also show that the direct coefficient on a net performance measure in a compensation regression will, in theory, yield either an upward or downward bias in the estimate of managerial incentives. We provide empirical evidence of, on average, a downward bias in a cross-sectional estimation of pay-for-performance sensitivity, which also suggests a possible explanation for surprisingly weak CEO incentives interpreted in prior studies.
Keywords: Executive Compensation, Relative Performance Evaluation, Performance Measurement
JEL Classification: J33, M40, M46
Suggested Citation: Suggested Citation