Relative Weights on Measures of Past Performance

Abstract

To investigate the role of measures of past performance and accomplishments in determining compensation, we develop a principal-agent model where the agent's ability is not observed directly by the principal. To the extent managerial ability persists over time, past performance and accomplishments signal how the agent's innate ability is likely to affect future performance which, in turn, should affect his value to the firm as well as the wage level he can command in the labor market for his services. We show that the principal's optimal solution for employing and compensating the agent can be implemented by a simple take-it-or-leave-it offer mechanism in which the optimal compensation level can be expressed as a nonlinear function of two linear aggregates of measures of past performance. We show that the optimal aggregate measure places a greater weight on measures of past performance that are more precise and more sensitive to the agent's ability. The results are analogous to those in moral hazard models, but they stem from the posterior expectation about the agent's ability in creating value for the firm and in commanding a higher market price rather than the slope of the loglikelihood function in inferring his unobservable effort.