Signals of ability in an agency model

This paper analyses the trade-offs associated with relying on performance versus ability measures in executive compensation. We propose a principal-agent model with moral hazard and adverse selection in which the principal designs the compensation scheme to be contingent on the outcome of interest to the principal along with a noisy signal of the agent's ability. The signal of ability may include traditional measures of human capital, and information gleaned from private interactions with the agent. We show that, under empirically plausible conditions, the weight placed on the signal of ability is negative, while the weight placed on the outcome is positive; and an increase in one weight tends to be associated with a reduction in the other weight. Consequently, the principal may in fact prefer agents with inferior qualifications, rejecting those who are 'overqualified'; conversely, she may select an agent with a distinguished pedigree, rejecting agents who are 'underqualified'.