Robust Performance Evaluation of Independent and Identical Agents

TL;DR

This paper investigates the design of optimal contracts for identical, independent agents when the principal cannot observe actions or know all available actions, revealing that nonlinear, performance-dependent incentives are necessary for worst-case optimality.

Contribution

It demonstrates that worst-case optimal contracts are inherently nonaffine and require agents' pay to depend on each other's performance, introducing a new channel for nonlinear team incentives.

Findings

Worst-case optimal contracts are nonaffine in performances.

Agents' pay depends on the performance of others.

Existence of such contracts is proven for two agents with binary output.

Abstract

A principal provides nondiscriminatory incentives for independent and identical agents. The principal cannot observe the agents' actions, nor does she know the entire set of actions available to them. It is shown, very generally, that any worst-case optimal contract is nonaffine in performances. In addition, each agent's pay must depend on the performance of another. In the case of two agents and binary output, existence of a worst-case optimal contract is established and it is proven that any such contract exhibits joint performance evaluation -- each agent's pay is strictly increasing in the performance of the other. The analysis identifies a fundamentally new channel leading to the optimality of nonlinear team-based incentive pay.