Indicator Choice in Pay-for-Performance

We study the classic principal-agent model when the signal observed by the principal is chosen by the agent. We fully characterize the optimal information structure from an agent's perspective in a general moral hazard setting with limited liability. Due to endogeneity of the contract chosen by the principal, the agent's choice of information is non-trivial. We show that the agent's problem can be mapped into a geometrical game between the principal and the agent in the space of likelihood ratios. We use this representation result to show that coarse contracts are sufficient: The agent can achieve her best with binary signals. Additionally, we can characterize conditions under which the agent is able to extract the entire surplus and implement the first-best efficient allocation. Finally, we show that when effort and performance are one-dimensional, under a general class of models, threshold signals are optimal. Our theory can thus provide a rationale for coarseness of contracts based on the bargaining power of the agent in negotiations.