Robust Contracting for Sequential Search

A principal contracts with an agent who can sequentially search over projects to generate a prize. The principal knows only one of the agent's available projects and evaluates a contract by its worst-case performance. We characterize the set of robustly optimal contracts, all of which involve a minimum debt level, i.e., the agent only receives payment if the prize exceeds a certain threshold. This debt requirement is essential to prevent the agent from terminating the search too early. Our characterization encompasses several commonly observed contract formats, including pure debt, debt-plus-equity, and tranches. We also study situations where each of these contracts emerges as the unique prediction. In contrast to much of the existing robust contracting literature, linear contracts are strictly sub-optimal because they dampen the agent's search incentive.