Robust Contracting Under Distributional Uncertainty

ABSTRACT We study contract design when the principal has limited information about the output distributions induced by the agent's actions. In a baseline model where only the means are known, we show that increasing affine contracts are robustly optimal. The mean restrictions accommodate a wide range of output distributions, including extreme cases that help establish this optimality. We then extend the analysis to environments with additional constraints on the distributions. Our main result shows that the robust optimality of increasing affine contracts persists even when the principal knows more—for example, that each action induces a distribution with full support.