Adaptive Incentive Design

We apply control theoretic and optimization techniques to adaptively design incentives for principal-agent problems in which the principal faces adverse selection its interaction with multiple agents. In particular, principal's objective depends on data from strategic decision makers (agents) whose decision-making process is unknown a priori. consider both cases where agents play best response one another (Nash) they employ myopic update rules. By parametrizing agents' utility functions offered, we develop an algorithm that can learn processes while simultaneously designing change their more desirable. provide convergence results this noise-free noisy present illustrative examples.