Dynamic monitoring under resource constraints

Often monitoring agencies (principal) do not have enough resources to monitor all agents, and violations are unavoidable. Questions arise regarding the structure of the monitoring scheme that minimizes the rate of violations. In dynamic monitoring problems, the principal can use the past behavior of agents to determine her monitoring policy. In this paper, we identify the optimal dynamic monitoring scheme when the principal has a commitment power, and show that in this scheme agents first "compete" in a tournament, where the one who is monitored more frequently wins. The winner of the tournament then enjoys lower monitoring intensity, and violates more in the long run. (C) 2021 Elsevier Inc. All rights reserved.