Cooperation with Network Monitoring

This paper studies the maximum level of cooperation that can be sustained in perfect Bayesian equilibrium in repeated games with network monitoring, where players observe each other's actions either perfectly or not at all. The foundational result is that the maximum level of cooperation can be robustly sustained in grim trigger strategies. If players are equally well monitored, comparative statics on the maximum level of cooperation are highly tractable and depend on the monitoring technology only through a simple statistic, its effective contagiousness. Typically, cooperation in the provision of pure public goods is greater in larger groups, while cooperation in the provision of divisible public goods is greater in smaller groups, and making monitoring less uncertain in the second-order stochastic dominance sense increases cooperation. For fixed monitoring networks, a new notion of network centrality is developed, which determines which players cooperate more in a given network, as well as which networks support greater cooperation.