Network Monitoring and Punishment in Public Goods Experiments

We report experimental findings on the impact of network structure on decentralized monitoring and punishment in public goods games. In the environment we study, individuals can only directly monitor and punish their immediate neighbors in an exogenously determined network. We examine contributions and punishment decisions in a Complete network, a Circle network, and an Asymmetric network. Average contributions are lower in the Asymmetric network, although this result is driven entirely by the player who faces only one potential punisher. We also examine whether asymmetry in the network leads some punishers to discriminate between their potential targets. After controlling for targets' contribution decisions, we find limited support for this hypothesis. However, the data indicate that some punishers may be deterred from issuing discriminatory punishment by undermonitored targets who retaliate against previous punishment more often than others. Thus, we identify an additional complication of asymmetry in the network - that it may facilitate more targeted revenge by under-monitored players.