How to organize monitoring and munishment: Experimental evidence

Punishment institutions for curtailing free-riding in social dilemmas rely on information about individuals' behavior collected through monitoring. We contribute to the experimental study of cooperation-enhancing institutions by examining how cooperation and efficiency in a social dilemma change in response to varying how monitoring and punishment are jointly organized. Specifically, we evaluate - against a no-monitoring baseline - combinations of two imperfect monitoring regimes (cen-tralized vs. decentralized) and three punishment regimes (self- vs. peer- vs. del-egated punishment) in a repeated public goods game. As hypothesized, we find that delegated punishment outperforms other punishment regimes, irrespective of the monitoring regime, both in terms of cooperation and efficiency. Monitoring, both centralized and decentralized, cannot raise cooperation relative to the baseline unless accompanied by a credible punishment. When combined with a punishment institution, both monitoring regime outperforms the baseline.