Non-Strategic Punishment when Monitoring is Costly: Experimental Evidence on Differences between Second and Third Party Behavior

This paper studies monitoring and punishment behavior by second and third parties in a cooperation experiment with endogenous information structures: Players are uninformed whether the target player cooperated or defected at the cooperation stage, but can decide to resolve the information imperfection at non-negative cost at the punishment stage. We examine how monitoring and punishment respond to changes in monitoring costs, and exploit the evidence to gain new insights about commonalities and differences between second and third party behavior. We establish three effects of positive monitoring costs relative to the zero-cost baseline and find that each one affects third parties differently than second parties: A «direct punishment cost effect» (the supply of non-strategic punishment decreases), a «blind punishment effect» (players punish without resolving the information imperfection) and a «diffusion effect» (defectors make up a smaller share of the punished and receive weaker punishment). The first effect affects third parties less, the other two more. As a result, third party punishment leads to increasingly weaker incentives for cooperation relative to second party punishment as monitoring costs rise. In addition, the differences between second and third parties suggest the presence of a «pure role effect»: Taking into account elicited beliefs and risk preferences, third parties punish differently from second parties, not just more weakly.