Reputation and Multilateral Punishment under Uncertainty

Principal-agent problems can reduce gains from exchange available in long distance trade. One solution to mitigate this problem is multilateral punishment, whereby groups of principals jointly punish cheating agents by giving them bad reputations. But how does such punishment work when there is uncertainty regarding whether an agent actually cheated or was just the victim of bad luck? And how might such uncertainty be mitigated---or exacerbated---by non-observable, pro-social behavioral characteristics? We address these questions by designing a simple modified trust game with uncertainty and the capacity for principals to employ multilateral punishment. We find that a modest amount of uncertainty increases overall welfare because principals are more willing to trust agents with bad reputations