In the long-run we are all dead: on the benefits of peer punishment in rich environments

We investigate whether peer punishment is an efficient mechanism for enforcing cooperation in an experiment with a long time horizon. Previous evidence suggests that the costs of peer punishment can be outweighed by the benefits of higher cooperation if (i) there is a sufficiently long time horizon and (ii) punishment cannot be avenged. However, in most instances in daily life, when individuals interact for an extended period of time, punishment can be retaliated. We use a design that imposes minimal restrictions on who can punish whom or when, and allows participants to employ a wide range of punishment strategies including retaliation of punishment. Similar to previous research, we find that, when punishment cannot be avenged, peer punishment leads to higher earnings relative to a baseline treatment without any punishment opportunities. However, in the more general setting, we find no evidence of group earnings increasing systematically or significantly over time relative to the baseline treatment. Our results raise questions under what conditions peer punishment can be an efficient mechanism for enforcing cooperation.