Altruistic punishment and the threat of feuds

Altruistic punishment may promote cooperation, but can also lead to costly feuds. We examine how the threat of feuds affects individuals' willingness to engage in altruistic punishment in a public good experiment in which the number of stages is determined by participants' actions. The design imposes minimal restrictions on who can punish whom and when, and therefore allows participants to use a range of punishment strategies. We find that participants recognize the threat of feuds and respond to it by employing strategies that prevent their breakout. When feuds can span several periods, the extent of altruistic punishment is greatly reduced. This leads to progressively lower levels of cooperation and earnings relative to a baseline treatment where punishment cannot be avenged.