Punishment and disclosure probabilities in an experimental deception game

Abstract

Previous findings have shown tha t punishment does not necessarily reduce deception i n principal agent - relationships . We shed further light on this issue by first identifying a punishment mechanism that substantially decreases deception in a sender - receiver game: the possibility of imposing severe sanctions that are cost - free for the enforcer. Keeping this effective combination of punishment costs and severity constant , we then investigate how a reduction in monitoring affects deception by c ompar ing assured revelation of s ender behavior ex post with a treatment in which it is disclosed with just 50% probability . We find a similarly strong deterrence effect in both treatments suggesting that punishment works in a part icular way in the deception context: o nce it is a credible threat, it does not require complete monit oring to be effective . We also find that receivers show s imilar trust level s in senders’ messages for both punishment treatments , which are significantly higher than in the corresponding baseline s without sanctions , further support ing ou r conclusion.