The Short Arm of Guilt – An Experiment on Group Identity and Guilt Aversion

In a laboratory experiment, I test whether guilt aversion, i.e., a preference to fulfill the expectations of others, plays out more strongly if agents share an induced social identity. Participants play a dictator game in which they can condition their amount sent on recipients' beliefs. Dictators either play with a recipient from their own group (ingroup treatment) or from the other group (outgroup treatment). I find that the positive influence of second-order beliefs on how much a dictator sends is stronger in the ingroup treatment. However, the way dictators react to very high expectations does not differ significantly between treatments. In contrast to previous work I do not find that amounts sent are an inversely u-shaped function of recipients' expectations. Rather, and independently of the treatment, participants tend to ignore very high expectations