The anatomy of distributional preferences with group identity

This paper dissects distributional preferences with group identity in a modified dictator game. I estimate individual-level utility functions with two parameters that govern the trade-offs between equity and efficiency and giving to self and to other. Subjects put on average less weight on income of the out-group, but overall only a minority behaves completely selfishly. Giving to the out-group also renders subjects more accepting of inequality. However, the experiment also uncovers a large heterogeneity of preferences. It seems that those who are social become slightly less social in the presence of the out-group. The number of selfish individuals is instead hardly affected. Moreover, choices in both treatments overwhelmingly stem from well-behaved, yet systematically different underlying social preference functionals. Hence this experiment suggests that the rational choice approach, which is predominantly used in the literature, is a useful tool to understand the effect of group identity on social preferences. As a side result, I find that the weight on self, but not the individual equity-efficiency trade-off predicts political left-right self-assessment as more conservative voters are more selfish. I also document gender differences: females put less weight on self, are more inequality averse and react more strongly to the treatment.