Heterogeneity of ex post fairness and ex ante fairness

We present an axiomatization of the preferences of a decision maker who is partially self-interested and cares about both ex ante and ex post fairness in situations involving risk and social interactions. Our work highlights the possibility of variations among the agents in the extent to which the decision maker is concerned with ex ante fairness relative to ex post fairness, which we explain using our axiomatized preferences. These variations cannot be explained by existing models including the EIA model proposed by Saito (American Econ. Rev. 103:3084–3101, 2013). Furthermore, we identify an important axiomatic difference between our model and the EIA model. Our model includes all preferences prioritizing ex ante fairness, which cannot be incorporated into the EIA model. Our model is tractable and testable through laboratory experiments, as all parameters can be determined experimentally. As a result, our model provides a theoretical foundation for understanding fairness-related behavior in groups during risky situations. It is applicable to a variety of games in risky and social situations.