A Bliss Point Model of Distributional Preferences

In this study, I propose a new model of distributional preferences, the “Bliss Point†model. According to this model, each person has a preferred payoff for others which is a linear function of his own payoff, and he suffers a utility penalty proportional to the distance between this bliss point for his co-players and the actual payoffs. To evaluate the empirical relevance of this model, I create an experiment in which subjects play several games which affect both their own payoff and that of an anonymous partner. I used their behavior to estimate individual-level parameters of my model as well as three others: inequality aversion (Fehr & Schmidt, 1999), quasi-maximin (Charness & Rabin, 2002), and constant elasticity of substitution (Andreoni & Miller, 2002). I then compare the models’ ability to predict behavior out-ofsample. I find that the Bliss Point model generally predicts about as well as the other models. Further, it is a better predictor than the other models for up to 7% of subjects, depending on the task. This suggests that the Bliss Point model is a useful alternative to existing models.