Using an experiment with material incentives, this paper investigates the violation of composite dominance relationships, viz. absolute Pareto dominance, Pareto rank dominance, transfer dominance, Lorenz dominance, and generalized Lorenz dominance. Moreover, we test tail independence. The experiment consists of two treatments, a self-concern mode (in which each subject expects payoffs according to her own choices), and a social-planner mode (in which subjects form their preferences without any chance of receiving payoffs when they became effective). The main focus of this paper centers on the behavioral shifts between the self-concern and the social-planner modes. We show, first, that subjects' behavior is different under the two treatments. Second, we show that there are less violations of the two Pareto dominance relations and of generalized Lorenz dominance and more violations of Lorenz dominance and of transfer dominance under the self-concern mode than under the social-planner mode. Within these groups, behavior is more similar under the self-concern mode than under the social-planner mode. Tail independence is widely rejected.
Income distributions dominance relations tail independence