OTHER-REGARDING PREFERENCES: OUTCOMES, INTENTIONS OR INTERDEPENDENCE?

Abstract. If preferences and beliefs are appropriately parametrized, different theories of “other-regarding ” preferences possess equilibria that are consistent with experimental results in a variety of settings. Our goal is to experimentally separate be-tween those theories, by studying their comparative-statics per-formance in the neighborhood of the classic Ultimatum Game, whose results are extremely robust. In order to perform this ex-ercise, we first characterize monotone Perfect Bayesian Equilibia in the Ultimatum Game if preferences are interdependent. We then show that in this model, setting a lower bound to the of-fer a proposer can make, may decrease the proposer’s offer and increase the responder’s acceptance probability. Outcome-based theories and intentions-based models have (weakly) opposite pre-dictions. We then design and execute an experiment that facilitates almost instantaneous learning and convergence by both proposers and responders. The experimental results are consistent with the predictions of the interdependent-preferences model. 1