The cost of fair divisions: An experimental investigation of Ultimatum Games with groups

I investigated the effect of the presence of a group of non-active subjects upon the behavior of active players in a Ultimatum bargaining game. In the experiment a subject with the role of P has to offer a share r of a sum S to a subject with the role of AR who belongs to a group and decides on behalf of his group’s members (players R). If AR rejects the P’s offer, both active and non-active players get zero, if AR accepts the offer then P gets S − r while r is equally divided between AR and the members of his group. Every subject assumes all the three roles (P, AR and R) and the group size is manipulated keeping constant the share S/N (with N=number of subjects, either active or non active, involved in the game) Data suggest that active players tend to behave as they were playing a standard two-person Ultima-tum game. A clear insensitivity to changes in group size by subjects playing as P, emerging in the main experiment, is compatible with the hyphotesis that at the basis of their behavior there is a willingness to gain a payoff which satisfies an ex-ante fixed aspiration level, that for most of them corresponds to about half of S. The interpretation of the decisions taken by subjects under the AR role is more complicated as, although most of them show a behavior which is compatible with the one observed in the standard Ultimatum Game, a non-negligible share of players fix very low acceptance thresholds that could be explainained in terms of a shift from the willingness to punish unfair behaviors to the responsibility for others ’ wellbeing. The experimental analysis of norms of fairness and reciprocity has become one of most promising strand of economic and decision theory research. The results gathered in twenty years of experiments have contributed to uncover the limits of models of human agents as motivated only by the maximization of their own material payoff. At the same time a number of theories have been introduced aiming at explaining the experimental evidence by means of models based on the hypothesis that agents have a multiplicity of motivations, maximization of their own material payoff being only on