Push, Don't Nudge : Behavioral Spillovers and Policy Instruments

Policy interventions are generally evaluated for their direct effectiveness. Little is known about their ability to persist over time and spill across contexts. These latter aspects can reinforce or offset the direct impacts depending on the policy instrument choice. Through an online experiment with 1,486 subjects, we compare four widely used policy instruments in terms of their ability to enforce a norm of fairness in the Dictator Game, and to persist over time (i.e., to a subsequent untreated Dictator Game) or spill over to a norm of cooperation (i.e., to a subsequent Prisoner's Dilemma): defaults, implemented by pre-setting dictators' donations at 50% of the endowment; social information, where dictators were informed that roughly half of the donations made by subjects in a pilot study were equal or above 50%; rebates, which returned half of the donation to dictators who gave at least 50% of their endowment; and a minimum donation rule, forcing dictators to donate at least 50% of the endowment. Our results show that (i) rebates, the minimum donation rule and social information have a significant positive direct effect on fairness, and that (ii) the effect of rebates and the minimum donation rule persists in the second game, but only within the same game type. Moreover, we measure observers' beliefs about dictators' donations and find that (iii) subjects overestimate the direct and the spillover effects of defaults and the spillover effect of social information