Fair procedures with naive agents: Who wants the Boston mechanism?

We study preferences over procedures in the presence of naive agents. We employ a school choice setting following Pathak and Sönmez (2008) who show that sophisticated agents are better off under the Boston mechanism than under a strategy-proof mechanism if some agents are sincere. We use lab experiments to study the preferences of subjects for the Boston mechanism or the assortative matching. We compare the preferences of stakeholders who know their own role with agents behind the veil of ignorance and spectators. As predicted, stakeholders vote for the Boston mechanism if it maximizes their payoffs and vote for the assortative matching otherwise. This is in line with the model of Pathak and Sönmez (2008). Subjects behind the veil of ignorance mainly choose the Boston mechanism when the priority at schools is determined randomly. In a second experiment with priorities based on performance in a real-effort task, spectators whose payoff does not depend on the choice of the mechanism are split in their vote for the Boston mechanism and the assortative matching. According to the spectators' statements in the post-experimental questionnaire, the main reason for preferring the Boston mechanism is that playing the game well deserves a higher payoff. These findings provide a novel explanation for the widespread use of the Boston mechanism.