Strategy-proof, efficient, and nonbossy quota allocations

We consider the problem of designing a mechanism to allocate objects to agents when each agent has a quota that must be filled exactly. Agents are assumed to have responsive preferences over items. We show that the only strategy-proof, Pareto optimal, and nonbossy mechanisms are sequential dictatorships. We also show that the only strategy-proof, Pareto optimal, nonbossy, and neutral mechanisms are serial dictatorships. Since these negative results hold for responsive preferences, they hold for more general preferences as well.