Rawlsian Matching

Abstract

We apply the Rawlsian principle to a canonical discrete object allocation problem. A matching is Rawlsian if it is impossible to improve the ranking of assignment for the worst-off agent or reduce the cardinality of the set of the worst-off agent-body. None of the well-known mechanisms are Rawlsian. We introduce an efficient and Rawlsian class of mechanisms. Strategy-proofness is incompatible with Rawlsianism; therefore, no Rawlsian mechanism is strategy-proof.