| 简介： | There are many contexts in economics where productivity and welfare performances of institutions and policies depend on who matches with whom. Examples include matching of caseworkers and job seekers in job search assistance programs, medical doctors and patients, teachers and students, attorneys and defendants, tax auditors and taxpayers, among others. Although reallocation of individuals through a change in matching policy can be less costly than directly training personnel or offering a new program, methods for learning optimal matching policies and their statistical performances are less studied. This paper develops a method to learn welfare optimal matching policies for two-sided matching problems in which a planner centrally prescribes who should match with whom based on individual’s observable characteristics of the two sides. We formulate the learning problem as an empirical optimal transport with the match cost function estimated from training data, and propose to estimate an optimal matching policy by optimizing the entropy regularized empirical welfare criterion. We derive a welfare regret bound of the estimated policy and characterize its convergence. We apply our proposal to the assignment problem of caseworkers to job seekers for a job search assistance program, and assess its welfare performance in a simulation study calibrated with French administrative data. |