Résumé en anglais

A stylized fact of most centralized matching markets (for example school choiceproblems is that participants tend to submit preference lists over a small number ofpotential matches. This paper is based on the premise that this fact provides additionalinformation that can be used to match agents. In a school choice context we showthat considering only schools’ priorities and the set of acceptable schools for each student (but not their submitted preferences), we can restrict the set of possible stable matchings that can arise for any preference profile of the students that leaves the setof acceptable schools unchanged. Our approach consists of linking Hall’s marriage condition to stable matchings and offers a methodology that can be used for data analysis when one has access to the preferences of only one side of the market. We use our technique to propose a new mechanism for school choice problems that Pareto dominates the Student Optimal Stable Mechanism, and show that in this mechanism it is a dominant strategy to always rank the schools that have been revealed as acceptablein the same order as in the true preferences.