🤖 AI Summary
This paper studies institutional admission decisions in decentralized many-to-one matching markets under persistent preference uncertainty: applicants gradually reveal their preferences only upon receiving offers, rendering classical stable matching mechanisms inapplicable. To address this, the paper proposes a probabilistic offer mechanism—where institutions send stochastic admission offers based on prior beliefs, and applicants subsequently update their preferences and respond. Combining randomized assignment design, game-theoretic analysis, and expected utility modeling, the paper establishes that this mechanism achieves ex ante market clearing and stability. It further identifies, for the first time, the counterintuitive phenomenon that increased information disclosure may reduce aggregate applicant welfare. The core contribution is the first decentralized matching framework for dynamically revealed preferences that simultaneously guarantees stability, implementability, and analyzable welfare properties.
📝 Abstract
This paper studies a decentralized many-to-one matching market where preferences remain uncertain during the matching process. Institutions initiate matching by sending offers, and applicants decide whether to accept upon receiving them. Since applicants learn their preferences only after receiving offers, institutions face a challenge in deciding how many offers to issue. I address this challenge by introducing probabilistic offers (admitting applicants with a probability less than one), which ensure that ex-ante market clearing and stability are achievable. However, the welfare effect of information is subtle: applicants may become worse off as they acquire more information.