Research
Work in progress
Draft available upon request
The Structure of Bayesian Stable Matchings with Transfers
Classical matching theory gives stable outcomes a distinct structure: an identical set of unmatched agents across stable outcomes (the Lone Wolf theorem) and a lattice of stable payoffs. Hu (2026) shows that under incomplete information matching without transfers, the Lone Wolf property survives whenever stable outcomes share a common information structure. I show that this is not generally true under incomplete information matching with transfers, and give a necessary and sufficient condition under which the Lone Wolf theorem holds generically.
In progress
Deliberation by Ballot: Communication and Consensus in Elections by Repeated Voting
Some of the oldest consensus institutions — conclaves, juries, conventions — do not separate deliberation from voting: members communicate by voting, casting repeated ballots until a supermajority forms. I model committees in which each ballot is semi-binding communication about private information and private biases.
In progress
Implementing Stable Matching under Private Information
This paper implements Bayesian stable matching (Liu, 2020) as a Markov perfect Bayesian equilibrium in a decentralized bargaining process following Elliott and Nava (2019).