Job Market Paper

Robustness and Simple Contracts in Sequential Screening (draft coming soon!)

How should a seller screen buyers when she observes the distribution of their valuations but not how each buyer learns them? Does the rich dynamic menu of contracts survive?

Abstract: I study optimal sequential screening when buyers gradually learn their valuations over time, as in Courty and Li (2000), but the seller observes only the distribution of valuations and is uncertain about how individual buyers learn. When this distribution is symmetric or right-skewed, a single option with full participation, a static contract that every buyer accepts but only some exercise, guarantees the seller the highest expected revenue against an adversarially chosen learning process. It is essentially the unique mechanism that does so. Whereas the Bayesian benchmark calls for a menu of infinitely many options, ambiguity about learning leaves no value in early-stage screening.

• Short note on Courty and Li (2000) (with Eric Gao) (coming soon!)

• First-order approach to robustness problem (coming soon!)

Working Papers

• Technology Speed Limits (with Andrew Koh), 2026 (New!)

Follow-up to "Robust Technology Regulation" (with Andrew Koh) in which the agent endogenously chooses how fast to learn.

• Memory Correlated Equilibrium (with Andrew Koh), 2025

How does memory affect equilibria in games? Memory structures can alleviate commitment issues.

• Robust Technology Regulation (with Andrew Koh), 2025

How should uncertain technologies be robustly regulated when a learning process is unknown?

• Attention Capture (with Andrew Koh), 2024

Revise and Resubmit, Journal of Political Economy

How to optimally extract attention from platform users through dynamic information design?

• Persuasion and Optimal Stopping (with Andrew Koh and Weijie Zhong), 2024

Submitted

A unified method for solving dynamic information design with optimal stopping

• Informational Puts (with Andrew Koh and Kei Uzui), 2024

Submitted

A dynamic information design using only public signals can guarantee the best equilibrium in a dynamic coordination game with switching frictions.

• Multiple Admission Tracks in the College Admissions Problem, 2022

When does a college prefer a multi-channel admissions policy to a test-only policy?

Work in Progress

• Implementable Dynamic Information Structures under Outside Learning

To what extent can the information designer persuade a receiver who learns additional information from an outside source?