Qiyuan Zheng
Ph.D. Candidate in Economics, Princeton University
Welcome! I am a microeconomic theorist with interests in information and mechanism design. My research examines how information structures shape incentives and welfare. I am on the 2026–2027 job market.
Job market paper
-
Regulating Discretion: Information Design with Privately Known Common InterestsPDF
Abstract: A principal regulates the information of an agent who privately knows which statistic of the state determines both parties’ payoffs from his action. The agent’s payoff is this shared component plus a bias, a second statistic of the state. When the statistics are correlated, information that helps the agent learn about the shared statistic may also serve his bias. Our main result characterizes the optimal mechanism, which exhibits two distortions that arise from screening. First, the agent evaluates the state by a virtual statistic rather than his own. Second, the agent conditions on the bias statistic when making his decision. We benchmark the optimal mechanism against simpler alternatives, such as full information and mechanisms that do not condition on the bias statistic. In particular, when the agent can exploit the correlation, forbidding the use of a protected characteristic distorts his evaluation of the state more than the optimal mechanism does.
Working papers
-
Strategically Ambiguous Deterrence
Draft coming soon
We study deterrence when agents are ambiguity averse. A principal chooses a schedule of penalties across levels of compliance and how much of that schedule to disclose. When only aggregates are disclosed — the marginal distributions of penalties and of compliance — agents cannot infer the penalty attached to each level of compliance, and evaluate each level at the worst case consistent with what they observe. Obfuscation is therefore efficient: it sustains any given profile of perceived penalties at a strictly lower expected penalty cost. We show that this does not imply obfuscation is optimal: since the distribution of compliance is itself determined by agents’ behavior, the profiles that obfuscation can sustain are restricted by an equilibrium requirement. We characterize the implementable set: obfuscation can sustain a perceived penalty profile if and only if the perceived penalty for partial compliance, weighted by the mass of partial compliers, covers the perceived gap between partial and zero compliance, weighted by the mass of zero compliers. Obfuscation therefore compresses perceived penalties across compliance levels, and a principal who needs a large marginal increase in penalties (to prevent escalation) may strictly prefer transparency. We give sufficient conditions for each.
-
Can Fact-Checking Backfire? Mixing Information When the Source Is Hidden
Draft coming soon
This paper studies the problem of a designer that tries to improve the decision making of an agent by injecting verified information into their news source. When the agent cannot observe the underlying signal structure, they cannot distinguish whether the signal they observe is drawn from the status-quo or the verified information source. I first show that mixing in Blackwell-dominant information sources, even perfect information, may reduce the agent’s payoff in simple decision problems. Therefore, naive fact-checking may be harmful to the agent. I then study a problem in which the designer may garble the verified information before injecting it into the agent’s news feed. I show that this problem reduces to choosing how to pool states into existing signal labels. When the designer’s influence over the news feed is small, I characterize exactly when unaltered fact-checking is optimal, and show how pooling states prevents the misreadings that make fact-checking costly.
Work in progress
-
Multi-Stage Application Processes for Distributional Concerns
Teaching
-
ECO 501: Graduate Microeconomics I
-
ECO 300: Undergraduate Microeconomics