Research

Working Papers

Grade Inflation, Signaling And Status Externality in the Ph.D. Admission Market (Job Market Paper)

Increasing competition in the labor market induces agents to take costly credentials that may enhance productivity, send signals to committees, yet fuel a rat race. I propose a model in which grade inflation reduces the informativeness of undergraduate records, raising the signaling value of predoctoral experience under fixed program capacity and generating status externalities. Applying the framework to new data on U.S. Economics and Business Ph.D. cohorts, I quantify both matching gains and competitive costs. I find that grade inflation explains about 30% of the rise in predoc take-up; signaling restores 48.5% of the efficiency gap relative to full information, but its private costs are 63.4% of that gap, implying a net welfare loss of 14.9% as credentials scale.

Agency Problem in Learning: Theory and Evidence from the Drug Innovation Market

2025 ASSA Annual Meeting

This paper examines how the agency problem in venture-backed drug development influences attrition decisions in the pharmaceutical industry. I develop a dynamic learning model of clinical trials in which large pharmaceutical firms self-finance their research, while biotech companies rely on milestone-based contracts with venture capital. The model predicts that, after receiving relatively poor trial signals, biotech firms are more likely than large pharmaceutical companies to continue development, reflecting distorted incentives to pursue lower-quality drugs. Using a dataset covering over 12,000 drug projects from 3,000 firms over the past three decades, I find that, conditional on negative signals, drugs held by biotech firms are 12.3% less likely to be discontinued than those held by large pharmaceutical companies. These results highlight how venture financing can exacerbate inefficient attrition decisions, with implications for welfare, contract design, and regulatory policy.

Evidence Acquisition and Positive-Biased Mediator

2023 International Conference on Game Theory, Stony Brook

This paper analyzes a sender–receiver game in which a third‑party mediator can incur effort to obtain verifiable information about an unknown state. The sender’s payoff increases in the receiver’s action, the receiver seeks to match the state, and the mediator always prefers disclosure. When the receiver observes neither the mediator’s effort nor whether evidence exists (covert acquisition), equilibrium replicates a standard voluntary disclosure outcome: the sender discloses only above a cutoff, and the mediator exerts no effort below that cutoff and constant effort above it. When the acquisition process is observable (overt acquisition), the mediator can reweight her effort toward high states. By making silence a stronger bad‑news signal, this positive bias relaxes the sender’s incentive to censor, leading to strictly more disclosure. I characterize the optimal overt effort function and show that overt acquisition increases ex ante information disclosure and welfare for the receiver while leaving the sender indifferent. The results suggest that transparency about information acquisition can mitigate censorship even when the mediator’s search is skewed toward favorable evidence.

Working in Progress

Adverse Selection with Partial Commitment (with Hongting Leng)

2024 International Conference on Game Theory, Stony Brook

Page updated

Google Sites

Report abuse