Welcome!
I'm Siqi Li (李思麒), an assistant professor in Economics at the School of Management and Economics, Chinese University of Hong Kong, Shenzhen.
My research interests lie in how information frictions and incentive misalignment shape behavior and welfare, from both a theoretical and empirical perspective.
I received my Ph.D. in Economics at the University of Pennsylvania in 2026.
Email: lisiqi@cuhk.edu.cn
Rising competition in imperfect markets pushes agents to invest in costly signals that differentiate themselves. Such investments can mitigate unraveling and improve matching efficiency, but also generate rat races that reallocate resources towards relative standing. I develop an empirical framework to quantify how competition affects signal adoption in matching markets and its welfare consequences, applying it to the role of pre-Ph.D. experiences—master’s and predoctoral programs—in Ph.D. admissions. These experiences help programs screen applicants and provide research training. Yet when capacity is limited and grade inflation reduces informativeness, students pursue additional research experience to stand out. Using LinkedIn data on Economics and Business Ph.D.s, I find that pre-Ph.D. experience improves admission outcomes, with 54% of the gain attributable to signaling and 46% to training. While signaling restores about half of the matching efficiency lost under pooling, its opportunity costs exceed benefits, yielding a 15% net welfare loss. Benefits are concentrated among economics majors from top colleges; other groups are worse off. Grade inflation explains roughly one-quarter of the rise in pre-Ph.D. experience.
2024 International Conference on Game Theory, Stony Brook
We study a durable-goods seller who offers buyers a price plan specifying current and future posted prices. Each plan binds until random expiration allows her to announce a new binding plan. Under public expiration, all perfect Bayesian equilibria yield the same seller revenue, attainable with non-decreasing price plans. Longer commitment can reduce profit below the no-commitment benchmark: stronger commitments after expiration can make future concessions more profitable, weakening buyers' initial willingness to pay. When only the seller observes expiration, scheduled sales can instead reward her for concealing expiration and continuing the announced plan. Reserving enough future revenue can sustain higher initial payments and raise profit above public expiration. Both regimes retain a Coasian conjecture.
[ Draft ]
2023 International Conference on Game Theory, Stony Brook
I study a disclosure game in which an investigator chooses state-contingent probability of finding evidence to reveal state, a gatekeeper may suppress any evidence found. The investigator values the publication, whereas the gatekeeper benefits from receiver's belief of high state. Covert acquisition uses a constant discovery probability wherever evidence would be disclosed, while overt acquisition searches more intensively at higher states. While high states are more likely be reported, silence induces a lower posterior, reducing the return to suppression and expanding disclosure. Investigator and receiver strictly prefer overt acquisition. The gatekeeper prefers overt when its payoff is convex, and covert when concave.
[ Draft ]
2025 ASSA Annual Meeting
I build and test a dynamic model of learning and moral hazard in drug innovation. Firms receive noisy trial outcomes, update beliefs about quality, and decide whether to continue or terminate development. Financing structure shapes continuation incentives: large pharmaceutical firms self-finance and internalize costs and rewards, while venture-backed biotechs face milestone-based contracts that attenuate downside risk and distort attrition. The model predicts biotechs are more likely to continue after unfavorable signals because continuation signals competence to investors. Using a panel of 12,000+ clinical projects from~3,000 firms (1990 — 2020), I find that, conditional on negative signals, biotech projects are 12% less likely to be discontinued than those of large pharmaceutical companies. Contractual agency problems thus distort dynamic learning and selection, with implications for innovation efficiency and policy design.
[ draft ]