Research

Working Papers

Early Mentors for Exceptional Students

Is the supply of extraordinary talent constrained by the scarcity of exceptional mentors, or can proficient but ordinary teachers suffice? I test these competing hypotheses in mathematics using newly assembled data spanning four decades of the American Mathematics Competitions (AMC) linked to long-run education and career outcomes. Exploiting variation in the timing and geographic availability of school-based mentors, I show that teachers who organize math competitions play a central role in revealing latent talent: mentor entry more than doubles the number of top-performing students identified at a school by the AMC, with comparable declines upon exit. Access to a high school mentor also shifts the trajectories of already-identified students, increasing the likelihood that they attend highly selective universities, hold patents, and work at elite tech firms. These mentors are overwhelmingly ordinary teachers, suggesting the binding constraint on extraordinary talent is not the scarcity of exceptional mentors but the underutilization of a broad pool of capable ones. The findings imply that expanding mentor access is a scalable, low-cost lever for increasing the long-run supply of talent in science and innovation.

Invited Presentations: NBER Economics of Talent; Midwest Economics of Education; Technology, Innovation, and Entrepreneurship Seminar (UC Berkeley, Haas)

Media Coverage: The Economist; Marginal Revolution; The Report Card with Nat Malkus

Selected Research in Progress

The Persistence of Early Scientific Interests: Evidence from the Science Talent Search

This paper examines whether exceptionally talented high school scientists persist in science as adults and whether their specific scientific interests carry forward. Using 80 years of data from the Science Talent Search (STS), a prestigious national science competition, I document substantial heterogeneity in long-term persistence across scientific fields: students whose high school projects focus on mathematics, physics, or computer science are significantly more likely to receive an NSF Graduate Research Fellowship than those in the life sciences or social sciences, and are also more likely to hold patents. A large gender gap in persistence exists across both measures, though it largely closed by the mid-1970s for fellowship receipt. Despite growing representation, Asian semifinalists are surprisingly less likely to persist on both margins. Conditional on persisting, students tend to remain in the same domain they chose in high school, a pattern visible across fellowship awards, patents, and academic publications. These findings suggest that the scientific interests of exceptionally talented students are quite ``sticky," with implications for the long-run distribution of talent across scientific fields.

Teacher Shortages and Labor Supply: Evidence from Application Data (with Michael Bates and Andrew C. Johnston)

Rigid salary schedules in public schools may generate a shortage of quality candidates for some openings and a surplus in others. We use novel applicant data from a hiring platform to investigate the supply of teachers to openings in different subjects and school types. Positions in science, mathematics, and low-income schools receive few quality applicants. Applicants are plentiful in early childhood education and suburban schools. We leverage changes in salary schedules to estimate how labor supply responds to salary for each type of opening. We use the estimates to calculate salaries that equalize quality applicants across schools and fields.

The Mathematical Olympiad Summer Program: Impacts on Later-in-Life Math Productivity

Mathematics Research

Approval Ballot Triangles and Strict-Sence Ballots (with Andrew Beveridge), under review, 2026.

De Finetti Lattices and Magog Triangles (with Andrew Beveridge and Kristin Heysse), The Electronic Journal of Combinatorics, 2021.

The Voter Basis and the Admissibility of Tree Characters (with Andrew Beveridge), Order: A Journal on the Theory of Ordered Sets and its Applications, 2021.