Research

Job Market Paper

Dynamic Treatment Effects in Education

Examines how unobserved education paths affect lifetime welfare and MVPF comparisons across 38 policy calculations.

Abstract for Dynamic Treatment Effects in Education

Lifetime welfare calculations often rely on policy effects on educational attainment measured in a study’s final observation window—for example, effects on college completion by age X. These estimates leave unresolved whether subsequent attainment widens or narrows the observed gap between the policy and comparison groups, and when that attainment occurs. This paper develops a framework to assess the welfare implications of this uncertainty. I document the timing information in existing studies and model possible continuations of observed education effects. Two deterministic simulations show how different education paths can match the same observed effect yet imply different lifetime earnings values, and how additional observations can narrow this range. I apply the framework to 38 policy calculations to reconstruct ranges for the marginal value of public funds (MVPF) and propose an ordering rule based on these ranges. Under the specified path and valuation assumptions, the same education evidence can support conflicting welfare judgments even when the estimated effects are treated as known. My contribution is a framework for assessing the robustness of welfare comparisons to unobserved education paths and quantifying how estimates from each additional observation window can narrow uncertainty.

Working Papers

Higher Education or Just Higher? The Unintended Effects of Recreational Marijuana Legalization

Uses staggered legalization across U.S. states to study degree completion, academic performance, and potential behavioral mechanisms.

Abstract for Higher Education or Just Higher? The Unintended Effects of Recreational Marijuana Legalization

This paper examines the effects of recreational marijuana legalization (RML) on educational attainment in the United States. Exploiting variation from the staggered adoption of RML across states, I find that RML reduces the probability of earning a bachelor’s degree within four, five, and six years by 3.8%, 2.0%, and 3.1%, respectively, relative to pre-treatment means. In addition, RML lowers the three-year graduation rate at two-year colleges by 6.0% relative to the pre-treatment mean. These effects suggest that RML not only delays degree completion but may also lower educational attainment in the longer run. I further find that RML reduces grade point average (GPA) among currently enrolled college students ages 18–24 and study time among individuals in the same age range. Increases in marijuana-use frequency and other substance use among college-age individuals are plausible mechanisms, although the data do not establish their contribution to the educational effects. These findings highlight the potential educational costs of marijuana legalization alongside the documented benefits identified in prior research.

New Sites, New Services: Evidence from a Randomized Dialysis Payment Reform

Uses a randomized Medicare payment reform to study the home dialysis services supplied by successive cohorts of new facilities after entry.

Abstract for New Sites, New Services: Evidence from a Randomized Dialysis Payment Reform

I use the geographic randomization of Medicare’s End-Stage Renal Disease Treatment Choices (ETC) Model to study how payment incentives shape the services that new facilities supply after entry. Linking facility entry histories to annual cost reports, I follow fixed entry cohorts and incumbents through 2024 and compare their contributions with covered-market supply. Statistically significant increases in home dialysis–providing facilities and Medicare home treatments are concentrated among early entrants. The first payment-period cohort contributes an additional 0.125 providing facilities and 183.01 Medicare home-treatment equivalents per hospital referral region on average, with positive treatment responses continuing through follow-up. A similar pattern is not evident among incumbents or across the covered market. For the first payment-period cohort, an accounting decomposition assigns most of the supply increase to entrant numbers, with a smaller contribution from service per entrant. Higher reported training and administrative staffing accompany this expansion. Supplementary market estimates show lower long-term catheter use and higher fistula use; effects on mortality, hospitalization, and geographic access remain imprecisely estimated. These findings suggest that new facilities can be a margin of recorded supply adjustment to payment reform, while net market expansion and broader welfare gains remain uncertain.