Working Papers
-
"The Human Capital Effects of Housing Assistance" (Job Market Paper) [draft]
This paper estimates the impact of means-tested housing subsidies on children’s human capital accumulation and early adult outcomes. I link applications to public housing and housing voucher wait lists in New Haven, Connecticut to statewide education, labor market, and criminal justice records, and exploit variation in when households reach the top of randomly-ordered wait lists as an instrument for childhood exposure to assistance. Receiving housing assistance significantly reduces school-reported homelessness and housing instability at younger ages and raises measures of early childhood development among boys. Housing assistance also improves longer-run outcomes. Each additional year of assistance raises high school graduation and college enrollment by 2 to 4 percentage points, and children that receive housing assistance are less likely to be convicted of a felony in early adulthood. To distinguish among mechanisms driving these results, I estimate short-run effects on school continuity, school quality, and adult labor supply. Finally, translating the attainment effects into predicted impacts on lifetime earnings and accounting for reductions in incarceration and sheltered homelessness, I calculate that providing housing assistance to a household with children produces roughly $2 in benefits per government dollar spent. Because of larger impacts on educational attainment, housing vouchers are more cost-effective than public housing units.
-
"A Sin Ban, Not a Sin Tax? Consumption and Incidence of SNAP Soda Restrictions" [NBER] [SSRN]
We study the first statewide restrictions on purchasing sweetened beverages and candy with Supplemental Nutrition Assistance Program (SNAP) benefits. Using a multi-retailer receipt panel and a stacked difference-in-differences design based on staggered implementation in 2026, we find that SNAP households' spending on restricted products falls by 11 percent. Substitution is limited and asymmetric: in states where sweetened fruit drinks remain eligible, substitution toward them partly offsets a soda decline, while candy restrictions reduce rather than increase purchases of still-eligible snacks. Calories and sugar from beverages and packaged snacks fall by 5 to 8 percent. The response is not driven by changes in households' budget sets: shelf prices barely change, and the loss of SNAP's sales tax exemption raises tax-inclusive prices by 1.9 percent. Additional evidence does not support nutrition signaling or checkout stigmas as primary drivers of the response. Instead, spending declines across store types increase monotonically in households' pre-policy SNAP spending, consistent with households treating SNAP benefits and cash as non-fungible. Under nationwide adoption, the restrictions impose an estimated $176 million in additional sales taxes on SNAP participants - only 13 percent of payments under excise taxes calibrated to achieve the same purchase reductions - and a back-of-the-envelope behavioral welfare calculation incorporating nutrition-related corrective benefits implies annual welfare gains of approximately $1.1 billion.
-
"Where Does the Tide Roll? The Causes and Consequences of Out-of-State Enrollment at Public Universities" [SSRN]
Public flagship universities increasingly use nonresident tuition revenues to offset declining state funding. Nationally, we find that public flagships facing lower state support enroll relatively more high-income nonresidents. We then link two decades of University of Alabama (UA) commencement records to admissions office recruiting visits and individual employment histories. These data allow us to document how universities pursue this revenue margin and to measure downstream effects on graduates and state budgets. We show UA boosts high-income out-of-state demand through targeted recruitment visits: in a differences-in-differences design, new visits generate 8–10 percent more future UA graduates. We then trace how nonresident enrollment redirects graduates across state labor markets. Using variation in the state-of-origin composition of UA cohorts and a shift-share instrument constructed from origin states’ non-UA out-of-state enrollment rates, we estimate that every 100 out-of-state students cause 9–15 additional in-state students to leave the state after college. These patterns are consistent with peer effects operating through expanded consideration sets and referral networks. The resulting fiscal losses to Alabama exceed the university's nonresident tuition margins. Finally, the out-of-state expansion appears to support in-state degree production. A budget accounting exercise shows that UA’s out-of-state enrollment margins are sufficient to cross-subsidize more than half of the in-state cohort, who receive an average annual subsidy of $13,000 per student.
Work in Progress
-
"The Demand for Income Share Agreements"
We study student demand for Income Share Agreements (ISAs), a private alternative to federal income-based student loan repayment plans, using proprietary application-level data covering more than 29,000 applications from prospective student borrowers. To identify how the offered repayment rate affects demand, we exploit a sharp change in the provider’s pricing algorithm among a subset of applicants receiving an initial quote, and we construct a machine learning model to measure applicants' exposure to the change in the pricing algorithm. Using this exposure measure as an instrument, we estimate that a 0.1-percentage-point increase in the share of income owed to the provider - equivalent to around $230 in additional payments in expectation - reduces take-up by roughly 10%.