Jori Barash

This paper estimates the effect of a physician’s number of registered patients ("enrollment”) on short-run treatment intensity in the context of Norwegian primary care. I instrument for enrollment with quasi-random administrative patient assignments. The estimated effect of enrollment is negative but small for several measures of treatment intensity. For example, with one new patient registration, the average physician spends 3 fewer minutes per month across incumbent patients. Descriptive evidence suggests that crowd-out exacerbates under-treatment. Crowd-out is larger among physicians who reach their stated capacity or initially work part-time. Drawing on a model of physician decision-making, this heterogeneity implies that capacity constraints dominate income effects in explaining crowd-out. With capacity constraints, increasing the number of physicians may more effectively reduce crowd-out than incentives for greater treatment per physician. Fixing physician supply, an alternative patient assignment rule could reduce crowd-out from administrative assignment by 86 percent.

Some universities are better than others at improving long-run outcomes, but recent policy changes limit the scope for admissions to equitably ration spots. I investigate whether selective universities can use graduation-contingent loan forgiveness to improve match quality. While raw means suggest a short-lived loan forgiveness program doubled graduation rates, I show that this difference was entirely driven by selection rather than effects on effort. Exploiting a discontinuity from Pell Grant eligibility, I find no detectable effect of participation on major choice, course load, part-time work, persistence, on-time graduation, or earnings. Moreover, consistent with variation in match quality, I find heterogeneity across students in the effect of selective university enrollment on graduation. I incorporate these findings into a structural model of college choice, loan forgiveness take-up, and graduation among admitted students. Given model estimates, I simulate welfare under counterfactual menus of financial aid. Improved revenue-neutral design of loan forgiveness would lead to large welfare gains including greater statewide graduation rates.

In times of crisis, means-tested government programs sometimes relax eligibility standards to deliver aid faster. With adverse selection and less time to screen beneficiaries, relaxed eligibility may increase expenditure on non-targeted populations and decrease pass-through to households from private intermediaries, with both mechanisms lowering efficacy. The tradeoff between speed and eligibility standards is motivated by the relatively untested premise that faster delivery meaningfully improves outcomes. This paper shows that timely subsidized modification of distressed mortgages could have further reduced U.S. foreclosures in the aftermath of the 2008 financial crisis. We exploit a simulated instrument based on the spatial distribution of financial shocks. Using a dynamic structural model of servicers' modification and foreclosure choices, we characterize how the optimal modification subsidy varies with time from delinquency, servicer volume, and market conditions.

Environmental policy frequently subsidizes low-pollution products, but this approach can be relatively expensive given information frictions and ineffective given consumer heterogeneity. For example, plant-based meat has a lower carbon footprint than most animal products, but it is unclear ex-ante whether economies of scale are sufficient to spur widespread adoption. Motivated by reduced-form evidence of large changes in behavior after first purchase, I estimate demand for these products with a 10-year nationwide household panel, allowing for imperfect information about quality and rich consumer heterogeneity. I evaluate the effect of counterfactual vouchers and marginal costs on consumer surplus and CO2 emissions. Contrary to conventional wisdom, lower prices are unlikely to induce large-scale substitution, but offering vouchers to inexperienced households accelerates the pace of adoption. The averted social cost of carbon is twice as large as direct voucher costs.

Selective universities may soon be limited to race-neutral approaches for promoting diversity. We provide evidence on the complementarity between guarantees and targeted outreach. The Texas Advance Commitment guarantees free tuition to admitted students with family adjusted gross incomes below $65,000. Exploiting this discontinuity among admitted students, we observe a large change in total aid, but no evidence of a change in enrollment. We investigate whether information frictions explain the null effect with a randomized control trial. The trial targets high school seniors eligible for free or reduced-price lunch and in the top 10% of their class. The first wave of the study included 32 high-poverty schools, with 16 randomized to receive outreach that included proactive guarantees of free tuition, housing placement, and a housing scholarship. Results are preliminary and subject to change: relative to outreach alone, outreach with guarantees increased applications by 18 percent and admissions by 10 percent. Point estimates are larger for students with guaranteed admission.