We study the effect of cash transfers on health disparities using data from the Department of Veteran Affairs’ Disability Compensation program, one of the nation’s largest cash transfer programs. Our instrumental variables strategy analyzes disability claims and medical records for over 800,000 veterans, leveraging the quasi-random assignment of these claims to two agents: caseworkers and physicians. Cash transfers reduce mortality on average, but there is considerable heterogeneity. Using only the caseworker instrument, we find that receiving $300 per month reduces 5-year mortality by 1.0 pp. Conversely, the physician instrument recovers a significantly smaller effect of 0.4 pp. We reconcile this discrepancy with a reverse selection-on-gains model, in which caseworkers and physicians differentially target cash transfers, but sicker veterans gain the least from transfers. Our findings help explain disparate findings in the literature and provide policy implications for targeting transfers to narrow disability-related health disparities.

Many medical innovations succeed in clinical trials but fail to make an impact outside controlled settings. This paper investigates why the introduction of pre-exposure prophylaxis (PrEP), whose trials demonstrate near-perfect efficacy in preventing HIV, coincided with a plateau in infections that halted decades of progress. Using epidemiological, pharmaceutical, and Census data, we estimate PrEP’s real-world effectiveness and investigate why the innovation failed to reduce population-level transmission. To address selection bias in uptake, we exploit two facts: PrEP is used almost exclusively by men who have sex with men, and regional concentrations of male same-sex partnerships vary widely. Counties with higher male same-sex partnership rates experienced greater PrEP uptake, which in turn lowered transmission rates. Our results imply that 76 additional users avert one new HIV diagnosis. However, while PrEP uptake is similar between Black and White men, eightfold higher incidence among Black men indicates substantial underutilization relative to need. Furthermore, PrEP’s introduction was associated with increased risky sexual behavior among both users and non-users.

Tracking allows instruction to be tailored to students’ abilities but may divert resources from lower-performing students, widening achievement gaps. We test these hypotheses by examining the most widespread tracking program in the U.S.: Advanced Placement. Leveraging its expansion in Texas following a statewide incentive policy and linking administrative microdata to long-run outcomes, we find that AP adoption increased college completion by 1.2 pp and adult earnings by 3.3%. Gains were largest for high-performing students, but lower-performing students also saw meaningful relative improvements. Our findings reject that tracking harms secondary, postsecondary, or labor-market outcomes for any academic, racial, or socioeconomic group.