Research

This paper examines how cyclical changes in state appropriations affect public universities’ decisions and student access. Using institution-level panel data from 2002–2019, I exploit cross-state differences in balanced-budget rules and their interaction with local unemployment as an instrument for higher-education appropriations. I find that in states with stricter balanced-budget requirements, downturns lead to substantially larger cuts in state support for public universities. In response to these cuts, institutions primarily adjust along the enrollment margin rather than tuition: undergraduate enrollment falls contemporaneously and in subsequent years, while tuition responses are limited and imprecisely estimated. Universities also shift their program mix toward lower-cost fields. The results highlight how fiscal rules can indirectly restrict access to public higher education during economic downturns.

This paper investigates how a temporary income shock brought about by the 2021 expansion of the U.S. Child Tax Credit (CTC) affects local crime rates. The expanded CTC provided a sizable, temporary increase in disposable income for low-income households with children. Using administrative arrest data from the 2021–2022 National Incident-Based Reporting System (NIBRS) and a difference-in-differences framework, we exploit cross-county variation in exposure to the CTC driven by the number of eligible children and household income. We find that each additional $1,000 in CTC payments received by eligible families at the county level is associated with 0.15 more arrests per 1,000 residents. We detect no effect on violent offense arrests and a small negative effect on property offense arrests. The results are consistent with previous literature in which cash transfers show little or no clear reduction in crime.

More than one-third of opioid-related deaths are attributed to secondary users of prescription opioids with two-thirds obtaining painkillers from family and friends. Using a randomized control trial, we test the effectiveness of a community-based medication return program. Treated patients were 3.0 percentage points more likely to return medication relative to the control group, yet these effects remained small even when coupled with a text reminder, financial incentive, mail back envelope, or clinician involvement. Our results suggest that while medication return programs may remove opioids from some households, they are unlikely to meaningfully reduce the supply of opioids within a community.

Using the public release of ChatGPT in November 2022 as a natural experiment, we explore the effects of generative AI on the software developer job market, documenting three new stylized facts. First, using the near-universe of online job vacancies, we find the widespread introduction of generative AI resulted in a 16.3 percent drop in the relative proportion of junior- versus senior-level software developer job vacancies. Second, we find that the relative decline in labor demand for junior- versus senior-level software developers appears concentrated among larger firms and bigger cities while industries with moderate exposure to the software industry were more insulated from the effects of ChatGPT. Third, we reveal alternative career pathways for potentially displaced junior software developers using an occupation similarity network to identify credible job opportunities for junior software developers that require minimal re-skilling. Our findings are consistent with both recent industry observations as well as the vast literature on skill-biased technological change, reinforcing the notion that technological disruptions do not uniformly eliminate employment but rather reshape the labor market.