"Pharmaceutical Drug Regulation and Mortality: Evidence from E-cigarettes" (with Michael Pesko) Submitted to RESTUD

Media Highlights: Marginal Revolution, Vox, National Affairs

This study evaluates drug regulation in the United States by examining the effects of the unexpected judicial exemption of e--cigarettes from drug regulation, compared to nicotine replacement therapy, which remained regulated as a drug. We find that this exemption led to significant increases in innovation, as evidenced by a rise in patent applications. Using variation in smoking rates across demographic groups prior to e-cigarette introduction, we estimate that from 2011 to 2019, e-cigarettes saved 677,000 life-years-approximately one-third of the estimated benefit of early HIV/AIDS drugs by year 2000-and increased social surplus by $8 billion annually. We show that reduced smoking is a key mechanism behind this mortality reduction, with statistically significant declines in mortality lagging smoking reductions by approximately four years.

"Sports Spectating and Drunk Driving Deaths: Evidence from the Super Bowl" Submitted to American Journal of Health Economics

Sports spectating is frequently linked to alcohol-related harms such as drunk driving. In the context of sporting events, alcohol is frequently consumed in social settings according to the emotional state of the drinker. Using the Super Bowl as a natural experiment, I study the effect of sports spectating on drunk driving fatalities across 42 Super Bowls using a regression discontinuity-in-time design. I find that the Super Bowl increases motor vehicle fatalities by 17 percent increase at an annual social cost of $161 million within a 24-hour span. Emotional cues play an important role in explaining drunk driving deaths: wins increase fatalities more than losses but upset losses are comparatively more harmful than upset wins. My results point to sports spectating and emotion-induced drinking as important determinants of drunk driving.

"Ridesharing and External-Cause Mortality" (with Keith Teltser and Conor Lennon) R&R at Journal of Public Economics

Media Highlights: National Affairs

Existing research suggests ridesharing affects transportation access, substance use, and exposure to crime. Ridesharing may therefore also have meaningful effects on mortality related to these underlying factors. To address this question, we use restricted-access data from the National Vital Statistics System to study how ridesharing affects mortality from external causes. Our identification strategy relies on spatial and temporal variation in UberX entry across U.S. counties. Among those aged 18 to 45, we find that UberX entry into an area is associated with 2.01 additional deaths per quarter per 100,000 population (roughly a 10% increase). We find that these deaths are primarily related to alcohol and drug use. We support a causal interpretation for our findings by presenting event studies, placebo analyses, sensitivity and heterogeneity analyses, and a variety of robustness checks, including differences-in-differences estimators that are robust to heterogeneous treatment effects. Our work contributes to the literature by addressing both the benefits and costs of ridesharing, and by informing stakeholders who are interested in the cumulative impact of ridesharing on public health and well-being.

"Natural Light and Fatal Crash Risk: Evidence from Civil Twilight"

Poor visibility is a key factor in pedestrian traffic collisions. It is difficult, however, to isolate the causal effect of light from confounding variables and endogenous roadway selection. Estimating crash risk also requires measures of vehicle and pedestrian exposure, which is often unobserved or measured coarsely. In this paper, I propose the use of civil twilight as a novel source of identification in which natural illuminance sharply changes and produces a kink in pedestrian traffic collisions. Using minute-level data on fatal-crashes with exact latitude and longitude from 2001-2019, I estimate a regression kink design at civil dawn and dusk. I find that adequate natural lighting reduces the probability of a driver fatally striking a pedestrian by 20 percent. I document benefits across roadways and built environments and find comparable reductions in both driver- and pedestrian-instigated crashes. I document only modest reductions in vehicle-only collisions in intersections. My results suggest substantial benefits to increased visibility and are consistent with natural light improving pedestrian detection and increasing stopping distance even in areas friendlier to pedestrians with greater artificial light sources.