Research Projects

  • Driving Safety: An Empirical Analysis of Ridesharing's Impact on Drunk Driving

    Estimating the effect of ridesharing services like Lyft and Uber on drunk driving. This is my job market paper, available to download on the left of this page.

    This paper examines the effect of ridesharing services such as Lyft and Uber on the incidence of drunk driving. Ridesharing services are convenient, low-cost alternatives to traditional taxi cabs. I use the gradual expansion of ridesharing to cities across the U.S. to identify the effect of their introduction on alcohol-related traffic fatalities and DUI arrests. Using a large sample of U.S. cities and a fixed effects difference-in-differences approach I find that ridesharing services significantly reduce fatal alcohol-related auto accidents and for a large subset of cities DUI arrests as well. I explore the possibility of heterogeneous effects based on the quality of public transit available as well as the duration over which ridesharing has been operating. I further find that ridesharing's presence corresponds to a significant reduction in other potentially alcohol-related crimes such as physical and sexual assaults.

  • Quantifying the Effect of Rapid Transit on Drunk Driving

    Estimating the effect of rapid transit systems such as subways, light rail, and rapid buses on drunk driving rates.

    This paper estimates the causal effect of rapid transit systems on drunk driving. Many rapid public transit systems in the U.S. were developed over the past 40 years. Among the benefits of such systems is the potential for them to offer an alternative to driving under the influence of alcohol. In this paper I take advantage of the gradual build out of rapid transit systems to estimate the causal effect of increasing the size and scope of such networks on the incidence of drunk driving in those cities. Using a fixed effects difference-in-differences methodology I estimate the effect of the number of lines, stations and the interconnectedness of the systems' lines on the number of fatal alcohol-related auto accidents. I find that adding an additional transit lines and stations significantly reduce such fatal accidents.

  • Late-Night Bus Transit's Impact on Drunk Driving

    Estimating the effect of late-night bus service on drunk driving arrests in Austin, TX.

    This study estimates the effect of Austin, Texas's late-night municipal bus services on drunk driving arrests. Using a unique dataset containing the home addresses of individuals arrested for drunk driving I estimate the effect of late-night bus service on the number of arrests. Using a difference-in-differences methodology I measure the change in number of drunk driving arrests from days without late night service to days with these services for neighborhoods within walking distance of late-night bus routes. Comparing this change to the same change for neighborhoods not served by these routes allows me to identify the causal effect of late-night bus services on drunk driving arrests.

  • Risk in Numbers: Parsing Ridesharing's Effect on the Number and Riskiness of Drunk Drivers

    Separating the effect of ridesharing on both the number of drunk drivers and their likelihood of causing fatal accidents.

    Building on the framework developed by Levitt and Porter (2001) I exploit the inherent richness contained in data on two-vehicle fatal accidents to separately estimate the effect of ridesharing on both the number of fatal drivers on the road as well as on the risk of fatal accidents posed by those drivers. Separating these changes allows me to investigate whether ridesharing's presence induces substitution away from drunk driving by drivers who are relatively more or less risky than average. Understanding how the risk composition of intoxicated drivers is impacted by ridesharing allows for a better understanding of the types of individuals who continue to drive drunk after these services are introduced.