The Impact of Smoking Bans in Bars and Restaurants on Alcohol Consumption and Smoking

Abstract: Governments implemented bar and restaurant smoking bans to target smoking-related externalities, but these bans may also affect drinking. This paper studies smoking bans’ effects on alcohol consumption and smoking behavior. I estimate a difference-in-differences model that exploits spatial and temporal variation in smoking bans. Bans result in a 1-drink-per-month (5%) increase in intensive-margin alcohol consumption, driven by changes in bar and restaurant consumption. I find no economically meaningful effects on extensive-margin smoking. These results imply that smoking bans lead to unintended consequences in the form of increased alcohol consumption.

Ban the Box and Cross-Border Spillovers

with David Wasser

Abstract: Ban-the-box (BTB) policies prevent employers from asking job applicants about their criminal history until very late in the hiring process. The intent of these policies is to help individuals with criminal records find employment by reducing the stigma associated with arrest or conviction. However, such a policy could induce employers to statistically discriminate on the basis of observable characteristics, such as race, if employers believe that certain racial groups are more likely to have a criminal history. This paper studies whether labor market effects spill over into jurisdictions bordering those with BTB policies. Using the 2005-2014 waves of the American Community Survey and a difference-in-differences method, we find that annual earnings for employed, young, non-college-educated black men increase by 9.5 percent following the implementation of BTB in neighboring labor markets. Earnings for young, non-college-educated Hispanic and white men are flat. The effects for black men do not appear to be driven by changes in the composition of the sample of employed workers.

Do Uber and Lyft Reduce Drunk-Driving Fatalities?

Abstract: This paper investigates whether Uber and Lyft lead to reductions in drunk driving, as measured by city-level drunk-driver-related motor vehicle fatalities and fatal crashes. I use a difference-in-differences method that exploits the variation in the timing of Uber and Lyft entry for the 100 most populous U.S. cities and a Poisson model to account for the fact that crashes and fatalities are count data. Using monthly city-level Fatality Analysis Reporting System (FARS) data for 2006 to 2016, I find small declines in drunk-driver-related fatal motor vehicle incidents and small increases in overall fatal motor vehicle incidents, but I cannot reject the null hypothesis of no effect of Uber or Lyft on these outcomes. Event studies suggest that drunk-driver-related and overall fatal motor vehicle incidents decline several years after the entry of Uber or Lyft into a city.