Research

We document a sharp rise in gunshots coupled with declining 911 call volume across thirteen major US cities in the aftermath of the murder of George Floyd. National survey data also indicate that victims of crime became less likely to report their victimization to law enforcement due to mistrust of police. Our results suggest that high-profile acts of police violence may erode community engagement with law enforcement and highlight the call-to-shot ratio as a natural measure of attitudes towards the police. 

Time scarcity is one of the strongest correlates of fast food consumption. To estimate the causal effect of time lost on food choice, we match daily store-specific foot traffic data traced via smartphones to plausibly exogenous shocks in highway traffic data in Los Angeles. We find that on days when highways are more congested, individuals are more likely to dine out and less likely to grocery shop. The effects are particularly pronounced for afternoon rush hour traffic. Our results imply a net reduction in healthy food store choice due to time lost.

In the United States, law enforcement officers serve as first responders to most health crises, allowing them to connect many more individuals to treatment services than other government actors, a fact that has come into increasing focus due to the opioid epidemic. In response, police departments across the country have begun to divert individuals that possess narcotics away from arrest and towards treatment and recovery. Evidence on whether these programs are able to engender meaningful change—initially by increasing participation in substance use treatment, and eventually by reducing the likelihood of continued drug use and criminal justice involvement—remains limited. This paper aims to shed light on the potential of these programs by exploiting the eligibility criteria for and staggered rollout of narcotics arrest diversion in Chicago between 2018 and 2020 using a difference-in-difference-in-differences framework. We find that the program reaches individuals with medically diagnosed substance use disorders, increases connections with substance use treatment, and reduces subsequent arrests. We conclude that Chicago's drug diversion program is able to simultaneously reduce the reach of the criminal justice system, expand the number of individuals with substance use disorders connected with treatment, and improve public safety.

Drug crimes continue to make up a large share of offenses for which individuals interact with the criminal justice system, and Black Americans are four times as likely to be arrested as white Americans despite similar usage rates. Policymakers across the country have introduced both deregulation and full legalization of the recreational use of marijuana with the intention of reducing these disparities. Yet, we do not know the impact of this approach on who and how the police arrest. In this paper, we exploit the rollout of four deregulatory policies from two branches of government across the three largest US cities, New York City, Los Angeles, and Chicago, using a difference-in-differences design. We find that executive-branch policies (nonprosecution and nonarrest) prove more effective in reducing the number of marijuana arrestees across all racial and ethnic groups than legislative-branch policies (decreased penalties and legalization) do. At the same time, once examining outcomes for specific racial and ethnic groups, we find the opposite— legislative branch policies decrease racial disparities, while executive branch policies exacerbate them. We develop a novel metric to identify marijuana possession’s complementary arrests, and find that these decrease after legalization. We find broadly similar patterns for substitute arrests and for arrests with the most police discretion as well. Lastly, we observe an extensive spillover effect of small-quantity marijuana possession becoming less regulated, where arrests for small quantity possession for those 18 to 21—an offense never deregulated—significantly decrease as well.

This study analyzes the impact of crime on mental health in the United Kingdom, using daily, small neighborhood level measures of reported crime and individual stress to estimate the impact of over 1 million crimes on the stress level of those in the crimes' vicinity. We find that violent and sexual crimes increase stress in their vicinity, and the impact lasts for approximately 3 days. Separating the impact of the neighborhood from that of a specific recent crime, we show that property crimes have no such effect. Further, we find a one day lag between the timing of the crime and the heightened stress response. Exploring the mechanism of news media as the mediator of information, we observe that on days with crime being covered as front page news individuals report heightened stress.