Dissertation Defense: “Essays in the Economics of Crime and Health” Michael Topper

This dissertation contains three chapters on the economics of crime and health. In Chapter 1, we study how technology is integral to police departments, automating officer tasks, but inherently changing their time allocation. We investigate this by studying ShotSpotter, a technology that automates gunfire detection. Following a detection, officers are dispatched to the scene, thereby reallocating their time. We leverage this shock to officers’ time allocation using the rollout of ShotSpotter across Chicago police districts to study the effects on 911 call response. We find substantial consequences---officers are dispatched to calls slower (23%), arrive on-scene later (13%), and the probability of arrest is decreased 9%. Consequently, police departments must evaluate their resource capacities prior to implementing technologies.

In Chapter 2, I exploit variation in timing from 44 temporary university-wide halts on all fraternity activity with alcohol (moratoriums) across 37 universities over 2014-2019. I construct a novel data set, merging incident-level crime logs from university police departments to provide the first causal estimates of the effect of moratoriums on reports of alcohol offenses and sexual assaults. In particular, I find robust evidence that moratoriums decrease alcohol offenses by 26%. Additionally, I find suggestive evidence that moratoriums decrease reports of sexual assault on the weekends by 29%. However, I do not find evidence of long-term changes once the moratorium is lifted.

Finally, in Chapter 3, we study the prevalence of gunfire, which results in loud and potentially stress-inducing sounds that may adversely affect critical stages of in utero development. However, gunfire is largely unreported, creating a unique challenge for researchers to understand its consequences. In this paper, we mitigate this shortcoming by leveraging data from ShotSpotter—an acoustic gunshot technology which uses an array of sensors placed on city structures to detect the sound of gunfire. We combine this unique data source with the universe of births from nine California cities, each matched to a mother's residence. Using the variation in gunfire detections from ShotSpotter at the census-block level, we employ a difference-in-differences methodology and find that gunshot noise creates substantial increases in very low birth weight (< 1,500 grams) and very pre-term births (< 32 weeks). These effects are driven by times of the day when mothers are likely to be at-home, and are particularly concentrated among mothers with low levels of education. These results suggest that gunshot noise is a major factor contributing to the income inequities in pregnancy outcomes.