Effects of Natural Hazards on Spatio-Temporal Patterns of (Violent) Crime in the United States
Abstract
The consensus that disasters do not cause an increase in crime rates is receiving renewed attention. In recent years, research has emerged that challenges this consensus by positing that crime rates and the type of crime may vary depending on the phase of the emergency. To address this, this research utilizes comprehensive crime data from the National Incident-Based Reporting System and hazard event data from the Spatial Hazard Events and Losses Database for the United States. Employing regression discontinuity design principles, swaths of linear regression models across different time scales are fitted, yielding nearly 200 statistically significant coefficients. The findings reveal correlations between certain natural hazard types and changes in crime rates. For instance, a correlation between winter weather hazard events and a subsequent drop in crime rates is observed whereas severe thunderstorms were associated with an increase in crime rates. Additionally, an increase in crime rates following natural hazard events that were observed in the shorter time scales (e.g., hail, tornadoes) did not persist into the longer time scale, where, in fact, negative treatment effects and a negative change in trend were found. These results shed light on the complex relationship between natural hazards and crime rates, providing valuable insights for policymakers, law enforcement agencies, and other stakeholders. Given that the intensity and frequency of natural hazards are on the rise, a better understanding of these dynamics can increase the efficiency of resource allocation for public safety and target the deployment of law enforcement more effectively.
