A large literature demonstrates the negative effects of stress during pregnancy on birth outcomes and the additional long-term negative educational, labor, and health effects of prenatal stressors (see e.g. Almond and Curry 2011). However, no previous study has identified the causal effect of criminal activity and the stress it potentially creates for mothers on birth outcomes in a US setting. In this study, we investigate the health effects of criminal activity. We make use of plausibly exogenous variation in criminal activity at a very fine geographic level (census tract or smaller) to examine how crime impacts natality outcomes. Specifically, if there are higher rates of face-to-face crimes, this could impact an expecting mother's willingness to exercise, travel to the store or attend medical appointments. To alleviate concerns about the exogeneity of criminal activity, we investigate plausibly exogenous shocks to criminal activity in the form of a shutdown in mass transit - namely, a bus strike in Los Angeles County - to determine how criminal activity responds to the lack of low cost, public transportation. This mass transit shutdown may impact natality outcomes directly by raising the opportunity cost of receiving prenatal medical care or shopping for nutritious, fresh foods, creating a negative income shock if expecting mothers or their partners cannot travel to work because of the shutdown or indirectly through shifting crime patterns created by the shutdown.
We use vital statistics restricted-access, geocoded natality data from California covering 1997-2013 and highly detailed criminal incident level data sets for Los Angeles County. We have access to both reported crimes and overall arrest rates for different types of crimes. The data measures criminal activity by incident hour, and is available at the reporting district level. This provides us with key variation required to identify our outcome measures. For instance, although crime at an aggregated level might be harder to be characterized as quasi-exogenous, daily variation in incidents in a given reporting district can, arguably, be considered to be plausibly exogenous to individual pregnant mothers living in those neighborhoods. Thus, using deviations from trend level of crime rates we measure the impact on pregnant mothers and their eventual birth outcomes as induced by stress related channels. Measuring the effect of criminal activity on health outcomes is an understudied, but important topic with substantial policy implications.
