Simon Voss (Freie Universität Berlin)
In this paper, I present causal evidence on the effect of homeless housing sites on local crime rates. Theoretically, homeless shelters could attract more homeless individualsand potentially more criminal activity turning these areas into crime hot spots. On the other hand, shelters may reduce street homelessness by increasing safety and surveillance, which could ultimately lower neighborhood crime, making the overall impact of homeless shelters on local crime ambiguous. Despite frequent community opposition to new shelter openings, empirical research on this relationship is limited. I use a unique dataset that geocodes all homeless housing sites in the city of Los Angeles. To identify a causal effect of these sites on crime, I implement a Difference-in-Differences design with spatial spillovers. This approach uses locations of crime incidents and shelter sites to define treatment and control areas based on concentric rings around shelter locations. My results show that the opening of a homeless shelter does not affect overall crime, while it significantly increases the prevalence of homeless-related crime. A possible explanation is that these shelters attract additional homeless people to the area, increasing both the pool of potential offenders and vulnerable targets.