Working Paper Available on SSRN

(with Uttara Ananthakrishnan and Anuj Kumar)

The prevalence of anti-Black bias in policing has long been of concern to social scientists and policymakers. This article studies a predecessor mechanism that constitutes an important source of policing bias in American society: calls by individuals to police to investigate “suspicious” behaviors, often of neighbors. We construct a novel data set of over 39 million 911 calls across 14 US cities from 2011 to 2019. This data, obtained through the digitization initiatives of local governments, provides us a unique opportunity to study neighborhood-level social cohesion and demonstrate how changes to a neighborhood’s composition lead to systematic biases against members of the minority community. Even in our most demanding specifications that used fixed-effect and instrumental variables, we find that the proportion of suspicion 911 calls and unfounded suspicion calls increase as more Non-Black residents move into neighborhoods, controlling for the population of other races and the overall crime in a tract. This effect is exacerbated in gentrifying neighborhoods and areas with high levels of internet penetration and Non-Black residents but can be mitigated through online activism, as in the case of the online #BlackLivesMatter movement. We conclude with a discussion of our results and implications for future work at the intersection of technology and policy.