Abstract
Using a two-stage least-squares procedure, we estimate the relationship between ethnic fragmentation and police spending using a cross-section of the US counties. Our results show that, when controlling for community characteristics and accounting for simultaneity bias, ethnic fragmentation is positively related to police spending. This article contributes to the understanding of the stylized fact that public spending on police increased over a period in which the incidence of crime decreased.
Acknowledgement
This research was supported by Rutgers University Research Council Grant # 202084.
Notes
1All crime statistics are from the Bureau of Justice Statistics.
2Throughout this article we use fragmentation as our measure of heterogeneity. Fragmentation is defined as the probability that two randomly selected people in a community are of a different ethnicity. We focus on ethnic fragmentation rather than racial fragmentation as we wish to distinguish between Hispanics and other racial groups.
3A number of authors have used it as a measure of diversity, for instance, Alesina et al. (Citation1999) and Hero and Tolbert (Citation1996).
4These are the customary variables used in the demand for public spending literature. For instance, see Poterba (Citation1997).
5Kovandzic and Sloan (Citation2002) perform a similar analysis of a different data set and arrive at similar conclusions.
6The results in Levitt (Citation1997) are not uncontroversial. See Levitt (Citation2002) and McCrary (Citation2002) for a discussion.
7The deployment of police to certain neighbourhoods has also been demonstrated to reduce crime in those particular neighbourhoods (Di Tella and Schargrodsky, Citation2004).
8Several counties were lost during the merge of the demographic and financial variables. Also not included are counties in Alaska, Hawaii and the District of Columbia.
9This method has also been used by Alesina et al. (Citation1999).