Did police expenditures and COPS grants reduce crime in large cities?
Before proceeding to the findings, an explanation about the presentation of the findings in table format is necessary. The crime dependent variables are measured on a per capita basis, so the coefficients for police expenditures and COPS grants are small. In order to improve the visual presentation of the findings, these coefficients and their standard errors are multiplied by 1 million. Table 5 presents the crime regression results. For the robbery and larceny 2SLS models, the standard errors are robust to heteroskedasticity and autocorrelation.[51] For all the other models, the standard errors are robust to heteroskedasticity.[52] For the murder models, the only law enforcement variable to deter murders is the innovative grant variable in the OLS model. A $1 increase in innovative grants per capita is associated with a reduction of 0.698 murders per 1 million persons. This finding is statistically significant at the 0.05 level. For the elasticity, a 1 percent increase in inno
