RACE ON POLICING: Results of Estimation 4
Table 3 presents a range of alternative specifications as a means of gauging the sensitivity of our estimates. The columns of Table 3 match those of Table 2. Each row represents a different specification. Only the differences in the own-race and cross-race arrest coefficients are reported (i.e. for odd columns the coefficient on white police minus the coefficient on black police, and for even columns the reverse). A negative value in Table 3 means that arrests are lower with own-race policing than with cross-race policing. The 72 entries in Table 3 (9 rows by 8 columns) represent coefficients from 72 different regressions. Coefficients that are statistically significant at the .05 level are highlighted in boldface.
For purposes of comparison, the top row of Table 3 presents the results of Table 2 as a baseline.21 The results obtained are robust to a wide range of specifications. Eliminating city-fixed-effects does remarkably little to change the results. Similarly, little changes when all of the covariates are eliminated except year and city dummies, or when region-year interactions (using the nine census regions) are added. The results are more sensitive to the inclusion of city-level trends and the standard errors rise as well.
Restricting the sample to cities with both a substantial Black population (>10 percent) and a small non-White, non-Black population has little systematic affect on the results. This suggests that lumping all non-whites into one category and including cities with few minorities in the sample is not greatly distorting the results. The coefficients shrink somewhat when robust regression techniques are used to reduce the influence of outliers. Nonetheless, three of the eight entries remain negative and statistically significant.
The two next-to-last rows of Table 3 present two stage least squares estimates using the racial composition of a city’s firefighters as an instrument for the racial composition of the police force. Even after controlling for the racial composition of city residents, the number of non-white municipal firefighters is a good predictor of non-white police, but is only weakly correlated with white police. Similarly, the number of white employees in those other functions is correlated with white police, but not with non-white police. Due to large standard errors, it is difficult to draw strong conclusions from the two-stage least squares estimates when city-fixed effects are included (the penultimate row). When city-fixed effects are dropped from the two-stage least squares regressions, however, statistically significant negative coefficients are obtained in four of the eight columns.
An important concern in interpreting the 2SLS coefficients is whether the exclusion of firefighters from the crime equation is valid. While one would certainly not expect a direct impact of the composition of the fire department on crime rates, it is possible that a large number of black firefighters is the consequence of other factors about a city that will influence crime, such as good race relations, or a thriving minority community. payday loan reviews