criminal offense. On the other hand, there is uncertainty involved, at the part of the offender, whether a particular property crime would actually yield a firearm or not. Therefore, this partial randomness in the acquisition of firearms through property crimes along with lagging the firearm proxy can together substantially help allay the simultaneity concerns outlined above. As one would expect, given the massive spatial heterogeneity in crime levels across the U.S, as evidenced in Section 4, Eq. (1) needs spatial level covariates to take care of the unobservable jurisdiction level criminogenic factors that would be correlated with both firearm prevalence and crime levels. Thus, I augment Eq. (1) with lagged property crime category to account for variation in these unobservable crime processes. Cook and Ludwig (2006) also use controls for robbery and burglary in a county motivated by similar concerns. To minimize serial correlation concerns, when the dependent variable is a property crime like burglary or motor vehicle theft, this aggregate is all property crime in the previous time period except the one under consideration as the dependent variable. This provides the following estimating equation:
การแปล กรุณารอสักครู่..
