The purpose of this paper is two-fold. First, we describe an estimation procedure that should be useful for spatial models which contain interactions between the dependent variables and autocorrelated error terms. Second, we apply that procedure to a spatial model relating to county police expenditures. Our estimation procedure does not require the specification of the error distribution, and its computational burden is modest. Large sample results are given.
No takes yet. Share an insight, caveat, or question.
Kelejian et al. (1993) studied this question.