Método dos momentos generalizados geograficamente ponderado

Abstract

Our study extends the technique of estimate of models linear regression geographically weighed models, popular in the literature of anaýsis of spaces datra, which allows estimative of local co-efficients using ordinary lesat squared or maximum likehood, generalized methods os moments models (GMM). This technique allows treat spatial of heterogeneity in cross-section models, common problem inspace studies; the estimative of models with presence of endogeneity by using instrumental variabels; the estimate of nonlinear models; and the estimative of parameters for each geographic unit or, in the case most general, for each unit of study. A pseudo-algorithm for implementation of approach proposal is presented, and two empirical aplications. One for estimative of labour and wage equations to study regional growth and development of the Brazilian municipalities, between 1991 and 2000, and another, to estimate insolvency risk of financial institutions in Brazil, with the purpose of demonstrate the aplicability this approach, in other fields of study out of space scope.