UMA ESTIMATIVA DOS DETERMINANTES DA TAXA DE CRIMINALIDADE BRASILEIRA: UMA APLICAÇÃO EM PAINEL DINÂMICO

Using a panel data for Brazilian states in the period 1984-1998, we estimate the determinants of Brazilian crime rate. The estimates was derived by System Generalized Method of Moments propose by Blundell and Bond (1998). In this way, it was possible to avoid econometric problems, that affect the majority of the works in this area, such as endogeneity and measure of error generated by the subnotification on crime rates. The results show that inequality and crime rates of the previous period have positive effect on crime rate of the present period, while the per capita GDP, the level of schooling, the degree of urbanization and the GDP growth have a negative effect. Moreover, the use of the First Difference Generalized Method of Moments estimator (Arellano and Bond, 1991), in this case, did not reveal to be appropriated due to weak presence of weak instruments. The coefficient was below the Within-Groups estimator. The use of System MGM show to be more informative in terms of bias and efficiency.