Condicionantes da taxa de homicídios nos estados brasileiros: novas evidências a partir de modelos dinâmicos de dados em painel

Abstract

This paper analyzes the determinants of the homicide rate in the Brazilian states from dynamic panel data models for the 27 states in the period between 1995 and 2009. The study has as contribution the use of robust panel estimators in the presence of endogeneity estimated by System-GMM. The results indicate that there is a strong persistence in the homicide rate; that inequality contributes to its growth; that poverty and extreme poverty have statistically zero effects and that education contributes significantly to coping with the violence that devastates the Brazilian states. Finally, the analysis of elasticities indicates that investment in education is more promising than spending on public safety to reduce the homicide rate.