Efeito do policiamento sobre a criminalidade: uma análise em painel para o municípios paulistas e estados brasileiros

This work had the aim of investigating the effect of policing over crime rates, analyzing both the municipalities of São Paulo as well the Brazilian states. The analysis of counties developed for four types of criminal practices classified as theft, burglary, robbery and vehicle theft and murders, while the second analysis of the Brazilian states was restricted to murders. The São Paulo counties crime data were extracted from the Bureau of Public Security of the State of São Paulo and the Brazilian state data has the Datasus/SIM as the main source. The model used is based on the economic theory of crime proposed by Becker (1968). The attempt of measuring the police effect over crime was carry on for 2 variables: the police force and the public security expenditure. Besides those variables, control variables were added to the regressions such as demographic and socio-economics, compiled from various sources. Models were estimated with fixed effects panel methodology, controlling for the simultaneity bias of the relationship between policing and crime in a two stages regression. The results, especially in the regressions for the Brazilian states, illustrate the positive bias of simultaneity between police and crime, once the firsts regressions, that did not control for this endogeneity, resulted in positive parameters, while all regressions in two stages resulted in negative coefficients for the police variable, which are significant when expenditure is used as public security measure. The São Paulo counties regressions, we found evidence that the existence of a municipality guard may have a reduction effect over thefts rate.