Prevenção e controle de homicídios: uma avaliação de impacto no Brasil

In this article we assess the impact of the Fica Vivo programme, for the prevention and control of criminality, set up in Brazil. Its design is inspired by the American programme CeaseFire and the principal objective is the reduction of homicides in areas where its incidence is high, in general, slums. The impact variable is the half-yearly homicide rate per one hundred thousand inhabitants and the methodology used is Double Difference Matching (DDM) with Propensity Score Matching (PSM). We utilize a specification of the model that allows us to consider the impact of the programme in a differentiated way for each area treated and for each stage of expansion. The data utilized are the georeferenced records of the Military Police of Minas Gerais running from 2000 to 2006, and data from the demographic census of 2000. The principal results indicate that Fica Vivo reduces criminality, although the effect of the programme is not homogeneous over the areas treated, and has increased with the passage of time during the period studied.