Commuting Time and Urban Violence in Brazil

Empirical evidence about the influence of exposure to public spaces on victimization strongly support the routine activities theory but, maybe reflecting the difficult of available data, specific evidence about the influence of the commuting on probability of victimization is not abundant. As registered by United Nation Office on Drugs and Crimes (UNODOC, 2012), Brazil is one of the most violent country of the world, with homicide rates around 27.1 (homicides per one hundred thousand people) in 2011, the third highest rate among Latin America countries (behind of only Colombia and Venezuela). This situation, in fact, reflect a general situation of high violence related to other kinds of crime in the country; as related to the violence associated to robbery, for example, the numbers of UNDOC (2012) for 2010 put Brazil, with rates (occurrences per one hundred thousand) of robbery and of theft among the three most violent Latin American Countries. But the problem of urban violence is neither the only substantive urban problem of Brazilian big urban centers, nor it is dissociated to other urban problems in these centers. Besides the risk of being victim of urban violence, visitors or inhabitants of Brazilian metropolitan regions must face with the problem of low mobility in these cities. The very bad quality of public transport together with public indirect subsidies for using individual transport make short distance locomotion a very high time demand action (IPEA, 2013). According to the more recent information of PNAD (PNAD 2012), the average commuting time for the inhabitant of Brazilian metropolitan regions was around 40.8 minutes in 2012, a very high number if compared to metropolitan regions around the world (Pereira and Schwanen, 2013; Silveira Neto et al. 2014). In this paper, we analyze this relationship using a large nationally representative cross-section sample of Brazilian individuals for 2009 using more traditional multivariate regressions and propensity score matching techniques to create counterfactuals. We also perform robustness checks, by applying different estimators (Abadie and Imbens, 2002), and implement a simulation-based sensitivity analysis that supports a causal interpretation of the results (Ichino wt al. 2008). We find that individuals with more than one hour of commuting have an overall 2.1% increase in the probability of being victim of robbery, with no robust impact on theft. Also, following the exposure literature we find larger effect on the probability of robbery victimization on women when compared with men, 2.5% and 2.2% respectively.