Ensaios em Criminalidade no Rio Grande do Sul

• Ano de defesa: 2017
• Autor(a) principal: Cortes, Renan Xavier
• Orientador(a): Fochezatto, Adelar
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Pontifícia Universidade Católica do Rio Grande do Sul
• Programa de Pós-Graduação: Programa de Pós-Graduação em Economia do Desenvolvimento
• Departamento: Escola de Negócios
• País: Brasil
• Palavras-chave em Português: Índice de Criminalidade Municipal; Inferência Bayesiana; Horário de Verão; Visualização de Dados; Cadeias de Markov
• Área do conhecimento CNPq: CIENCIAS SOCIAIS APLICADAS::ECONOMIA
• Link de acesso: http://tede2.pucrs.br/tede2/handle/tede/7797

This thesis approaches the theme of crime in Rio Grande do Sul (RS) through four essays. The first one is devoted to an in-depth discussion of crimes in the municipalities of Rio Grande do Sul, emphasizing the problems related to the raw rates estimations, especially in small municipalities. This article, in addition to making an extensive descriptive analysis of municipal crime in RS, seeks to overcome the problems of estimation by using the Bayesian Integrated Nested Laplace Approximation (INLA) approach to create aggregate crime rates. Using the law penalties for each crime type, we also created disaggregated indexes against economic patrimony and against life and the approach used was efficient in reducing the variability of municipalities with high volatility. The second article assess the effect of luminosity on RS crimes, estimating the effect of the Daylight Savnig Time (DST). Several discontinuous regressions, both at the beggining of DST and at the end, were estimated. Possible effects were controlled, as weekday and climatological variables, and the results indicate that, mostly, there is no significant effect of DST on RS crimes. These results were robust to different model specifications. The third refers to a new tool for interactive data visualization of criminal occurrences in RS, CrimeVis. This visualization platform is constructed using Shiny technology and the R language with several types of visualizations. This article discusses the importance of data visualization, exemplifies the construction of CrimeVis through codes and shows how it is a powerful device for understanding the criminal dynamics of RS. The fourth, and last, essay seeks to study the temporality and space-temporality of crimes in RS through Markov Chains. This article estimates the transition probabilities and odds ratios of municipalities between absence/presence of crime of an initial period in relation to a presence/absence of a final period. In addition, the spatial neighborhood joint effect of the initial period is also measured. Evidence has shown that there is a displacement of cars robbery to the metropolitan region of Porto Alegre and a strong neighborhood effect in the most recent transition from 2015-2016, a temporal stability in homicides (with space-time neighborhood effects) and narcotics tracking with denned regions of occurrence between the north and the south of the state and a highlighted structural break in the crimes related to the arms and ammunition between the year of 2003 and 2004.