UC San Diego

This study investigates two important types of violence, with a central focus on the case of Brazil. As in many countries, political assassinations are prevalent across Brazil, but receive minimal attention especially when involving local politicians. This study develops a typology to conceptualize political assassinations and introduces an unprecedented dataset of assassinations of municipal politicians in Brazil. To generate this dataset, I obtained official permission to cross-reference undisclosed vital statistics data on over 10 million homicides with a registry of all political candidates between 2001 and 2017. Using these data, descriptive statistics and regressions provide insights about patterns and correlates of political assassinations, at the individual as well as municipal level. Moreover, the present study elaborates and tests why corruption is a causal determinant of political assassinations. Corruption increases political assassinations for two key reasons: (1) politicians use assassinations as a way of enforcing their corrupt agreements with other politicians, and (2) corrupt politicians kill other politicians who may subject them to prosecution by engaging in whistleblowing. The case of Brazil presents an opportunity to test this causal relationship rigorously, as random anti-corruption audits have been shown to exogenously reduce municipal corruption. Findings are consistent with the theoretical argument. Not only are fewer politicians assassinated in municipalities that recently experienced a random anti-corruption audit, but these effects are also amplified when larger government revenues render it more worthwhile to engage in this form of costly punishment. This study then shifts to examine another crucial topic: criminal violence. Using a precise definition of criminal groups, it constructs a dataset mapping the temporal and geospatial distribution of organized criminal groups in Rio de Janeiro, as well as their economic activities. Of central focus is how criminal groups' interactions with the state and other criminal groups determine their use of violence and their capacity to diversify economically. Altogether, this study advances scholarly understanding of two fundamental aspects of violence, which have important substantive consequences for Brazil and beyond.