Violence, Vote Choice, and the Provision of Public Security in Brazilian States

This dissertation examines the relationship between politics and violence in Brazilian states. I challenge the assumption that ideology alone leads political parties from the left and right to implement different public security policies. Using a mixed method approach, I argue that politicians’ responses to violence depend on how security issues affect their electoral success. This process occurs in a four-stage cycle that involves elections, the provision of resources for law enforcement, the distribution of these resources and violence. Chapter 1 uses the lens of retrospective voting to examine how violence influences elections. I show that contrary to Western democracies, where conservative parties gain when security issues arise, security is a double-edged sword for Brazilian political parties. Voters punish/reward right parties more harshly than they punish/reward leftist parties for security failures/successes. In Chapter 2, I then analyze the policies that governors implement to address crime rates. Using state budgetary data, I show that while left and right parties advocate different security policies during the campaign, in office, they both turn to law enforcement to address crime. However, their spending priorities depend on political competition. Right parties spend more on law enforcement when they face leftist competitors that they can paint as being soft on security, while leftist parties spend more on security when they face centrist or other leftist competitors. Chapter 3 turns to the issue of how governors distribute law enforcement resources. I show that governors send more police officers to large municipalities that are important for their electoral success. Chapter 4 then integrates the findings from previous chapters to examine how resources and ideology influence municipal homicide rates. I show that homicide rates decline in states where the governor is from a right party and when states spend more on education. However, the effects that law enforcement has on crime depends on state wealth and the cities voting behavior. In the concluding chapter, I summarize the main findings and discuss the project’s overall importance