Corruption and electoral accountability in Brazil

This dissertation examines how voters react towards candidates with records of misuse of public funds in the context of sub-national elections in Brazil. Its contribution to the extant literature on corruption and electoral accountability is twofold. First, it is the first study to inquire whether voters punish candidates with malfeasance records running for both executive and legislative office in the same electoral context and whether a number of contextual factors affect electoral accountability in these offices. Second, it presents and tests new hypotheses on the type of motivation that ground voters' rejection towards corrupt candidates. In chapter 2, I examine whether voters punish candidates for mayor and city councilman with accounts rejected by the Brazilian Audit Courts and whether additional contextual factors affect electoral accountability. In particular, I study whether electoral accountability decreases as candidates (for mayor) have better records of social provision; whether local media promotes electoral accountability; and whether candidates with negative antecedents receive fewer campaign donations and are less likely to re-run. I combine large-N observational analysis, using an original dataset with candidates' accounts rejection records, with interviews with Brazilian Audit Court members and local politicians. In chapter 3 I use three online survey experiments with a convenience sample of Brazilian voters to examine whether likelihood to support a corrupt incumbent is affected by the details that subjects learn about the corruption incident. I use these additional details to inquire whether subjects are sensitive to information emphasizing the public costs of corruption, the candidate's moral misbehavior, or his illicit enrichment. Results presented in chapter 2 suggest that prior records of misuse of public funds have electoral consequences both for candidates for mayor and for city councilman. In addition, they suggest that the existence of local media does not increase electoral punishment; that public spending does not reduce electoral punishment; and that candidates with accounts rejected often receive fewer funds and are less likely to re-run. Results presented in chapter 3 suggest that voters' rejection towards corrupt candidates is stronger when they learn additional details on the candidate's ilegal enrichment