Subnational Determinants of Corruption: The Case of Brazil
MetadataShow full item record
This dissertation project assesses the role that different factors play in curbing municipal- and individual-level corruption. Over the course of three essays, this project explores the extant literature on determinants of corruption with the goal of both building on this literature and proposing novel insights to help further our understanding of the causes of corruption. The first essay explores the effects of government transparency on levels of municipal corruption. I leverage a random municipal auditing program in Brazil to create an original data set and construct a novel natural experiment that assesses how revelations of mayoral corruption affect future levels of corruption in the audited municipalities. The results indicate that, in municipalities where government malfeasance is exposed, this exposure eventuates a decrease in future levels of corruption; however, this attenuating effect decreases over time. In the second essay, I probe for the existence of an unintended benefit of participatory budgeting institutions, which are a form of direct democracy in which citizens vote on the allocation of part of their municipalities’ budgets. I argue that the increased participation and availability of information inherent to participatory budgeting institutions allow citizens to better monitor, and hold accountable, their governments. Using a panel data set of Brazilian municipalities, I find that municipalities with participatory budgeting institutions are less likely to suffer from government corruption, compared to those municipalities that do not practice participatory budgeting. Finally, the third essay introduces a new theoretical and empirical explanatory factor underlying individual-level motivations to engage in corruption: namely, context. Specifically, I argue that existing work on this subject, which largely explains corruption as a function of economic or cultural considerations, fails to account for the immediate context within which the decision-making process occurs. This gap in the study of individual-level determinants of corruption is both theoretical, as it does not fully describe the process through which individuals decide to behave in a corrupt manner, and empirical, as it suffers from a key omitted explanatory factor, leading to biased results. I deploy a unique survey experiment in which context is introduced on both theoretical and empirical levels. The results show that context does matter, and further indicate that failing to consider context has a distortionary effect on empirical outcomes.