Political corruption: Theory and evidence from the Brazilian experience

In the classic principal-agent model of corruption, citizens cannot observe the public servant's corrupt behavior. Corruption results from the combination of asymmetric information and the opportunistic behavior of the public servant. 1 In my dissertation, I introduce the assumption that voters observe that politicians are corrupt. I develop a model of political corruption with a heterogeneous electorate. Middle class voters pay taxes and prefer politicians who spend tax revenue on public goods. Lower class voters do not pay taxes, and prefer public and private goods. Private goods are given from individual politicians to individual voters and typically take the form of basic necessities, such as foodbaskets. To get elected, politicians need campaign contributions, which come from clients and interest groups. Clients provide illegal kickbacks from income received from over-invoiced government contracts. Interest groups provide legal campaign contributions. I show that the introduction of a lower class that values private goods decreases the costs of corruption. The model also highlights that the costs and benefits of corruption depend on the relative sizes of the middle and lower class voters.;I test the model with data collected during a year of field research in Brazil. I supplement the statistical work with qualitative evidence received from interviews. In the empirical analysis I segment the data by education levels, which are my proxies for economic classes. The dependent variable is percentage of the electorate by district who voted for known corrupt politicians. The data come from elections in 1998 from two regions in Brazil, one election determining a senator and the other a governor. The empirical results in both cases are consistent with the hypothesis that corrupt politicians receive more votes in electoral zones characterized by lower class voters. In particular, illiterate voters overwhelmingly voted for the corrupt candidate. Middle class zones characterized by relatively higher levels of education voted for the non- or relatively less-corrupt candidate.;1Rose-Ackerman (1978) and Shleifer and Vishny (1993)...