Working Paper: NBER ID: w28683
Authors: Frederico Finan; Maurizio Mazzocco
Abstract: In this paper, we develop a dynamic model of politicians who can engage in corruption. The model offers important insights into what determines corruption and how to design policy to combat it. We estimate the model using data from Brazil to measure voters' willingness to pay for various commonly-proposed anti-corruption policies, such as increasing audit probabilities, increasing politicians' wages, and extending term limits. We document that while voters have a high willingness to pay for audit policies, due to their effectiveness in reducing corruption, a multi-pronged approach that bundles certain policies can achieve similar welfare gains at a fraction of the cost.
Keywords: No keywords provided
JEL Codes: D72; D73; O1
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Increasing the probability of federal audits from 5% to 16.8% (H83) | Decrease in corruption (H57) |
| Extending the number of consecutive terms a mayor can hold (K16) | Decrease in corruption (H57) |
| Raising wages (J38) | Incentivize corruption among politicians not seeking reelection (D72) |
| Combining the clean record act with term limits (K16) | Increase in average willingness to pay for policies (G52) |
| Selection effects account for differences in stealing between first and second-term mayors (C34) | Differences in stealing (K42) |