A Dynamic Programming Approach for the Game of Corruption with Non-Trial Resolutions: The Brazilian Example

Sanction reduction policies for self-reporters and collaborators may be effective in deterring collusive corruption. In order to test this hypothesis, this paper builds a theoretical model that solves the game of corruption with and without sanction reduction policies, using a dynamic programming setting to find equilibrium strategies. Results show that these policies may effectively deter bribery. Additionally, it provides a way to measure policy effectiveness over non-observable corruption crimes by utilizing the number of corruption crimes. The model predicts that a successful anti-corruption policy shock will lead to an increase in corruption detections, even if the probability of detection remains unchanged. The paper tests the introduction of the Brazilian anti-corruption policy of 2013, finding significant signs of immediate anti-corruption shock and evidence of further deterrence.