We estimate the causal real economic effects of a randomized anticorruption crackdown on local governments in Brazil over the period 2003-2014.After anti-corruption audits, municipalities experience an increase in economic activity concentrated in sectors most dependent on government relationships. Theseeffects spill over to nearby municipalities and are larger when the audits are coveredby the media. Back-of-the-envelope estimates suggest that $1 away from corruption generates more than $3 in local value added. Using administrative matchedemployer-employee and firm-level datasets and novel face-to-face firm surveys weargue that corruption mostly acts as a barrier to entry, and by introducing costsand distortions on local government-dependent firms. The political misallocation ofresources across firms plays a seemingly secondary role, indicating that at the locallevel most rents are captured by politicians and public officials rather than firms.
