Corruption and Firms: Evidence from Randomized Audits in Brazil

Posted: 14 Mar 2017 Last revised: 4 Nov 2017

Emanuele Colonnelli

Stanford University, Department of Economics, Students

Mounu Prem

Universidad del Rosario

Date Written: March 8, 2017


We exploit spatial variation in randomized anti-corruption audits related to government procurement contracts in Brazil to assess how corruption affects resource al- location, firm performance, and local economic activity. In an event study framework, we find that after an anti-corruption crackdown, regions experience more entrepreneurship, improved access to finance, and higher levels of economic activity. This is inconsistent with corruption acting as “grease in the wheel.” We find that two channels explain these facts: (i) allocation of resources to less efficient firms, and (ii) distortions in government dependent firms. Using firms involved in corrupt business with the municipality, i.e. “corrupt firms,” we show that the second channel is more important. Difference in difference estimation suggests that, after audits, the performance of corrupt firms improves relative to a similar set of unaffected firms. Corrupt firms invest more, increase borrowing and leverage, reallocate capital and labor within the organization, rely less on government contracts, and grow more. Our findings provide novel micro-evidence on why corruption acts as an institutional failure that is detrimental to firm performance and economic growth.

Keywords: Corruption, Firms, Audits, Public Procurement, Misallocation, Labor Reallocation, Political Connections

JEL Classification: D22, D72, D73, G30, G38, H57, K00, L22, O10, O43

Suggested Citation

Colonnelli, Emanuele and Prem, Mounu, Corruption and Firms: Evidence from Randomized Audits in Brazil (March 8, 2017). Available at SSRN: or

Emanuele Colonnelli (Contact Author)

Stanford University, Department of Economics, Students ( email )

Stanford, CA
United States

Mounu Prem

Universidad del Rosario ( email )

Casa Pedro Fermín
Calle 14 # 4-69

Register to save articles to
your library


Paper statistics

Abstract Views