94 Pages Posted: 14 Mar 2017
Date Written: March 8, 2017
Is corruption harmful to firms? We study this question by estimating the private sector impact of a large government program aimed at fighting corruption and inefficiency in Brazilian local bureaucracies. Exploiting a special feature of the auditors’ instruction manual, we are able to manually construct a novel dataset on 34,000 corruption cases involving firms. This dataset is then linked to matched employer-employee data, to the universe of federal procurement contracts and government loans, to the manufacturing Census, as well as to public prosecutions and electoral data. Our identification strategy mainly relies on the randomization of the audits to estimate dynamic difference-in-difference models. First, we show that the audits lead to a partial reallocation of economic activity away from incumbent business establishments towards new entrants. Our evidence suggests that this is primarily driven by a reduction in private sector corruption leading to higher entry in the local economy. Second, we uncover striking heterogeneity at the firm level. Firms that are directly affected by government corruption strongly benefit from the audits, as reflected by lower exit rates and higher employment in the short and the long-term. Third, we test several theories of corruption by investigating the mechanisms behind the positive effects of the anti-corruption program on firms. We find evidence for three of them: (i) access to finance; (ii) composition of government contracts; and (ii) within-firm restructuring. We conclude by discussing efficiency considerations and policy implications.
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: Suggested Citation