Consequences of an Anti‐Corruption Experiment for Local Government Performance in Brazil

Decentralization reforms, implemented to improve efficiency and service provision, pose a challenge for federal governments that would like to ensure that federal resources are used appropriately by local governments. To overcome this challenge, some federal governments have implemented costly oversight programs aimed at improving municipal governance. For instance, in 2003, the Brazilian federal government introduced a randomized auditing program with the goal of improving municipal performance by exposing episodes of corruption and mismanagement. Yet, we know little about whether these types of programs actually lead to improvements in local outcomes, especially in terms of service delivery. We argue that audits provide opportunities for learning that should improve performance outcomes. To test this argument, we examine municipal performance in over 5,000 Brazilian municipalities from 2001 to 2012. We utilize the random assignment of audits and estimate difference‐in‐differences regressions. We find that audited municipalities experience greater improvements in performance overall compared to unaudited municipalities, though the effect size is modest. We also find evidence that the auditing program indirectly improves municipal performance. These results indicate that top‐down oversight programs, such as the Brazilian one, are useful not only for improving transparency and accountability, but also for the provision of public services as well.