Playing Whac-a-mole in the Fight Against Corruption: Evidence from Random Audits in Brazil

Abstract: Evidence shows that audits can reduce corruption, but also that corrupt officials are able to substitute to alternate forms of corruption and to adapt over time to the anticorruption system in place—especially when the anticorruption policy imposes controls only on certain types of corruption. This paper studies the possible unintended consequences of such selective anticorruption monitoring, in the context of a large random audits program in Brazil that has proved successful in reducing the misuse of federal funds at the municipal level. By exploiting the random assignment of Brazilian municipalities to the program, we document the existence of unintended (and undesirable) consequences: audited municipalities employ more labor in water provision, and this translates into a more inefficient service, in the sense that providers in audited municipalities use more labor for a given level of output and other inputs. We also provide additional evidence consistent with the idea that local officials may be using their discretion in hiring to substitute between different forms of corruption.