Under which conditions do anti-corruption policies reduce bureaucratic corruption? Previous studies find that anti-corruption audits are effective in disciplining politicians, but their impact on bureaucrats is unclear. We leverage 10 years of randomized audits and the careers of 275 thousand Brazilian municipal officials. We find that audits trigger a multi-pronged response, with increased risks of dismissal and departure, slower career progression, and a spike in hiring. Non-tenured, less educated bureaucrats are more at risk of dismissal, consistent with a form of scapegoating in which the program does not necessarily punish the most corrupt bureaucrats, but instead the ones who are easiest to punish. Finally, the response hinges on electoral accountability, with mayors in their first term and mayors audited in the first year of their term displaying stronger responses.
About the speaker:
Romain Ferrali is an Assistant Professor of Economics at the Aix-Marseille School of Economics. He studies how social networks influence political and economic outcomes. I am particularly interested in issues related to development, especially in North and Sub-Saharan Africa. My research uses game theory and a variety of quantitative methods including network analysis, causal inference, and structural estimation. My main project examines the relationship between the network structure of an organization and corruption. I explain why corruption sometimes involves vast conspiracies, and sometimes isolated individuals. I also propose ways of designing organizations that are more resilient to corruption. His research spans across disciplinary boundaries and has been published in a variety of outlets such as the American Journal of Political Science, Comparative Political Studies, Games and Economic Behavior, and the Journal of Public Administration Research and Theory. Aside from his academic work, Romain is also the scientific advisor of Tafra, an NGO based in Rabat, Morocco that advocates for the rule of law through access to data.