Partners in crime? Corruption as a criminal network

How does the structure of an organization affect corruption? This paper analyzes a model that views organizations as networks on which coalitions of corrupt accomplices may form. This network approach to corruption provides new insights into the problem: (i) corruption will arise in enclaves, i.e. coalitions that minimize joint exposure to witnesses, (ii) making the organization more connected may increase corruption, and (iii) corruption will involve larger coalitions under better monitoring. Simulation results also suggest that more hierarchical organizations are more corrupt than flatter organizations. I test these predictions in the lab. Results confirm the predictions and reveal a systematic deviation that has implications for why better monitoring reduces corruption: participants disproportionately fail to realize larger coalitions, which are more necessary under good monitoring. Results suggest it would be sensible to redesign public agencies to puncture the isolation of enclaves.