I Won’t Get Caught: An Agent-Based Model of Corruption with Incomplete Information

Corruption remains one of the biggest challenges facing Indonesia. Rampant bribery leads to the misallocation of resources, unfairly privileging those in power and hindering economic development. Existing studies tend to focus on top-down strategies involving prevention and law enforcement in the larger context of building national integrity. In a budding democracy such as Indonesia, however, the creation of an anti-corruption agency will not be sufficient to combat persistent official malpractice. In this chapter, I propose an agent-based model that combines optimizing behavior with imperfect information to highlight the decentralized mechanisms implicit in the culture of bribery. The model’s bottom-up approach demonstrates how choices at the microlevel give rise to the macrolevel dynamics. Numerical simulations show that aggregate corruption levels diverge from a steady state into permanent oscillations when arrest probability crosses a critical threshold. Consistent with the economic theory of crime, a negative correlation exists between the chance of arrest and the occurrence of embezzlement in situations where the risk of getting caught is minimal. Thus, a higher likelihood of detection lowers aggregate corruption. However, when the chance of arrest exceeds a critical point, the corrupt population alternates between the state where everyone denounces bribery and the state where everyone participates in corruption. The policy implications of the nonlinear relationship between arrest probability and crime are discussed.