Modeling sanction choices on fraudulent benefit exchanges in public service delivery

Public service delivery programs are not free from players' opportunistic behaviors, such as fraudulent benefit exchanges. The standard methods used to detect such misbehaviors are static, less effective in uncovering interactions between corrupt agents, and easy to evade because of corrupt agents' familiarity with detection procedures. Current fraud detection efforts do not match the dynamics and adaptive processes they are supposed to monitor and regulate. In this paper, an agent-based simulation model is built to gain insight on sanction choices to deter fraudulent activities in public service delivery programs. The simulation outputs demonstrate that sanctions with low certainty must be accompanied by prompt action in order to observe a reduction in fraudulent vendors. However, a similar level of reduction in fraudulent vendors may be achieved once a certain number of fraudulent vendors are sanctioned, even if the public agency's action is relatively delayed. These characteristics of sanctions provide strategic choices that public service delivery program managers can consider based on their priorities and resources.