Computational Simulation of Benefit Fraud and Community Resilience in the Wake of Disaster

Abstract

The monetary assistance provided for disaster relief creates opportunities for fraudulent behavior. Historical records have shown that the loss of recovery funds due to improper and fraudulent payments could reach hundreds of millions of dollars per event, siphoning support away from those who need it the most and potentially slowing down the economic resurgence of a disaster-stricken community. Focusing specifically on benefit fraud, a common type of postdisaster crime, an agent-based computational model based upon criminology theory is proposed to investigate how such behavior affects recovery during the postevent period. The simulation environment models a community facing a natural disaster, the presence of fraudsters and application inspectors, and the interactions between them. Data from the Hurricane Katrina and Rita disasters is used for calibration. The proposed model accounts for both microlevel disaster demands caused by building damage and meso-level social variables. It estimates the cost to communities associated with benefit fraud. Parametric studies quantify how reducing application review errors, decreasing disaster demands, and increasing oversight can help lessen the losses caused by benefit fraud. They demonstrate how computational simulation can be used to achieve a meaningful balance between the loss of fraudulent payments and the speed of distributing aid in order to improve the overall resilience performance of communities.