Insurance Fraud Detection: A Statistically-Validated Network Approach

Fraud is a social phenomenon, and fraudsters oftencollaborate with other fraudsters, taking on differentroles. The challenge for insurance companies is toimplement claim assessment and improve frauddetection accuracy. We developed an investigativesystem based on bipartite networks, highlighting therelationships between subjects and accidents or vehi-cles and accidents. We formalize filtering rules throughprobability models and test specific methods to assessthe existence of communities in extensive networksand propose new alert metrics for suspicious struc-tures. We apply the methodology to a real database—the Italian Antifraud Integrated Archive—and compare the results to out‐of‐sample fraud scams underinvestigation by the judicial authorities