Reputation Effects in Repeated Audits, with Application to Insurance Fraud Deterrence

In principal-agent problems, the repetition of interactions in a dynamic setting may alter the equilibrium outcomes. In insurance fraud, the frequency of auditor-auditee interactions is higher when there is collusion between policyholders and service providers (e.g., car repairers, health care providers...). The same service provider usually handles claims filed by many policyholders affiliated to the same insurer, and thus the insurer-service provider interactions are repeated with reputation effects. We analyze this issue in a repeated game where the insurer may potentially face a dishonest service provider who colludes with policyholders. The insurer has beliefs about the type (honest or dishonest) of the service provider and she may verify the truthfulness of the claim through costly audits. The reputation of the service provider corresponds to these beliefs and changes over time, and misbehaving deteriorates this reputation. In the end, it may lead to a breach of contract and thus represents a threat that may deter from defrauding. We show that, at early periods, the insurer audits agents who would not be monitored in a static setting because their reputation is good enough. Corresponding dishonest agents who slipped under the radar and have an initially good reputation do not defraud systematically at early periods, as opposed to the instantaneous game. In addition, auditing efforts for medium reputations are lower as dishonest agents want to preserve the possibility of defrauding later. Both aspects corresponds to a reputation-based deterrence mechanism, where the fear of deteriorating one's reputation acts as a discipline device for dishonest service providers.