Fraud-prevention agencies typically have far more potential cases than they have resources to investigate.  How can they effectively target their resources?  What are the benefits of intensive investigations compared to less intensive investigations?

This paper reports on a field experiment conducted by a state/local government agency with oversight authority in Medicaid.  One of the agency’s tasks is to identify Medicaid recipients whose incomes far exceed eligibility thresholds.  The agency can remove these violators from the Medicaid rolls, recover costs, and refer cases for prosecution.  The agency’s systems flag potential violators, but it can currently investigate fewer than 10% of flagged cases each year.

The agency is developing more refined targeting criteria to prioritize investigations.  In this study, the agency randomly chooses a sample of 300 “high-probability” and “moderate-probability” targets to investigate.  It also randomly assigns targets to either a “high-intensity” or a “low-intensity” investigation.  This paper reports on the effectiveness of the targeting criteria and investigation strategies at identifying actual Medicaid violators and at recovering costs.