Case Study: Maryland achieves major reduction in tax-refund fraud and faster refunds with Teradata

Targeting Tax Fraud with Advanced Analytics

Maryland’s Comptroller’s Office, through its Questionable Return Detection Team (QRDT), was struggling with a surge in identity‑theft tax refunds and an inefficient, rule‑based process that suspended about 110,000 returns a year but identified only roughly 10% as fraudulent—delaying honest taxpayers’ refunds and tying up resources. To improve detection, the state partnered with Teradata and ASR Analytics to apply a data warehouse and advanced analytics to its tax data.

The team built models that combine many metrics, score and group suspicious returns, and prioritize investigations, which reduced suspensions to an expected 40,000–50,000 and raised detection accuracy to about 65% versus the prior 10%. The new approach sped refunds to legitimate taxpayers, recovered roughly $10 million in the 2015 filing season (compared with about $3 million under the old system at the same point), and enabled the QRDT to handle more complex identity‑theft workflows and multi‑year W‑2 comparisons.

Maryland

Andy Schaufele

Director of the Bureau of Revenue Estimates