Treasury Expands Data Analytics to Combat Fraud, Improper Payments

Why It Matters

A unified, data‑driven approach can dramatically reduce waste and protect taxpayer dollars across federal programs, strengthening fiscal stewardship and public trust.

Key Takeaways

• Treasury's Do Not Pay program flags fraudulent payments before disbursement
• Shared data services enable cross‑agency fraud detection
• Machine‑learning models analyze eligibility, identity, and payment history
• Death Master File updates daily to catch payments to deceased
• Data gaps remain for fraud types lacking existing information sources

Pulse Analysis

In recent months the U.S. Treasury has moved from a siloed approach to a centralized, government‑wide fraud‑prevention model. Executive orders issued in 2025 and 2026 mandated a national strategy to curb waste, fraud, and abuse across federal benefit programs, pushing agencies to share data and standardize payment oversight. By positioning the Treasury’s Fiscal Service as the steward of federal funds, the administration hopes to create a unified front that can spot irregularities before money leaves the Treasury’s accounts. The centralized service also reduces duplication of effort, freeing resources for higher‑value analysis.

The centerpiece of the new effort is the Do Not Pay program, which layers eligibility checks, identity verification, and pre‑disbursement reviews using advanced analytics and machine learning. Agencies feed transaction data into a shared platform where algorithms flag anomalies in real time, allowing auditors to intervene before funds are released. Early pilots have shown a measurable drop in improper payments, and the Treasury reports that the system has already prevented millions of dollars in erroneous disbursements. Moreover, the platform integrates with the Social Security Administration’s Death Master File, providing daily updates that automatically block payments to deceased recipients.

Despite the progress, significant data gaps limit the system’s reach. Certain fraud typologies, such as payments to undisclosed beneficiaries or sophisticated identity theft schemes, lack reliable source data, forcing analysts to rely on proxy indicators. The Treasury is actively negotiating access to additional registries, including updated death records and private sector risk databases, to enrich its models. If these enhancements succeed, the government could cut the $400 billion annual improper‑payment estimate by a substantial margin, delivering savings directly to taxpayers. Long‑term, the Treasury aims to embed these analytics into the procurement cycle, ensuring that new contracts are vetted for fraud risk before award.