Large error estimates without full fraud-risk maps

GAO’s 2026 comparison reported fiscal 2025 improper-payment estimates of $608 million for Veterans Affairs community care and $23.7 billion for Medicare Advantage. Together they equal $24.31 billion.

The error rates differ sharply—2.4% for VA community care and 6.1% for Medicare Advantage—but GAO’s central governance finding applies to both: neither program had a comprehensive fraud-risk assessment.

Fiscal 2025 improper-payment estimates
ProgramEstimateRate
VA community care$608M2.4%
Medicare Advantage$23.7B6.1%
Combined$24.31BNot comparable as one rate
Sources: GAO-26-107946: Program Integrity

Improper does not automatically mean fraudulent

Federal improper-payment estimates include payments made in the wrong amount, to an ineligible recipient, or without enough documentation. They are not a finding that the entire amount was stolen. GAO separately examined whether agencies followed its fraud-risk framework.

For Medicare Advantage, GAO also pointed to delays in Risk Adjustment Data Validation audits and the lack of a detailed plan for clearing the backlog. For VA, GAO recommended a comprehensive assessment; VA concurred. CMS neither agreed nor disagreed with the new actions, according to GAO.

The oversight gap is measurable

The original contribution is to place the two programs on the same control matrix without collapsing their different payment systems. Both reported substantial errors, and both lacked the full assessment needed to identify schemes, rank risk and document responses across the program.

A fraud-risk assessment would not eliminate every improper payment. It would create a documented, repeatable basis for deciding which vulnerabilities deserve controls and resources. The absence is a management gap, not proof that managers knowingly permitted fraud.

Method and limits

This newsroom used GAO’s printed fiscal 2025 estimates, added the two dollar amounts and reviewed the status of the three recommendations on the official page. The rates were not averaged because the programs have different payment bases.

GAO’s improper-payment figures are statistical estimates and may be revised. They should not be presented as adjudicated losses, criminal fraud or cash that agencies can automatically recover.

Sources and further reading

GAO-26-107946 and recommendation status