The U.S. Department of Agriculture (USDA) reported that the Supplemental Nutrition Assistance Program (SNAP) had a 10.6% payment error rate in fiscal year 2025, nearly double the 6% threshold set by the 2025 One Big Beautiful Bill Act (OBBBA). The error rate amounts to $10 billion in improper payments, according to the USDA. Meanwhile, Tennessee faces potential financial penalties if its 9.44% error rate does not improve, which could cost the state $162 million to $171 million in federal funding.
Payment Errors vs. Fraud
The USDA clarifies that payment errors—over- or under-payments—are not necessarily fraudulent. Experts say most errors stem from administrative mistakes or eligibility miscalculations. Fraud, such as trafficking benefits for cash, is a separate issue. A 2024 Government Accountability Office (GAO) report found that payment errors are largely unintentional, while fraud involves deliberate deception.
State-Level Consequences
Under OBBBA, states with error rates exceeding 6% could face penalties starting in October 2027. Tennessee’s error rate of 9.44%—comprising 7.84% overpayments and 1.60% underpayments—puts it at risk. The USDA emphasizes that error rates do not indicate fraud, but Agriculture Secretary Brooke L. Rollins argues they reflect a lack of state accountability.
Advocates Push Back
Anti-hunger groups argue that framing SNAP errors as fraud or waste is misleading. They contend that payment errors are not equivalent to fraud and that penalties could disrupt food assistance for millions of low-income families. The USDA’s metric does not capture common fraud types, such as EBT card skimming, according to the American Public Human Services Association (APHSA).
Long-Term Implications
If states like Tennessee fail to reduce error rates, they may need to cover a portion of SNAP costs previously funded by the federal government. The USDA has taken steps to improve program oversight, but critics warn that penalties could reduce benefits for vulnerable populations without addressing root causes of errors.