Nine states came in under the federal accuracy threshold for food assistance last fiscal year. The other 41 did not, and that arithmetic is about to stop being a bookkeeping curiosity and start being a line in state budgets. A law signed in 2025 converted the payment error rate from a performance statistic into a bill, and the first year of data eligible to calculate that bill is already on the books. What those 41 states do before the current measurement year closes on September 30 determines how large it gets.
What the 6 percent line in Public Law 119-21 actually measures
The payment error rate is not a fraud statistic, which is the single most misread thing about it. It measures how accurately a state agency determines who qualifies for the Supplemental Nutrition Assistance Program and how much they receive, and it counts errors in both directions. A caseworker who issues a household $40 too little damages the rate exactly as much as one who issues $40 too much.
For fiscal year 2025 the national rate landed at 10.62 percent, against a congressional threshold of 6 percent. Overpayments accounted for 9.28 points of that and underpayments for 1.33, and the errors above a $57 tolerance floor summed to roughly $10.1 billion nationwide, according to the Food and Nutrition Administration’s announcement of the figures. Agriculture Secretary Brooke L. Rollins called the numbers “further proof that state accountability is severely lacking in SNAP.”
Sections 10105 and 10106 of the One Big Beautiful Bill Act attached money to that number. Under the benefit provision, a state at or above 6 percent must cover a share of its own SNAP benefit costs on a sliding scale: 5 percent for rates between 6 and 8, 10 percent between 8 and 10, and 15 percent at 10 percent and above. In most cases that obligation begins October 1, 2027, and fiscal 2025 is the first year whose rate can be used to calculate it.
One detail keeps this from being a verdict. States may base their first cost share on either their fiscal 2025 rate or their fiscal 2026 rate, whichever serves them better, so the 41 that missed last year get one more measured attempt before the bill is set. That reprieve is narrow and it expires: from fiscal 2029 onward each year’s share is fixed by the error rate three years earlier, with no choice involved.
Alaska at 23 percent, South Dakota at 2.5 percent
The official state-by-state table shows a spread wide enough to make a national average close to meaningless. Alaska recorded 23.15 percent. New Mexico came in at 16.81, Delaware at 16.00, Georgia at 15.21, Illinois at 14.67 and Oregon at 14.14. The District of Columbia, which is measured alongside the states, posted 18.66.
At the other end, South Dakota finished at 2.47 percent, Idaho at 3.85, Wyoming at 3.96 and Kentucky at 4.70. Those four, plus Iowa, Nebraska, Utah, Vermont and Wisconsin, make up the entire set of states that cleared the threshold. Several states missed it by almost nothing: Nevada at 6.22, West Virginia at 6.69 and Ohio at 6.76 sit close enough that a single good year of case reviews would move them across.
The distinction matters because the penalty tiers are cliffs rather than slopes. A state at 9.99 percent owes 10 percent of its benefit costs. A state at 10.01 percent owes 15. On a large caseload, two hundredths of a percentage point is worth tens of millions of dollars, which gives every agency in the country an intense interest in the exact number it reports.
The worst performers get a stay rather than a pass. States above 13.34 percent receive a temporary exemption from cost sharing, which is why Alaska, New Mexico, Delaware, Georgia, Illinois and Oregon are not in the first wave. That exemption ends in fiscal 2030, at which point they pay on their fiscal 2027 numbers. There is also a pattern in who cleared the bar: all nine states that did have comparatively small caseloads, and none of the largest programs in the country, including California, Texas, New York and Florida, came in under 6 percent.
The administrative match that drops to 25 percent on October 1
Running underneath the benefit penalties is a second, nearer change that has drawn far less attention. Section 10106 cuts the federal government’s share of state SNAP administrative costs from 50 percent to 25 percent beginning in fiscal year 2027, which starts October 1 of this year. States pick up the difference regardless of how accurate their casework is.
The Food and Nutrition Administration has published a proposed rule codifying that reduction, and its own regulatory impact analysis puts the transfer at approximately $16.9 billion over fiscal years 2027 through 2031, an average of $3.4 billion a year moved from federal ledgers to state ones. The department calculates the net change in total spending at zero, because nothing is being saved. The cost is being relocated. Written comments on that rule are due by August 24.
A narrow set of carve-outs survives. Employment and training administrative costs stay at a 50 percent federal match, and tribal organizations administering the program on reservations continue to receive 75 percent. Everything else in the administrative column moves to a three-to-one state burden.
Why states argue an error rate is not a performance grade
State agencies have spent the year making a version of the same argument: that error rates measure the complexity of the rules more than the competence of the staff applying them. Eligibility turns on household composition, fluctuating wages, shelter costs and reporting timelines, and a caseworker processing a high volume of applications under those rules will generate errors that involve no bad actor at all.
The dollar estimates cluster tightly. Federal Funds Information for States, which models grant flows for state budget offices, put the annual shift at $9.4 billion under the fiscal 2025 rates. The Center on Budget and Policy Priorities, which opposes the change, lands near $9 billion for fiscal 2028 and calculates that close to half of states could owe $100 million or more apiece. It also notes that 79 percent of the exposed households include a child, an older adult or a person with a disability.
The counterargument is not frivolous, and it comes with a larger number attached. Angela Rachidi of the American Enterprise Institute calls the cost share a blunt but necessary tool, arguing that a program in which Washington funds 100 percent of benefits gives states no financial reason to police accuracy. She points out that the 10.62 percent rate is far above the 6.6 percent states averaged before the pandemic, and estimates that once the temporary exemptions lapse, states would carry roughly 12 percent of benefit costs, or about $11 billion a year. Both readings survive the evidence: the old incentive was genuinely weak, and the instrument chosen to fix it is genuinely crude.
What a state can actually cut when the bill arrives
The options available to a state facing a nine-figure obligation are limited and mostly unattractive. It can tighten eligibility so fewer residents qualify, which shrinks the benefit base the percentage is applied to. It can cut spending elsewhere in the budget. It can raise revenue. Or it can absorb the hit and carry a structural deficit.
Some have floated a fifth option. A number of state agencies have signaled they might suspend SNAP operations or stop administering the program altogether rather than absorb a cost share, a threat that would strand recipients rather than fix the underlying errors. Others are lobbying for time instead, and members of Congress from both the affected states and the minority have pushed to delay the fiscal 2028 requirement in the coming Farm Bill. Neither route addresses why the errors happen.
None of it is hypothetical. The administrative match changes in roughly two months, the comment window on the rule closes August 24, and the fiscal 2026 measurement year, the one that gives those 41 states their second and final chance at the threshold, ends September 30. More than 40 million Americans draw SNAP in a given month. The states deciding how hard to chase a decimal point are being scored on work they are doing right now.
Free retirement updates: Want plain-English help keeping more of your money in retirement? The free Retirement Shield newsletter covers scams, benefits, and money many retirees may be owed, a couple times a week. Subscribe free.
This article was researched and written with AI assistance and reviewed prior to publication.
Julian Harrow specializes in taxation, IRS rules, and compliance strategy. His work helps readers navigate complex tax codes, deadlines, and reporting requirements while identifying opportunities for efficiency and risk reduction. At The Daily Overview, Julian breaks down tax-related topics with precision and clarity, making a traditionally dense subject easier to understand.
