The Agriculture Department published a proposed rule on Nov. 10 that would end SNAP for 6.3 million people by fiscal 2029, and the notice is franker about where those people came from than the coverage of it has been.
- The rule, docket FNA-2026-0661, would limit automatic SNAP eligibility to households that receive TANF “assistance” as defined at 45 CFR 260.31, ending it for households that qualify through TANF-funded services that are not cash.
- USDA’s own analysis puts the loss at 17.6 percent of participants, 3.2 million households and 6.3 million people, in fiscal 2029, the first year of full effect.
- Federal benefit spending falls by about $44.649 billion across fiscal 2027 to 2031.
- Comments are due 30 days after publication, which by our count is Dec. 10.
- A 2019 version of the same idea was withdrawn by the Biden administration on June 10, 2021.
To see why the notice reads the way it does, start in 2008. SNAP participation had dropped after the 1996 welfare law, and USDA wanted it back up. Its tool was broad-based categorical eligibility, which lets a state treat a household as eligible for SNAP if it receives a benefit from a TANF-funded program. USDA’s September 2009 guidance told states that benefit could be as small as, in the notice’s quotation of that memo, “an informational pamphlet or an 800-number to call to locate TANF-funded services.” The number of states using the option went from 18 in fiscal 2008 to 43 in fiscal 2011, and the notice itself concedes how that happened:
“…was not organic but was encouraged formally by USDA under the Obama administration explicitly ‘to increase SNAP participation.'”
USDA’s proposed rule, Federal Register, Nov. 10
The sentence reads as a jab at the previous administration, and it makes a different point. A policy built to increase participation did increase it, and by USDA’s own accounting some of the households it brought in sit above the federal lines. The notice calls the result a loophole, and a loophole is ordinarily something nobody meant to build. USDA put this one in writing in 2009 and told states how to use it.
Now the part that has not been in the coverage. For fiscal 2024 the notice counts 5.4 million participants in households over the federal asset limit and 2.0 million over an income limit, and its footnote says anyone who fails both tests is counted only in the income group. By our arithmetic that makes almost three in four of the people above a federal line (5.4 of 7.4 million) people whose problem is what they have saved, not what they earn. The notice does not say that sentence, and it does not need to, because the rule’s argument is that the line should bind. The smallest case shows the scale of the thing being called a loophole: more than 840,000 one- and two-person households got the minimum benefit in 2024 only through this option, 94 percent of them would otherwise have been ineligible, and the total came to “nearly $214 million.” Spread across all 840,000 households that is about $21 a month, and across the 94 percent it is about $23.
The School Lunch Figure Is Smaller Than Its Headline
The headline version, repeated by CNBC among others, is that 1.8 million children would lose eligibility for free school lunch. USDA’s number is real, but what it measures is children who lose automatic eligibility, not children who lose lunch. The notice splits the 1.78 million into three groups: about 1.66 million would still get free meals by applying another way, 93,000 would move to reduced price, and 28,000 would pay full price. Those add to 1.781 million, and the share who keep free or reduced-price meals works out to 98 percent by our arithmetic (1.753 million of 1.781 million), against the “approximately 97 percent” the notice prints. What those families lose is a paperwork shortcut, and school districts would process about 197,000 more applications, an estimated 30,000 additional hours of work, to replace it. The notice also estimates that about 42,000 children could lose Summer EBT beginning in the summer of 2029. We would rather get that right than inflate it, because the case against this rule does not need the school lunch number.
Who Pays and Who Saves
The notice has a line about money that rarely makes the article. States would spend $928.5 million more over five years administering the change while sending about $6.172 billion less in SNAP transfers, a net saving of $5.243 billion, because the 2025 reconciliation law (P.L. 119-21, section 10105) ties state spending to benefit costs. State budgets come out ahead on paper, and that matters when the officials who would normally object to a federal cut are the ones whose books it improves. The extra $928.5 million lands on top of the shift we covered when 75 percent of SNAP’s administrative costs moved to the states. It also arrives after more than 5.5 million people left SNAP in the months since the July 2025 tax law, so this is another cut, not the first.
We think the framing is the tell. A rule that ends benefits for 6.3 million people and takes $44.6 billion out of the program, introduced as a loophole closing, is a cut with a vocabulary. USDA is entitled to argue that the federal limits should bind, and that is a real argument. It is not entitled to present the option as an accident when its own notice says the department built it to bring households in. The place to have the argument is the 5.4 million people in the asset group, because a household that is over the line on savings alone is not what most readers picture when they hear the word loophole. Comments are open on the docket until Dec. 10 by our count, and a final rule has to answer the substantial ones. Read the final rule’s answer to the asset question. If it has none, the loophole language was doing the work the analysis could not.