When the Aid Stops
In 2021, with public support at its peak, American hunger was the lowest in USDA's yearly series, which begins in 2001. In 2023, ending one piece of that support pushed hunger measurably back up. In 2025, the country chose to cut food aid again, this time in permanent law. We already ran this experiment, and the result is on file.
Most policy debates argue about what would happen if you changed something. This one does not have to. On food aid, the country ran the test in plain sight over the last five years: it raised benefits, watched hunger hold near the lowest level in USDA's yearly series through a pandemic, then let part of the increase expire and watched hunger climb back. The second half was measured: when the extra SNAP benefits ended in March 2023 in the 35 states and territories still paying them, food insufficiency among SNAP participants there rose.
That makes the current round of cuts unusual: their likely outcome has already been measured once and written down. (For the basics of how food insecurity works, see the companion primer, How to Think About Food Insecurity.)
This is not a prediction. It is a rerun. We have the tape from the last time the aid stopped.
Hunger held near the lowest level in USDA's yearly series through the middle of a pandemic, which should have pushed it up.
In 2021, the share of U.S. households that were food insecure was 10.2 percent, the lowest in USDA's yearly series, which begins in 2001, and statistically no different from 2019, before the pandemic. The economy was still unsteady, so prosperity does not explain why hunger did not rise. An unusual stack of public support was in place: expanded SNAP, stimulus checks, an enlarged Child Tax Credit, free school meals. By the Center on Budget and Policy Priorities' reading, the SNAP increases and other temporary policies helped keep hunger from rising. Money reached households, and the pandemic did not push more of them into running short of food. For one year, the country saw what generous aid does to hunger, and the answer was that it keeps hunger from rising.
The experiment, in one line
Share of U.S. households that were food insecure, 2020 to 2024. Drawn to scale.
Hunger was at the lowest in USDA's yearly series when aid was highest in 2021, then climbed as pandemic supports were withdrawn and prices rose. Source: USDA Economic Research Service.
When one specific benefit ended, researchers could watch what happened next, cleanly.
During the pandemic, SNAP added temporary emergency allotments, between $95 and more than $250 a month for a household. In March 2023, those allotments ended in the 35 states and territories still paying them. That created a natural experiment: a large group of families lost at least $95 a month in food money at about the same time, while a group of people with incomes low enough to qualify but not on SNAP did not.
A peer-reviewed study in Health Affairs tracked the result with Census survey data. Among SNAP participants, the end of emergency allotments led to an 8.4 percentage-point rise in food insufficiency, not having enough to eat. It also brought more food-pantry use and more trouble paying other bills. An earlier study by University of Pennsylvania researchers, of the 18 states that ended the allotments early, applied its results to the whole country and estimated roughly 2 million more Americans went without enough food once the extra benefit stopped everywhere. A permanent benefit increase from 2021 was already in place and did not prevent it. Cut the support, and hunger rose, within months, by a measurable amount.
The 2023 cut ended a temporary pandemic benefit. The 2025 cut goes after the regular program, and it is permanent.
In July 2025, a federal reconciliation law made what the Center on Budget and Policy Priorities calls the deepest cuts to SNAP in the program's roughly sixty-year history. Analyses of the law describe about $187 billion in federal SNAP funding withdrawn through 2034, about a sixth of the program. For the first time, states can be made to pay a share of the benefit costs themselves. How much depends on how much of its benefit money the state pays out wrong, too much or too little. Score well and it pays nothing. Score badly and it can owe up to 15 percent of its entire food bill. That starts in October 2027. Several states cannot easily absorb it, and some may meet it by tightening eligibility. Work requirements were expanded to include older adults up to 64 and adults whose household's youngest child is 14 or older, and exemptions for veterans, people experiencing homelessness, and young adults who aged out of foster care were removed; the law added one for American Indians. The one route to a real increase was closed: the plan behind the benefit can still be re-priced for inflation each October, but a review of what it should contain can no longer raise what it costs.
The details are complicated; the shape is not. This is the 2023 experiment again, run deliberately and with no end date. The 2023 version ended a temporary pandemic top-up that had run for about three years. The 2025 version cuts into the regular program, by narrowing who qualifies, lowering some households' benefits, and shifting costs to the states, and it is written into permanent law. If ending a temporary top-up produced a measurable jump in hunger, the direction of a permanent cut is not in question. Only the size is.
Supporters of the law argue it reins in federal spending and pushes more recipients toward work, and the program's cost is real. The 2023 evidence does not weigh in on whether those goals have merit. It establishes only what pursuing them by cutting benefits has been shown to cost: a measurable rise in hunger.
This is no longer a forecast. By March 2026, about eight months after the July 2025 law, SNAP enrollment had already fallen by more than four million people, a 10 percent drop and the steepest in decades.
In June 2026, the Senate's draft farm bill did not delay the new state share of benefit costs or restore any of the cuts. On September 16, 2026, the Senate Agriculture Committee passed a version that would push the usual start of that share back one year, to October 2028. For the states with the worst scores, it would also raise the top rate from 15 to 20 percent of their food bill, starting in October 2030. Separately, the July 2025 law raised the state share of the cost of running the program on October 1, 2026. The committee's version would not have delayed that increase. The farm bill the House passed in April 2026 has neither the delay nor the higher rate. As of September 28, 2026, the full Senate had not voted on its version.
A region still recovering from a flood is a bad place to run a hunger experiment.
Western North Carolina enters this with little slack. About 25,000 Buncombe residents rely on SNAP, down from about 29,000 in 2025. In some of the rural counties around it the share was higher in 2025, about 1 in 6 residents in McDowell and close to 1 in 5 in Rutherford. After Helene, the food bank says need is at the highest level in its history. When the same families faced a brief SNAP pause during the federal shutdown in late 2025, the local response was real and could not cover the gap, which is the lesson of a separate piece, Why Food Banks Can't Fix Hunger. A permanent cut does not pause. It stays.
North Carolina has started to price its share. The state's first full budget since 2023 puts money toward carrying out the new federal rules and cutting payment errors. The legislature's fiscal analysts put the state's share of benefit costs at about $150 million for the first year, at the error rate the state posted for the year ending September 2025. At the top rate, NC Health News calculated, it could run to about $420 million a year. The budget hands that bill to the counties, in any year the state owes one, by holding back part of each county's sales tax from October 2027, money the counties share with their towns and special districts. The same budget funds a restart of the Healthy Opportunities Pilot. The pilot has been suspended since July 2025. Before that, it used Medicaid dollars for non-medical help such as food, housing repairs, and transportation. Under its new federal terms, it can pay for transportation only as a ride to another pilot service in a person's care plan. That means help such as car repairs will not come back in its old form. The restart is funded with about $9 million in state money, against the roughly $80 million in state money that had been requested. Federal matching money brings that one-time funding to $25 million. As of early September 2026, the state had not told the organizations that lead the pilot in each region when it would pay out the $25 million. Nor had it told them exactly which services the money would fund. Washington is paying less, and the state has passed the new costs on to the counties.
We have the result. We are choosing to repeat the cause.
The last five years drew the line about as clearly as social policy ever gets to. Aid up, and hunger held at the lowest in USDA's yearly series in 2021. Aid partly withdrawn, hunger measurably up by 2023, with about 2 million more people going without enough to eat. There is no mystery left in the mechanism, and no need to wait and see.
So the 2025 cuts are best understood not as a gamble but as a decision made with the evidence in hand. Hunger is set by how much the public helps cover the gap between wages and the price of food. That makes it a dial the public sets, and it turns both ways. We watched it stay low when we wanted it to. We are about to watch it do the opposite, on purpose, in a region that can least afford the lesson a second time.
Every group in our directory is local, vetted, and doing this work right now. Pick one and back it. →