Cut the Food, Kill the Count

Introduction

On July 4, 2025, Congress signed the largest food assistance cut in American history: $186 billion stripped from SNAP over 10 years, funded into the same bill as a $52-54 billion increase in commodity farm subsidies. Two months later, USDA Secretary Brooke Rollins cancelled the 30-year Household Food Security Report, the only annual government measurement of American hunger, and her department escorted the researchers who produced it out of the building. Archer-Daniels-Midland, whose Q1 2025 lobbying team included former staff of both the House and Senate Agriculture Committees that wrote the bill, spent $1.98 million lobbying Congress in those three months alone, a 71% jump from the prior quarter.

The Sequence Is the Story

Most coverage treated these as separate stories: SNAP cuts in July, the hunger report cancellation in September, researchers escorted out two days later, the SNAP funding freeze in November. Lined up as a single fiscal sequence, the move reads cleanly: cut the program, redirect the savings to the lobbyists who wrote the bill, delete the only independent count of who got hurt.

Start with the cut itself. The One Big Beautiful Bill Act expanded SNAP work requirements from age 54 to age 64 and eliminated the bipartisan 2023 exemptions that had protected veterans, people experiencing homelessness, and adults aging out of foster care from those rules. CBO estimated about 270,000 people in those three categories lose eligibility nationally. Senator Tammy Duckworth, an Army veteran who lost both legs in Iraq, tried to amend the bill to restore the protections. Senate Republicans blocked it.

The actual impact arrived faster than the projections. CBO's August 2025 analysis projected 2.4 million fewer Americans on SNAP in an average month. USDA's own preliminary data, confirmed by AP fact check in May 2026, shows enrollment dropped from 42.83 million in January 2025 to 38.55 million in January 2026 — a fall of 4.3 million recipients, and the largest enrollment decline in the program's 60-year history. Part of that traces to people losing eligibility outright, and part to administrative churn as the new requirements rolled out.

What USDA Did to the Measurement

The annual Household Food Security Report ran continuously from 1995 to 2024 on a consistent methodology: a 74% response rate and an 18-question instrument administered through the Current Population Survey to roughly 45,000 households a year. It was the only nationally representative, longitudinal dataset on American hunger that academics, advocates, members of Congress, and USDA itself all used as the canonical reference. The lineage is bipartisan: it originated from Reagan's 1984 Task Force on Food Assistance, was authorized by a 1990 law signed by George H.W. Bush, and ran its first full survey under Clinton.

On Saturday, September 20, 2025, USDA cancelled it. The press release called the report "redundant, costly and politicized" and added that "extraneous studies do nothing more than fear monger." Two days later, on Monday, September 22, about a dozen Economic Research Service economists and administrators were escorted from their offices, placed on paid administrative leave for what USDA described as "unauthorized disclosure," and stripped of their laptops. The American Federation of Government Employees, which represents some of them, confirmed the action and said the researchers' work product had been removed from agency servers.

The "redundant" framing is the one piece of the story USDA gave a substantive defense of, so it's worth taking seriously. Census Bureau's Household Pulse Survey does include food-sufficiency questions, and USDA itself maintains administrative SNAP data. CSIS's analysis of the cancellation walks through why neither substitutes: Census runs at a 1-10% response rate against the ERS survey's 74%, uses one screening question instead of the ERS 18-item instrument, and breaks longitudinal continuity because its methodology has changed repeatedly since 2020. Administrative SNAP data measures program participation rather than hunger, and by definition cannot capture people who left the program.

The final 2024 report came out on December 30, 2025, three months delayed by the shutdown. 13.7% of American households were food insecure and 5.4% had very low food security. Those are the last numbers from this series, and the next edition would have measured what the OBBBA did to those percentages.

The Money on the Other Side

The same legislation that pulled $186 billion out of SNAP put $65.6 billion back into agricultural spending, including $52-54 billion in expanded ARC and PLC commodity subsidies, the price-support programs for corn, soybeans, wheat, cotton, and rice. The PLC reference price hike alone scored at $50.5 billion over 10 years per CBO Publication 61570. The structure of those payments is the part that doesn't make it into most coverage. According to the EWG Farm Subsidy Database, the top 10% of commodity subsidy recipients collected 65% of all payments in 2024, while the top 1% alone collected 23%, and the bottom 80% averaged $1,180 per recipient. The OBBBA increases flow through that same proportional structure, which means small fruit and vegetable growers and most regenerative farms see nothing from the bump.

ADM's Q1 2025 lobbying disclosure, filed with Congress and aggregated by OpenSecrets, shows $1.98 million spent in the three months when the OBBBA was being drafted, a 71% increase over the prior quarter. The lobbying team includes former staff of both the House Agriculture Committee and the Senate Agriculture Committee, the two committees that wrote the bill's commodity provisions. ADM's full-year 2025 federal lobbying spend is on track to exceed $7 million, and the Q1 spike tracks the markup window for the bill almost exactly.

Whether the OBBBA's commodity provisions are good agricultural policy is a debate. Whether they raised reference prices for the dominant crops by 10-21% (per CBO scoring, they did) is not, and whether ADM benefits from higher reference prices for the crops it processes is also not in dispute, given the company is the largest buyer of US corn and soybeans and a primary trader for both. The committee staff who drafted the language are now on ADM's payroll, lobbying on the same provisions they wrote.

Defying the Court Order on the Way Out

The pattern continued past the bill itself. On November 1, 2025, during the federal shutdown, USDA froze SNAP disbursements for the first time in the program's 60-year history, warning states that paid full benefits anyway that USDA would claw the money back. Federal Judges John McConnell (D.R.I.) and Indira Talwani (D. Mass.) ordered restoration by November 7. USDA complied.

Five weeks later, on December 2, 2025, Rollins escalated again. She used a Cabinet meeting to threaten to cut off SNAP funding to 21 Democratic-led states that had refused to share immigration data on SNAP recipients. Federal Judge Maxine Chesney (N.D. Cal.) had already issued a preliminary injunction in October 2025 blocking USDA from doing exactly that. Rollins issued the threat anyway, with the injunction sitting in place. Her X post the same day read: "29 states stepped up. 21 blue states refused — and two SUED US FOR ASKING!"

The Campaign Legal Center filed a formal ethics complaint against Rollins on April 23, 2026, citing four documented appearances at the America First Policy Institute (her former employer) and the Texas Public Policy Foundation in 2025, despite an ethics agreement prohibiting personal appearances at former employers for one year. AFPI's website at one of those appearances referred to her as a member of "AFPI's cabinet." She also received a $300,000 bonus from AFPI in 2024, confirmed against AFPI's own 990, shortly after her nomination was reported. USDA disputes the violations and the complaint is unresolved.

Who Benefits, and How the Money Flows

The beneficiaries are all visible on the public record.

The cash goes to large commodity agribusiness. The $52-54 billion in expanded ARC/PLC payments flows primarily to the operations that already collect the bulk of farm subsidies, the same top 10% that took 65% of 2024 payments. ADM and the major commodity traders benefit upstream, because higher reference prices for corn, soybeans, and wheat support stable input prices for the products those traders move. Federal taxpayer money came out of SNAP and went through commodity programs to the largest agricultural operations and their downstream buyers, with the lobbying spend that secured it visible in quarterly disclosures.

Politically, the Trump administration gets cover. SNAP enrollment fell by 4.3 million people, and the hunger report that would have measured the human cost of that fall was cancelled before its first post-OBBBA edition could publish. CSIS's analysis says it plainly: there is no comparable replacement, and the timing of the cancellation, two months after the cut, sits inside the window when the first measurement of harm would have started showing up. If the report was genuinely redundant, the agency's response of escorting researchers out and confiscating laptops on September 22 is harder to explain than if the report was producing data the cut couldn't survive.

For Rollins personally, the payoff is access she signed an ethics agreement to forfeit. The $300K AFPI bonus, the four post-confirmation appearances, and a sitting ethics complaint don't prove the federal-policy positions she took were paid for. They do prove that the personal financial relationships with the outside organizations she came from are still operating while her department executes the policy agenda those organizations spent years writing.

What Policy Accountability Looks Like Without the Measurement

The empirical question this leaves open is the one nobody on the federal payroll will be answering after 2024. The Household Food Security Report was the instrument that let independent researchers, members of Congress, and journalists measure whether SNAP changes increased or decreased hunger over time. Without it, the available substitutes (administrative SNAP rolls, the Census Pulse Survey, whatever academic surveys universities can fund without federal cooperation) cannot produce a defensible national figure. The next public estimate of how many Americans went hungry in 2026 will arrive, if at all, two to three years late and from sources that USDA can credibly describe as advocacy.

The 30-year report was cancelled because the next edition would have measured a transfer this administration prefers not to be measurable. Federal taxpayers funded both ends of it: the $186 billion pulled out of SNAP, and the $52-54 billion routed to commodity operators with revolving-door lobbying teams. The instrument that would have put those two figures on opposite sides of the same ledger is the one USDA escorted out of the building.

The forward question is whether any of the academic, philanthropic, or state-level efforts now scrambling to fill the data vacuum can produce something defensible enough to land in a CBO score, a CRS report, or a federal court ruling. If something defensible emerges, accountability survives the cancellation. Without it, the policy that produced the harm will be measured against a baseline that no longer exists, and the answer to "how many Americans went hungry after the cut?" will be whatever USDA decides to say it was.