In the first six months of 2026, SNAP households living under new state purchase restrictions bought about 12 percent less soda than comparable recipients elsewhere, according to a National Bureau of Economic Research working paper that STAT reported on Monday. That is roughly 34 fewer twelve-ounce cans per person over the course of a year. On its face, the finding is exactly what the policy's architects wanted: a restriction that changes nothing about what people buy is a restriction with no health case at all.
The same paper, though, contains a second number that deserves at least as much attention. Among households that cut back on soda, up to 39 percent of the money they would have spent on it moved to other sugary drinks and juices that the bans do not cover. The restrictions changed which product people bought. What they did not do, by the study's own accounting, is deliver anything like a proportional change in sugar.
The policy measured one product, not the nutrient
The working paper, co-authored by University of Chicago Booth School economist Matt Notowidigdo and colleagues, draws on grocery purchase data covering 15,000 SNAP households in the first half of 2026, of which 3,291 live in the ten states where restrictions were in effect. The research was funded by a Bloomberg Philanthropies grant. The design compares purchases in restricted states before and after the rules took hold against households in states without them, which makes the 12 percent figure credible as a measure of soda, specifically.
The substitution finding is where the interpretation lives. Households did not simply spend less. A large share of the redirected money went to sweetened drinks outside the ban's category definitions. Notowidigdo's own conclusion, as STAT reported, is direct: "you want the ban to be more comprehensive, not less." That is the paper's caution against its own headline, and it mirrors how these restrictions were designed. They name products, and the product list is not the same thing as the nutrient.
The nutrient is worth taking seriously. A 2025 study in Nature Medicine led by Laura Lara-Castor of Tufts University estimated that sugar-sweetened beverages account for about 2.2 million new cases of type 2 diabetes and 1.2 million new cases of cardiovascular disease worldwide each year. SNAP serves roughly 42 million low-income Americans, and about 11 percent of SNAP dollars nationally go to sugary beverages and candy. There is a real target here.
The health math is real but small
The paper's authors estimate that the observed reduction, if it persisted, would lower type 2 diabetes risk among recipients by about 2.6 percent over ten years, on the order of 34,000 fewer new U.S. cases, and save roughly $1 billion a year in health care spending. Notowidigdo described the savings as small relative to the $5.3 trillion the United States spends on health care annually, while allowing that a billion dollars a year is still a billion dollars.
Neither side of the policy fight should overclaim these numbers. For supporters, 34,000 avoided cases over a decade is real and meaningful, but it is a small corner of a national burden measured in millions. For skeptics, the estimate assumes the restrictions survive legal challenge, remain in force, and are not further eroded by the substitution the paper itself documents. The most defensible reading of the study is that a soda restriction produces a modest, genuine reduction in the sale of one product and a considerably smaller reduction in the substance that actually drives the health outcomes.
That distinction would matter less if the policy debate were actually about sugar. It mostly is not. The debate is about whether a program that buys food for 42 million people should also be an instrument for changing how they eat, and the product-level design is precisely what keeps that argument alive.
The cost that does not appear in the sales data
There is a second result in the paper that is harder to put a dollar figure on: in a survey of recipients, people living under the restrictions were more likely to report feeling judged or disrespected. The finding is qualitative, but it is not decorative. A benefit restriction is visible in a way a price change is not. It attaches a condition to assistance, and it changes the experience of the program even for recipients who never bought soda in the first place.
Benjamin Chrisinger, an assistant professor of community health at Tufts, told STAT that the study's households may not represent all SNAP recipients and called the stigma finding concerning. Robert Paarlberg, professor emeritus at Wellesley, was blunter. Noting that the measured cut does not add up to a large national health gain, he questioned whether the health gains justify the social cost. His preferred instrument is a tax on sugary drinks, and he points to Philadelphia, where a beverage tax cut consumption by 31 percent across the whole population, not just among benefit recipients, without the same stigmatizing edge. He argues the tax also avoided stigma in part because the mayor framed it around revenue rather than around the diets of the poor.
The contrast matters more than either man's policy preference. A tax raises the price for everyone and lets people adjust quietly, on their own terms. A benefit restriction singles out one population and makes the adjustment public. Both are defensible positions. They are not the same experience, and only one of them shows up in a purchase dataset.
The states are running their own experiments
The policy itself is a moving patchwork, which the study's averages partly hide. The waivers began in January, with Indiana and Utah, followed by Idaho and Oklahoma in February, Louisiana later that month, Texas in April, Florida later in April, and Arkansas on July 1. South Carolina and North Dakota were scheduled for the end of August, with Montana, Ohio, and Virginia in October, and further states staggered into 2028. The rules differ from state to state: some cover soda and candy, others add energy drinks, and the boundary lines are drawn differently everywhere. The NBER paper's ten states is a snapshot of a program whose membership changes by the month.
The administration has treated the June ruling as a setback, not an ending. Agriculture Secretary Brooke Rollins called Judge Jackson an activist judge and pledged to keep fighting, and Health and Human Services Secretary Robert F. Kennedy Jr. promised the same. Whether the waivers survive will depend on the appeals, and whether the health effects survive will depend on which states remain in the experiment long enough to produce data. The study everyone is citing describes a policy that may not exist in its current form by the time the next study arrives.
The legal floor under the whole experiment
The experiment also rests on a legal foundation that is visibly cracking. On June 22, Judge Amy Berman Jackson of the U.S. District Court for the District of Columbia blocked the restrictions in five states, finding that the USDA's pilot-project authority allows testing of program efficiency, not reshaping what counts as food under the statute. The ruling suspended the bans in Colorado, Iowa, Nebraska, Tennessee, and West Virginia. The other 18 states with approved waivers are proceeding anyway, with Arkansas enforcing its ban from July 1 and additional states scheduled through 2028. The administration has pledged to keep fighting, and the case may be appealed.
This analysis takes no position on whether states should restrict SNAP purchases. The strongest case for the policy is straightforward: taxpayer dollars should not subsidize products linked to obesity and chronic disease, and a program built around nutrition should serve nutrition. The strongest case against it is equally straightforward: the restrictions are paternalistic, they burden the people least able to absorb the change, and Congress never authorized the USDA to run the experiment. Both cases can be stated in good faith, and the courts, not the sales data, will settle the authorization question.
What the legal fight exposes is that the product-by-product design is also the policy's vulnerability. A ban that must carve exemptions for every adjacent sweetened drink is a ban on a category with no natural boundary, written onto a statute that Congress did not write.
The choice is between regulating a product and regulating sugar
Read carefully, the NBER paper is less a verdict on the SNAP experiment than a demonstration of the difference between banning a product and taxing a nutrient. The ban reduced the product. A substantial share of the spending simply moved to the next shelf over. The tax, in Philadelphia's experience, reduced the nutrient across the entire population, including households far above the poverty line, and raised revenue while doing it.
The global evidence points the same direction. The authors of the Nature Medicine analysis recommended taxes among the interventions worth trying, and Mexico's 2014 tax on sugary drinks, the best-studied example, reduced consumption especially among lower-income households, the same group the SNAP restrictions target. A tax also generates revenue that can fund health programs for the population it burdens, while a purchase restriction generates administrative cost and a legal fight. The instrument debate has a literature, and it does not favor product bans.
A policy aimed at sugar should be evaluated by sugar. On that measure, the first evidence from the state experiments says the effect is real but far smaller than the soda headline suggests. The states now running these restrictions will spend the next year generating more data, and both sides will cite it selectively. The honest reading of what exists so far is that the bans changed what people bought, not what they consumed. The fight over whether that counts is not really about soda at all. It is about whether food assistance should be a tool for behavior change, and the numbers, so far, say the tool works mainly on the label.
Primary sources
- STAT's reporting by Sarah Todd for the study's headline findings, the substitution estimate, the health impact numbers, and the comments from Matt Notowidigdo, Benjamin Chrisinger, and Robert Paarlberg.
- The Nature Medicine analysis of the global burden of sugar-sweetened beverages by Laura Lara-Castor and colleagues at Tufts University.
- USA Today and CNN's coverage of the state waiver status and Judge Amy Berman Jackson's June 22 ruling.