A Purchasing Result, Not a Health Result
Soda purchases fell roughly 12 percent among households receiving food benefits in states that made soda ineligible, according to a new economic analysis, and the finding answers a narrower question than the headline suggests.
The study, circulated as a National Bureau of Economic Research working paper and reported by STAT, drew on grocery purchase data from the first six months of 2026 covering 15,000 households using Supplemental Nutrition Assistance Program benefits. Of those, 3,291 were in 10 states that had implemented new restrictions. The paper has not been peer-reviewed, and it was supported by a grant from Bloomberg Philanthropies, which campaigns for taxes on sugary beverages. Readers should weigh that funding relationship alongside the findings.
What was measured is what people bought, tracked through grocery purchase data, which the authors translate to roughly 34 fewer 12-ounce cans of soda per person per year. What was not measured was what anyone drank, what happened to their weight, or what happened to their blood sugar. Those are different variables requiring different study designs and far longer follow-up.
The distinction is not pedantic. The central economic objection to these policies has always been substitution: that households would simply pay for soda with other money, leaving consumption unchanged. This analysis directly tests that objection and finds it partly right and partly wrong. That is a genuinely useful result, and it concerns shopping baskets.
Substitution Happened, Just Not Completely
The most consequential number in the paper may not be the 12 percent. It is what happened to the money.
SNAP recipients redirected up to 39 percent of the funds they did not spend on soda toward other sugary drinks and fruit juices that state rules did not cover, the authors reported. Roughly two-fifths of the dollars that stopped going to soda went to something comparable.
That pattern was visible earlier in other data. Analysis from the marketing firm Ibotta, described in trade coverage, found that soda purchases among SNAP participants in the first five waiver states fell twice as fast as in states without the change, while powdered beverage mixes gained ground.
Study coauthor Matthew Notowidigdo told STAT that if the goal is reducing sugar intake, "you want the ban to be more comprehensive, not less." That is a design observation rather than an endorsement, and it points to the definitional problem underneath these policies. State waivers vary considerably in what counts. Some exclude drinks based on added sugar content, others on fruit juice content, and some restrict candy or prepared foods as well. Different lines produce different substitution paths.
The policy map is also unsettled. Twenty-three states have received waivers from the Department of Agriculture, but the restrictions are currently suspended in five of them by federal court order, which means the population living under these rules is smaller than the waiver count implies and could change again.
The Projected Health Effect Is a Model, Not a Measurement
The authors did estimate a downstream health effect, and readers should understand exactly what kind of number that is.
They projected that reducing soda consumption would lower the risk of type 2 diabetes by 2.6 percent over 10 years, resulting in roughly 34,000 fewer new cases nationally, with Notowidigdo estimating about $1 billion in annual health system savings. He characterized that against total United States health spending of about $5.3 trillion in 2024, noting it is small in that context while still being a billion dollars. Research published in Nature Medicine has attributed roughly 2 million new type 2 diabetes cases worldwide each year to sugar-sweetened beverages, which is the broader backdrop these projections sit against.
This is a modeled projection built on a measured change in purchasing, layered with assumptions about how purchasing maps to consumption and how consumption maps to disease risk over a decade. Each layer carries uncertainty. Nobody has observed 34,000 cases of avoided diabetes.
Limitations run further. Benjamin Chrisinger, an assistant professor of community health at Tufts University who was not involved in the work, told STAT the purchase data may not represent the broader population of SNAP recipients. Six months is a short window for a behavior change that could either deepen or erode. And the analysis cannot see purchases made outside tracked grocery channels.
A survey conducted as part of the study found something the purchasing data cannot capture: recipients reported feeling more judged or disrespected under the restrictions. Chrisinger called that finding "concerning," adding that stigma is difficult to weigh against other health indicators. Whether stigma affects program participation is an open question with real consequences, since a food benefit only helps people who enroll and use it.
The Evidence Base Is Still Being Built
Not every outside reader came away convinced the trade is worth it. Robert Paarlberg, a professor emeritus at Wellesley focused on food policy, told STAT the study looks sound but that a 12 percent cut in soda purchases, among the roughly 12 percent of Americans who use SNAP, does not add up to a large national health gain. "I wonder if it is worth the stigma," he said, arguing that beverage taxes reach everyone rather than one group and citing Philadelphia's tax, which was associated with a much larger drop in consumption.
Older federal analysis anticipated much of this debate. A USDA Economic Research Service review concluded years ago that restricting sugar-sweetened beverages in SNAP was unlikely to reduce consumption, partly because SNAP benefits cover only a portion of a household's food spending. The new analysis suggests that reasoning was incomplete rather than wrong, since purchasing did fall, though a meaningful share of the money moved rather than disappeared.
For households on SNAP in a waiver state, the practical facts are unchanged by this paper. Restricted items can still be bought with other money. Restrictions differ by state, are paused in several states, and the state agency administering benefits is the authoritative source for what applies locally. Peer review of this working paper and a longer follow-up are the next things worth waiting for, alongside separate federal moves on what goes into the food supply.
Key Questions Answered
What did the study find? Soda purchases fell about 12 percent among SNAP households in 10 states with new restrictions, based on grocery purchase data from the first half of 2026, or roughly 34 fewer 12-ounce cans per person per year.
Does that mean people are drinking less soda? The analysis measured purchases, not consumption or health outcomes. Those require different study designs and longer follow-up.
Did people just buy soda with other money? Partly. Recipients redirected up to 39 percent of the money not spent on soda to other sugary drinks and juices not covered by the restrictions.
Who funded the study, and has it been peer-reviewed? It was supported by a grant from Bloomberg Philanthropies, which advocates for sugary drink taxes, and was circulated as a National Bureau of Economic Research working paper, a preliminary format that has not been peer-reviewed.
What about the 34,000 fewer diabetes cases? That is a modeled 10-year projection built on the purchasing findings, not an observed outcome. It carries the uncertainty of every assumption in the model.
Were there other findings? A survey run alongside the study found recipients were more likely to report feeling judged or disrespected under the restrictions, which an outside researcher called concerning.
Can SNAP recipients still buy soda? Yes, using money other than benefits. Rules differ by state and are currently suspended in five, and the state agency administering benefits is the authoritative source for local specifics.
