Since last fall, a specific type of letter has been showing up in mailboxes in small Louisiana towns, suburban Ohio, and rural Arkansas. It is issued by the state, printed on thin paper bearing a government seal, and informs the recipient of something rather direct: either show that you are employed, or your food benefits will be terminated. It is like an eviction notice for the refrigerator for families who have been dependent on SNAP for years.
These letters have a well-known political message. Supporters contend that work requirements will encourage physically fit adults to leave the couch and enter the workforce. It’s the kind of neat narrative that works well in a campaign advertisement. The evidence consistently refuses to cooperate, which is the problem. Work requirements do not significantly increase employment, according to a recent analysis from The Hamilton Project, which is housed at the left-leaning Brookings Institution. They consistently reduce the number of people receiving assistance.
This distinction is more important than it might seem. Approximately one in four enrollees lost their coverage when Arkansas briefly experimented with Medicaid work requirements a few years ago. Not because they didn’t want to work. Many were already employed, taking care of a parent, or mired in the bureaucratic maze of attempting to upload documents from a phone with a spotty signal. The disqualifier was the paperwork itself. Speaking with researchers who investigate this gives the impression that the regulations are more intended to thin the rolls through attrition than to promote employment.
In a recent interview, Lauren Bauer, who oversees some of this work at Brookings, made a startling statement. SNAP could serve as a workforce program, she once thought. She no longer does. She told Stateline, “I am now of the mind that SNAP should be an anti-hunger program,” implying that job placement, career training, and food assistance shouldn’t be combined. It’s a subtle but important admission from someone who has worked with data for years.
Meanwhile, the data continues to pile up. The Center on Budget and Policy Priorities estimates that since the One Big Beautiful Bill Act was signed last summer, at least 2.5 million low-income Americans have lost their SNAP benefits. That represents about 6% of all enrolled individuals. For the first time since the pandemic, program rolls have dropped below 40 million, according to USDA, which has presented the decline as a success. In a June announcement, Speaker Mike Johnson referred to SNAP as “bloated” while assuring those who genuinely need assistance that they will still receive it. Depending on who you ask, that promise may or may not be fulfilled.
The logic of the policy during a downturn is more difficult to defend. Historically, SNAP has been one of the quickest ways to help someone who recently lost their job. You apply, receive assistance, and eat while you search for a new position. That is reversed by the new regulations. If you lose your job, you might also lose your food benefits under the stricter regime because you would no longer be employed. “I lost my job, I need food benefits,” Bauer said in describing the encounter with the dire economy. You can only receive food benefits if you are employed. It reads like a committee-written Catch-22.
The underlying premise of work requirements—that recipients are inactive and require prodding—has never quite aligned with reality. The majority of SNAP-eligible working-age adults are already employed, frequently in positions that lack regular schedules, paid time off, or consistent hours. Others are taking care of a parent who is developing dementia, a sibling with a disability, or a child. There is labor. It simply doesn’t appear on a pay stub.
It’s difficult to ignore the discrepancy between the rhetoric and the receipts as this policy develops. Workplace dignity is promised in the talking points. People are losing groceries, according to the data. A generation of families is learning to cut back on their spending somewhere in the middle.
