Changes in the SNAP program as a result of last year’s Big Beautiful Bill include a provision mandating that states will be required to pay a share of SNAP benefits starting next October, with the required share determined solely by the state’s “payment error rate” for the relevant fiscal year. A new paper from the Hamilton Project by Lauren Bauer, the Project’s associate director, and Diane Whitmore Schanzenbach, McCourt Chair, McCourt School of Public Policy, Georgetown University, argues that the new rules may put the entire program in jeopardy for some states. Spotlight spoke recently with Bauer and Schanzenbach; the transcript of that conversation has been lightly edited for length and clarity.
Why don’t we start by walking through the changes that have resulted from the Big Beautiful Bill as background and then we can talk more about what you’ve found and what you’re advocating for.
Lauren Bauer: I can summarize the changes in the legislation if that's helpful to you. So, the One Big Beautiful Bill Act, which was passed last July, made substantial and sweeping changes to the Supplemental Nutrition Assistance Program, formerly the food stamp program. The change that we're going to talk the most about today is for the first time, SNAP benefits are not going to be fully federally funded, depending on where a state's payment error rate lands against this policy change.
Where your payment error rates falls into a series of bands—below six percent, between six and eight percent, eight to 10%, 10 to 13.33%, and over 13.3%—will determine whether you pay five, 10 or 15% of SNAP benefits starting in October of 2027.
That's the change we're going to talk the most about in this interview, but there were other changes. There were changes to SNAP work requirements. There had previously been a waiver for SNAP work requirements for veterans, youth who had aged out of foster care, as well as the homeless. The waivers for people with those characteristics went away. We’ve expanded the population that are subject to work requirements all the way up to age 64. So, between the ages of 18 and 64, if you're an able-bodied adult without dependents, you are subject to work requirements. But in fact, some people with dependents are also now subject to work requirements. If you're in that age range and your youngest child is between the ages of 14 and 17, you are also now subject to work requirements, both parents in the household.
Finally, they took away the safety valve for work requirements, work requirement place-based waivers. Previously, if an area had particularly high unemployment or relatively high unemployment against the national average, that place could be exempt from work requirements. Anyone who lived in that location could be exempt. And most of the rules that allowed states to qualify or locations to qualify for those waivers were taken away. Now only if you have an elevated unemployment rate of 10% or higher can you qualify for a waiver, which functionally negated the capacity for places to qualify for waivers anymore because that is an exceptionally high unemployment rate.
Our focus is going to be on the payment error rates and just to hit the headline, you’re predicting that for some states, this could mean they may have to opt out of SNAP?
Diane Schanzenbach: That's right. These are going to be large, unprecedented dollar amounts added to state budgets. And as we show, it's going to be really difficult to predict how much money state legislatures are going to have to set aside because so much about the rate is about noise in that measure. But then also, one of the things that has functioned so well with SNAP in the past is when the economy turns downward, SNAP can automatically expand. And some of these changes will dampen the policy's ability to respond when there's an economic downturn.
And I want you to get into a little bit more detail on that. It's very difficult to calculate the payment error rate—explain that a bit more if you could.
Schanzenbach: SNAP is a complicated program to administer, and there are lots of things that you need to account for—for example, how much of a housing payment do you have, how much do you spend on child care, etc. Sometimes, those things are not included correctly. And as a result, when they run your case through the payment formula, you end up getting too much benefit or too little benefit. When that happens, there's some error tolerance, about $58, that's already built in. But if because of a mistake, you get more than that or less than that, the difference between how much you received and that $58 counts toward the payment error rate.
But it doesn't just stop there. For example, you might have seen that Alaska had very, very high payment error rates, but it was because somebody doing the calculation accidentally typed in 0.2 instead of 0.3. States were given a lot of flexibility around COVID, and some of those flexibilities got taken away, and the state didn't adjust quickly enough or there was a miscommunication. And as a result, people who did not re-certify in enough time, who wouldn't have had to under the additional flexibility, all of the benefits paid to them count in error. And so, because this is a complicated rate, when you have someone that gets benefits but technically should not have because they didn't re-certify in time or because they didn't sign a correct form, that really adds a lot to the payment error rate.
That's just the first part, about the administration of the program. And I want to just say that overpayments and underpayments aren't equal and opposite. There are many reasons why we would expect to see the overpayment rates higher. For example, if you didn't sign in the right place or if you missed your recertification window by a couple of days, that means everything counts towards this overpayment.
States work on quality control, and they go back and do a deep dive of some number of cases to try to improve the integrity of the program. We've had payment error rate and quality assurance programs for a very long time, but we're putting a lot of additional pressure on these programs now because of HR1. It's basically more weight than the structure can bear.
The way the process works, the states do their own audits, and the feds come and do an audit of some of those remaining cases. At that point, we've got relatively small sample sizes, like somewhere between 250 and 350 cases in the entire state. And as a result of that, there's a lot of statistical uncertainty. And the statistical uncertainty is sufficiently large that when you compare it to the very narrow bands over which the cost share rate goes up, the amount of statistical noise relative to the bandwidth is out of whack.
What happens then is there's a good chance that, let’s say you’ve found a 7.9% payment error rate in your statistical sample, but that could have just as easily been 6% or 9% because of this. But if you fall in 6% versus 9%, your state is saddled with a much different payment portion.
And what are we talking about? Millions of dollars, tens of millions of dollars?
Bauer: Hundreds of millions of dollars.
Schanzenbach: And in the piece, we benchmark this to all the other spending a state does on social benefits and things like that. This isn't just like a speed bump on your budget. It's a lot of money that then would need to be set aside, but then it's also subject to kind of the luck of the draw. You don't even know your potential dollar amount. We've provided ranges of dollar amounts because the statistical uncertainty is such that you could end up paying nothing or you could end up paying something. It is very hard to predict year to year where that's going to land.
I’m assuming the final arbiter of this would be the Department of Agriculture. Is there any kind of appeal mechanism for a state?
Bauer: I'm not aware of an appeal process here. One thing that's important to know right now is there's been no regulation or rule making around this policy yet. It is an unclickable link at USDA's One Day Beautiful Bill Act webpage. And so, there's a lot of speculation and a lot of concern, and many people have raised pretty fundamental questions about the advisability of this policy. We are among them. And the fact that we are getting no guidance from USDA about how they are handling any of these questions, period, much less a detailed FAQ for states is certainly contributing to concern and misunderstanding.
And I assume that the other changes in the program, particularly, the work requirements, only increases the level of, to use your analogy, Diane, the noise in the system and makes it even more possible that you get confusion and chaos.
Bauer: Right. The harder it is to administer the program, the more likely it is that you're going to be making a payment error.
Schanzenbach: And a lot of those people won't really be eligible in the end. And so, all of their benefits would count against the payment error rate. So, it's going to set up a system that encourages states to deny benefits. We've seen more than five million people lose benefits since the One Big Beautiful Bill Act was signed. We think at least part of that is due to this incentive to deny benefits to eligible households because to put them into the system makes them possibly an error.
Your recommendation is to repeal this provision in the bill. Short of that, what’s the status of any congressional action?
Bauer: The Senate Agriculture Committee passed its version of the Farm Bill a few weeks ago that calls for a one-year delay and an increase in the top cost shift rate to 20 percent. There are no changes in the House bill.
And from a 20,000-foot perspective, there is this perception being pushed by the administration that this is a program that is riddled with fraud. You folks are experts in this. Is that true?
Bauer: No. A SNAP participant is more likely to be a victim of fraud than a perpetrator of it. This really is a very basic level of support. You don't want to be living on SNAP benefits alone. There's really very little evidence that many people are dissuaded from work. It helps many, many people afford the foods that they need. The gap between what we know from the research and how the program has been talked about, I've never seen it so large.