SNAP Waste & Fraud: A Conversation with Keith Hall

In this episode, Veronique de Rugy, a Senior Research Fellow and George Gibbs Chair in Political Economy here at Mercatus, chats with Distinguished Visiting Fellow and former CBO Director, Keith Hall, about his latest research documenting SNAP waste, explain why fraud exists in the program, and present solutions for reforms. This episode features audio from a recent Mercatus webinar. 

If you would like to connect with a scholar featured in this episode, please email the Mercatus Outreach team at [email protected]. 

Check out Keith's research on SNAP.

Note: While transcripts are lightly edited, they are not rigorously proofed for accuracy. If you notice an error, please send us an email.

VERONIQUE DE RUGY: My name is Véronique de Rugy, I'm a Senior Research Fellow here at the Mercatus Center and I hold the Gibbs Chair in Political Economy and it's my pleasure today to do this webinar with my colleague, Keith Hall. Keith was the Director of the Congressional Budget Office and he held that position from 2015 to 2019, but he's also held many other positions at the ITC, at the Department of Commerce, at the Department of Labor and today, we're going to talk about program integrity and in particular, program integrity in SNAP.

This is not the first work that Keith has done on program integrity and it's extremely timely because anyone who has been following the reconciliation bills, the one in the House, the one in the Senate, knows that they are actually aspect of that issue that are trying to be addressed there, not necessarily as purely as Keith would do but still, they're steps in the right direction, so it shows that it's on the mind of a lot of people.

Welcome to all of you. I'm going to be talking a little bit with Keith and then we'll open it up for questions for you guys. Drop your question in the chat, and here we go. Keith, so you wrote this paper for Mercatus called Reducing Waste and Fraud in SNAP and it just came out and you've been talking to legislators and you've been talking to different groups. Why did you write this paper and in particular, why did you write it about this particular program?

KEITH HALL: Well, I'm generally worried about waste and fraud in the government. The waste and fraud levels have been at record levels, quite a lot of them coming off of the pandemic spending. The General Accountability Office has estimated that there's just 260 plus billion dollars worth of waste and fraud just this year alone. That's a tremendous amount of money, and the SNAP program is a big part of that. It's a large program and it has an incredible level of waste and fraud.

The improper payments, which are payments that the program either made to somebody who wasn't eligible or didn't make to somebody who was eligible, those improper payments are 11%-12% of the entire program. I focus in particular on the overpayments because that causes the federal government money, but it's also what dominates their improper payments. 10% of the program is waste and fraud, it's tremendously high, and so I took a look at the program as an example of all the waste and fraud that we're experiencing and there are a number of things, aspects of the program that I think just lead to this high level of waste and fraud that could be fixed.

DE RUGY: The other thing that you highlight in your paper is how little efforts it seems is done at the state or federal level to actually go after these improper payments. If anything, improper payments, they grow every year.

HALL: That's right, and that's the other notable thing on the program. The level of waste and fraud is high but literally, it grows every year. It has now grown for over 10 straight years. The level of waste and fraud, the error in their delivery just increases and they don't seem to be able to even slow the increase let alone reduce the waste and fraud. That's a frustrating thing.

Certainly, part of that, as you mentioned is something that I highlight in the paper, is that the program is entirely funded. The benefits are entirely funded by the federal government but it's entirely administered by states and states don't even have any contribution at all to the benefit level. I do think there is potentially an issue there with incentives. States are tasked with trying to stop waste and fraud, they have to spend money on it but they don't reap any of the benefits from doing that, the federal government does. I think there's some issue there with what economists call the agency problem.

DE RUGY: They may also have maybe that feel that there's an incentive or a political advantage in expending, actually, eligibility or not cracking down on fraud and waste.

HALL: That's right. I think it's an interesting indicator that if you look at the errors, the improper payments, the overpayments are five times as large as the underpayments, which is suggesting that it's something much more than just sloppiness in the waste and fraud. That there's a real bias towards overpayments. Part of that is almost surely fraud and even trafficking in SNAP benefits. The other, of course, is the incentive, as you mentioned, for the states.

DE RUGY: What's interesting to me about your paper, which I hope everyone will read, it's actually extremely accessible, it's phrased, it's not wonky language, it's easy to understand, is how it mirrors the improper payment problem at the federal level with the same exact profile, which is it's big, it's growing year after year after year for the same programs.

5% of improper payments are underpayment. Most of it is overpayments. Very little effort to go after. Part of philosophy, that is you pay and then you chase which is when there's a lot of fraud, is not very effective because the guys have disappeared. It's also this kind of fear of saying, we'd rather overpay than not pay someone who-- it's really striking because while it's administered by the states, it is really a government mindset which, of course, wouldn't happen in the private sector. Because they would go under if they actually face these rates of improper payment. Let's talk about incentives. In your paper, you identify a series of ways the program is designed that creates wrong incentives, like threshold, so can you talk to us about all the incentives that lead to this problem?

HALL: I'll start with the delivery of the program. The electronic benefits transfer. People get their SNAP benefits with what looks like a credit card. They take it to the store, to the retailer, use it to pay for food. Well, if you look in your wallet, every credit card in your wallet has a little chip on it, a security chip, that helps prevent the number from being stolen.

Well, these SNAP cards don't have any security on them at all. Literally, you have all 50 states, each of the states is issuing their own EBT card, their own SNAP card, and there isn't even any security. There's probably a very high level of fraud and trafficking and nothing's been done about it. They've used these cards for 20 years. Like I say, every other card, every credit card has these chips, but there doesn't seem to be any incentive to put security on these cards. Then you've got issues where people move from state to state and there's an issue with people collecting cards from different states. You could move from one state to another, get a second EBT card, and use both. You've doubled your benefits. That's a level of fraud that's probably the biggest way fraud occurs in the program.

Eligibility, enforcing eligibility is a problem. There is now something called broad-based eligibility where local government or state government will issue some simple little benefit from TANF. Here's a phone number to call if you have problems. Here's a brochure. That, making a brochure accessible, makes somebody eligible for SNAP, which seems ridiculous, but that's how most of the eligibility for SNAP occurs, is people get some sort of simple benefit like that, that if you were cynical, you'd say was designed to make people eligible for SNAP.

There are issues of how they measure improper payments. The program has been pushed to measure improper payments more precisely. They've got to measure now where they go back and resample benefits and make sure that people are actually eligible. Well, one of the troubles with that is there's a threshold where when they find overpayments that are below a certain level, below $57, they don't even count it.

The trouble with that is the General Accountability Office looked at this habit of using this threshold and say they're missing, at least they were in 2014, 38% of the improper payments. When we talk about this very high level of overpayments, like 10% of the program, you add in 38%, that's more like 17% or 18% of the program. Even how they measure improper payments is a problem and seems to be reducing the measure of just how big a problem this is.

DE RUGY: Do you know how they came up with that $57 threshold? It seems like a random number.

HALL: Well, they've had a threshold for years and they've adjusted it. In 2014, when the General Accountability Office looked at it and said, hey, this threshold is missing a lot, the threshold at that time was $50. Then the next year they dropped it down to 14 and a lower level. Now they just raised it. The logic is interesting. They want the states to focus on large overpayments, but I don't see how using this threshold, which doesn't even count small overpayments, I don't see how that benefits the program. Again, it has this side effect of making the problem seem less severe than it really is.

DE RUGY: Yes. If it's almost 40% that is missing, it is a real problem. Let's talk about your solutions. You have a series of solution that you highlight in the paper. For instance, for this threshold, what would you do?

HALL: Well there's no reason to have the threshold. I would just eliminate the threshold. You're measuring all the overpayments, you've got a more accurate number. If you want states to focus on big overpayments, they can do that. It's not distracting to have this bigger measure, it's quite the opposite. One of the things I think is a fundamental problem is the fact that states have no responsibility for paying the benefits.

There's an incentive problem. For them to track down fraud and waste, for them to track it down, they don't get anything out of doing that. One of the things I would do is I would have states share with the federal government in the cost of the program, in the cost of providing benefits for the program, even a modest amount. I've seen proposals as small as 5% going up to 25%. [crosstalk]

DE RUGY: That's actually in the House. The House Reconciliation Bill, that is the change for cost-sharing for the states, for SNAP is exactly this one, this 5%. For states with high error rate, it would go to 25%.

HALL: That's right. That's trying to get the incentives back in line. It's hard to estimate how big an effect that would have, but it's clearly a problem with incentives the way it is now.

DE RUGY: CBU has tried to score, actually, the impact that this would have, right?

HALL: Yes, that's right. They did put down a score. They talked about, for that particular part of the bill would save about $130 billion over 10 years. It's not a huge amount, but it's a reasonable amount of money. A lot of the savings, this is their estimate, they don't know for sure, is they think that states might actually pull back on how many benefits they provide. It's actually not the federal government who'd reduce benefits. That whole bill part, people have pointed out how many people would drop their SNAP benefits. That would come from state decisions whether they want to do that or not.

DE RUGY: Yes, because one of the things that I learned yesterday from talking with you is that the states, they have an option. They have a range of benefits that they can pay based on the poverty line. It's not everyone below the poverty line gets this. It's actually a range that actually goes to simple to double, right?

HALL: That's right. The core of the program is people earning 130% of poverty or lower, and states can just raise that criteria, and include a lot more people in the program if they like. Again, this is part of the incentive issue is states are doing that, but they're not actually paying the bill. When they do that, the federal government is picking up the tab.

DE RUGY: On the Senate, the Senate Reconciliation Bill has also a cost-sharing reform. It's more modest. It's 5% with a maximum of 15%. I think it's worth noting this. We were told yesterday by a staffer that, for political reason, they felt in the Senate they couldn't go above 15%. It's worth mentioning, these are reforms that are under consideration right now. The GAO, the Government Accountability Office, has written a lot about this problem, not just for SNAP, but for SNAP in particular, and actually has suggested change, integrity reforms and stuff. Why do you think, until now, no one is taking up GAO and its reforms?

HALL: Well, that's a good question. I think this incentive issue is at the problem somehow too, but the SNAP program itself just doesn't seem to have been very systematic in their efforts to reduce the waste and fraud. What the GAO has done is they haven't just lectured SNAP on precisely what they should be doing, but they've talked to SNAP about, gee, you need to find out what the root causes are of all this waste and fraud and figure out ways to counter it, impose those steps, and then see how it works.

This sounds really commonsensical, but it's really just GAO pushing the SNAP program itself to be accountable for this high waste and fraud rate. They just need to do it and see what works and go from there. The thing I would point out is they've tried some things, yet the waste and fraud rate--

DE RUGY: Because they spend money on the-- they've increased money on integrity, and yet--

HALL: Yet the waste and fraud goes up and up. It's not working very effectively if it constantly goes up. This is an issue really of pushing an agency to be more transparent about what they're trying to do to reduce waste and fraud, and be real clear about what they've done or what they're doing and how much it helps.

DE RUGY: You talked about return on investment for integrity measures. Can you flesh that out a little for us?

HALL: I'm a big fan on thinking about efforts on program integrity, which are efforts to reduce waste and fraud, to approach it by trying to measure how effective these program integrity steps are. If you spend $1 in program integrity, how much does it save you in waste and fraud? That's really what the return on investment approach is like. If you spend $1 and you reduce waste and fraud by $2, that's a ratio of 2:1. That's a return on investment 2:1. That's something that's worth doing.

It works both directions. If you calculate a return on investment on a number of activities in the SNAP program, some activities don't have much of a return on investment that that ROI could be less than one. Well, they should stop doing that. That's a waste of taxpayer dollars. If the return on investment's 100, well, gee, you can reduce waste and fraud 100 times what it costs to go after it, you should be doing those things.

It's a real transparency tool that other agencies have used, a few other agencies have used, and GAO has pushed them to use this. This is a sort of approach that I think the SNAP program ought to be thinking about, is how to calculate a return on investment, how to understand what their efforts are producing now, and what their future efforts might produce if they try new things.

DE RUGY: In your paper, you have a figure that's quite impressive about the administrative cost per beneficiary per state. It's an enormously wide range. What do you think explains this wide range and how much of a role does it play in states' reluctance to do any program integrity, really, measure?

HALL: It's a remarkable thing if you look at how much states spend per beneficiary on delivering the program, and it ranges. Florida, I think it's something like $85 per beneficiary a year. In Wyoming, it's $850 per beneficiary. You've got this huge variance in how much it costs for a state to deliver the program.

What's most notable about that is there's not an obvious explanation. If you could look at it and see, well, gee, it's economic conditions. Some states are in better shape than others, or it's the size of the program in that state. If that was an easy explanation, that would be more understandable. There's not a straightforward explanation, which makes you worry that it's literally part of the waste and fraud problem. It's where states aren't being very penny-wise in delivering the program.

DE RUGY: One of your recommendations is basically like a clearinghouse that all 50 states would have to participate in, right?

HALL: Right.

DE RUGY: You say that there is such a clearinghouse, but only seven states are participating in. Can you explain the concept of the clearinghouse? To the best of your knowledge, what explains that states are not even trying?

HALL: Well, the biggest single problem with improper payments on the program is people using EBT cards from two different states. Because each state works somewhat independently on producing the card, it would be extremely helpful if a state could look at a new application for a card and look into this clearinghouse that we're talking about and see if that same person is showing up in some other state. Then they could flag it, okay, here's a case of somebody who may commit fraud and use two different cards.

It seems somewhat of a no-brainer to produce this clearinghouse where all the states put their cardholders' information in there so other states can look at it. Of course, that only works if you've got all 50 states putting their information in so that any particular state could check that clearinghouse to make sure they're not having somebody trying to commit fraud. Well, this clearinghouse has been set up. That's been good. It's been there since early 2014. Only seven states have joined it. Now, if you're a state, if you're the State of Virginia, let's say, you want to see, well, here's somebody's name, let's see if they show up in some other state, this clearinghouse is not that much use because it's only showing you six other states.

DE RUGY: This actually echoes some of what people have explained to me about how it works at the federal level with improper payment and fraud. There is such a thing as a do-not-pay list, which may not be as up-to-date as it should be. What you find is that most agencies don't even actually ping it before they make any payments. Sometimes, to the extent that they ping it, they actually pay before they even get the answer. It seems that there's really no incentives to use even the resources that are put in the system to try to prevent integrity. The broad-based categorical eligibility, if you can explain what that is, and if you can explain to us how this plays a role in the integrity of the program.

HALL: Well, as it turns out, to be eligible for SNAP benefits, you can fill out forms and establish your eligibility directly with SNAP. A second thing you can do is be eligible through some other program, TANF, or some other federal program, and establish your eligibility through the other program. To reduce the burden, the SNAP program will just say, oh, well, you get benefits from another program, we'll give you benefits too. You don't have to establish your eligibility with us. This is the categorical eligibility. The vast majority of people who get SNAP just do it by declaring, hey, I'm eligible for something else, I should get SNAP too. Which is fine. I guess it reduces the burden. They don't have to do it that way.

There's something called broad-based eligibility, where some local governments have said, okay, we're going to create something. We're going to say, gee, if you're eligible for TANF, we're going to send you our brochure on your benefits. Just the fact that you get that brochure makes you eligible for SNAP. You cannot be eligible for anything else but just a brochure, and SNAP will give you benefits. That's just a much broader eligibility than the program ever intended.

DE RUGY: Which is weird, because if, really, that's all it took was to be eligible for another program, then they would not have different and varied eligibility measures. They would have just one across the board for all welfare programs, right?

HALL: That's right.

DE RUGY: Then if that really was the case, then why have all these categorical programs? Why not have just one cash transfer for poor people below a certain level? It makes no sense. It really makes no sense, because each program has eligibility standards, and they're different for a reason.

HALL: This makes it feel like a local government is trying to take advantage of the federal government and get people eligibility to SNAP when they wouldn't otherwise be eligible by this very trivial, here's a hotline if you have concerns about anything, making people eligible that way purely to get them SNAP benefits.

DE RUGY: I assume that the broad-based categorical eligibility complicates the measure of improper payments, and so probably improper payments are much broader, much bigger than we think because of that difficulty of measuring.

HALL: It might be. On that, I haven't thought that through. It's hard to know how that affects it. It certainly is expanding eligibility to some folks in ways that makes it hard to figure out the improper payments because you catch somebody who's gotten a brochure and is now eligible, you've got to now try to figure out if they meet the criteria for the program, and they haven't had to provide that information.

DE RUGY: Well, I want to turn it to our Q&A now, and so please, fill out your questions in the chat. We have a first question. How will the benefit cost-sharing with states prevent states from simply segregating their funding to certain individuals, which they will monitor closely, and then allowing overpayment to continue on the federal money?

HALL: I'm not sure I understand that question, to be honest. I like it because it puts the incentives in the right place. I don't know how much effect it will have. To be honest, I don't know what you mean by segregating their funding. Oh, okay. You're talking about just separating the SNAP funding out. Yes, I like the fact that it costs states money if they expand eligibility, and they have to pay some of the burden of that. I do think if problems arise, like you're describing here with having the states share in the benefit, then this thing can be fine-tuned. You can figure out some ways to stop this from happening.

DE RUGY: Any other questions from the audience? What are the seven states that report to the fraud list?

HALL: I don't have that in front of me. It's not that-

DE RUGY: Is it in the report?

HALL: Probably. I don't know for sure. My big point is that it should be everybody, not just seven states. I'm hoping that that number will go up over time so that seven states is not set forever, that over the next year that will increase from seven to hopefully all 50.

DE RUGY: If there were some cost-sharing, you would think that an easy step would be to actually join the clearinghouse, right?

HALL: No, that's a really good point. I think this is, if nothing else, is an example of what happens when you don't have the state share in the benefits. They don't pay any sort of cost in having overpayments on benefits. It's not their problem, almost.

DE RUGY: Is it California and Oklahoma that have started the process of putting chips in their SNAP cards?

HALL: That's right. They both declared that they're going to do that.

DE RUGY: They haven't done it yet, but they have--

HALL: That's right. They've announced they're going to. I don't know where they are in the process of doing that. It takes a little time, because you've got to develop the the cards, you've got to make sure that all the retailers have readers that will read these chips, which is just the same as every grocery store that takes your credit card that's got a chip reader on it. It shouldn't be that much, but they've got to worry about that. Implementing security on the cards might take a little bit of time, but states just have to do it. They've had plenty of time to do it.

DE RUGY: I have another question, actually. Do states join the clearinghouse legislatively or by agency decision, or does it depend on the state?

HALL: I suppose it depends upon the state. Presumably, states already have a list of people who are getting their SNAP benefits. Really, it's really whoever's holding that list just needs to make that available in the clearinghouse.

DE RUGY: Do we know the share of the improper payment that is the product of this double-state eligibility?

HALL: Well, the program says it's the biggest cause of overpayments, but the problem is when you measure improper payments, you're sampling people who get the cards and then you're trying to figure out who's getting more than they're supposed to. That doesn't really tell you how they got more than they're supposed to. That takes a bit more research. All you know, with the improper payments, is people are getting too much. Doesn't tell you as much about whether it's results from fraud or it's from trafficking. It doesn't really tell you how it occurred.

DE RUGY: Actually, this is the next question. Could you briefly explain trafficking and how that works between recipients and SNAP-accepting stores?

HALL: Sure. The idea is that if you steal the card's number through cloning or whatever, you can now have what's essentially an imitation card. Then if you're a dishonest retailer or dishonest individual, you can now use this number to get SNAP benefits or you could sell them to somebody else. Obviously, making a fake card with a fake number on it and selling it to people at a discount, that's clearly trafficking. That's the sort of thing.

The program itself says trafficking is pretty significant. When they've tried to look at it, they've found that about 1.7% of all benefits are trafficked, which is a big chunk of that 10% overpayments. They really need to develop their trafficking estimate better and get a better idea of exactly what the trafficking number is. Of course, that would help you understand exactly what the efforts are like on the trafficking, exactly what's happening, and put an end to it. This is treated like, this is how the program works. This is a $110 billion program. We're talking about over $10 billion worth of waste and fraud every year for the program. This is a huge amount of money. It's worth spending the time to lower this level of waste.

DE RUGY: Of course, you remind me that the increase in the program has been exponential between 2019 and even this year. It was from 60 billion, I think, a year to now it's 100 billion and it's heading towards 110. Am I getting this right?

HALL: Yes, it's at about 110 now. The Congressional Budget Office anticipates the program's going to stay about that size, but we'll see.

DE RUGY: Of course, the increase in the program is extremely attractive for more fraud and more elaborate schemes to actually tap into that money, right?

HALL: That's right. The program itself is countercyclical, meaning when you go into a recession, the program gets bigger. More people go onto the SNAP program to get benefits. Whatever size it is now, if we have a recession, some economic issues at some point in the future, we will in some point in the future, the size of the program will increase and the opportunities for waste and fraud and trafficking and all that will increase.

DE RUGY: I have another question. Who are the individuals typically selling benefits to? Do we know that?

HALL: I don't know. I didn't look really carefully of exactly that. The program itself hasn't spent a lot of time talking about exactly where the problems are, I think because they have measurement issues, but they've had instances where this fraud is occurring from individuals, but it's occurring by retailers. At times it's actually occurred by the government, local governments have been part of the defraud activities in this. It's just pretty widespread. It's remarkably widespread.

Again, they don't have a good measure of the fraud because they just have this notion of improper payments and they haven't drilled down perhaps, at least in my view, they should in figuring out this problem. That's part of why I talk about a more coordinated effort to go after waste and fraud, maybe creating an office of program integrity for the program that actually exists to focus on this. That doesn't take a lot of money. It's a $110 billion program. They can afford to put together a small office to go after this.

DE RUGY: There were draft reforms in-

HALL: In the draft Farm Bill.

DE RUGY: In the Farm Bill. Can you talk a little bit about that?

HALL: Sure. The Farm Bill had some good ideas. The Farm Bill talked about these problems. They talked about the threshold problem, that $57 problem. They talked about creating an office of program integrity. I thought that was a great idea. In fact, it's an idea that ought to be used in a lot of other programs, not just the SNAP program. A lot of what that bill was pushing them towards is pushing more responsibility onto the states to actually work on this problem and systematically go after the problem. Again, if you look at the level of waste and fraud, it goes up and up. It's not been successful. Their efforts to improve their integrity of the program just haven't been successful.

DE RUGY: $110 billion a year, really, it's large, especially considering the fact that these are on top of many other programs. As you talked with the broad-based category called eligibility, there was actually a ton of double dipping, right? This must create this incentive to work, for sure. In your paper, you don't talk about work requirements because it's not an integrity issue. Work requirements exist in both the House Reconciliation Bill and the Senate Reconciliation Bill. I assume you're probably in favor of some sort of work requirement?

HALL: Yes, that makes sense. What I've tried to do though here is focus on the waste and fraud aspect, but you're right, there may be this sort of secondary effect from reducing waste and fraud, they push people into the program properly. With work requirements there, it might actually help with the employment level.

DE RUGY: I have one final question. If you could only recommend one reform, one of all the reform and you were guaranteed that it would go into effect, which one do you think would have the greatest fiscal impact of all of the reforms that you proposed?

HALL: I like giving the states incentive to go after this problem. The SNAP program tries to find states if they get the error rates too high, things like that, but just giving them part of the responsibility for the benefits just changes the context of what they're doing. I don't know how to estimate how much that would help, but it's got to go change the culture in state SNAP programs, that they now have incentives to spend money and spend wisely to properly administer the program. Somebody has got to do that.

To me, that's something where you just get it started and you see how it works, and you can actually change the state contribution over time. If you think 5% is too low, you can raise it over time. If 5% looks like it's a burden, you can lower it a little bit, that sort of thing. I think getting that established is worth doing because it gets the incentives in line.

DE RUGY: Okay. Well, if there are no more questions, guys, we're going to wrap it up. I want to thank you, Keith, for writing this paper because I think it's very important. I recommend that you actually check Keith's other work on integrity, including scoring program integrity, which is broader and probably affects all of the programs. Check this paper. I've put it in the chat, I think. I'm not sure who can see what I put. Someone asked me to put the link, so I put it there. Otherwise, just google Reducing Waste and Fraud in SNAP,  Mercatus Center, and you will find the paper. If you have any questions, just email us, and we'll be more than happy to answer your questions. Thank you very much, and thank you, Keith.

HALL: Yes, thank you for spending time with us. I appreciate it.