Spotlight Exclusives

At Least 100,000 More Kids Could Be Eligible for SSI Payments, Study Finds

Michael Levere and David Wittenburg Michael Levere and David Wittenburg, posted on

Children’s participation in the federal Supplemental Security Income (SSI) program has declined substantially over the past decade.  Many children with disabilities might be eligible for SSI, yet barriers such as a lack of knowledge of the program or perceived challenges with applying may limit participation.  In a new paper, David Wittenburg, a senior fellow at Mathematica, and Michael Levere, an assistant professor of economics at Colgate University and a senior researcher at Mathematica, use machine learning models on Medicaid administrative data to estimate that as many as 650,000 kids could be eligible but are not receiving SSI payments. Wittenburg and Levere spoke to Spotlight recently about the study; the transcript has been lightly edited for length and clarity.

Why don’t we start with an explanation of what SSI is.

David Wittenburg: Supplemental Security Income is the nation’s largest federal program that provides supports to low-income families with disabilities and elderly adults. There are different eligibility criteria for children, adults, and families. One thing that is consistent for SSI is that you have to have limited means and assets to qualify and there’s been a lot of media attention around the asset limits, which have not changed in a very long time. In addition to those, there are disability eligibility criteria for children and adults. For children, those criteria center around the child’s disability, and for adults, it’s around work capacity and the eligibility changes at age 18. So, if you qualify as a child for SSI at age 18, you have to get redetermined under the adult definitions and a number of children will continue on, but a number of lose benefits. Mike, is there anything that you want to add?

Michael Levere: I think the only thing that you didn’t mention is that SSI offers people who are eligible a monthly cash payment and the maximum payment right now is $943 per month in 2024. Almost all of the time, it also means that you’ll automatically qualify for Medicaid coverage.

So, participation by kids had been going down even prior to the pandemic. Talk a little bit about that.

Levere: SSI was established in 1974 and for most of the program’s history, participation generally increased. It really increased in the early 90s, had a slight tick down in 1996, and then increased through about 2013. But since 2013, the number of child SSI participants has gone down. We’ve done some research related to this, but I would say it’s not exactly clear what’s been the driving force. The economy was generally improving during that time but the asset limit of $2,000 remained at a fixed nominal amount. So, probably fewer people lived in families that met that $2,000 limit. There were also increases in continuing disability reviews during this time and a large backlog built up.

But none of those factors can fully explain things because then during the pandemic, the economy got worse, and participation kept going down. Those continuing disability reviews slowed down, and participation kept going down. So, there’s a lot of uncertainty around exactly what’s causing it. From studies of other programs, we know that things like complications in signing up for programs can be a barrier, or a lot of people may not know about the program, or there may be a stigma associated with the program.

Wittenburg:  To reinforce what Mike just said, one way to think about it is that SSI is a heavily networked program. And so, when you think about Supplemental Security Income, you can’t just think about it as income supports for families with children with disabilities. You have to think about what other supports are in the area. And so, when you talk about the pandemic, you had things that happened during the pandemic, namely the SSI field office closings were huge because you had to go to those field offices to do some of the eligibility requirements.

When you look at maps of SSI participation, one of the things you see is that there’s heavy clusters of SSI participation across counties. And one of the implications is that people really know about SSI in certain areas and not so much in other areas. So, even though it’s this federal program, it very much operates like a state program. And when it interacts with those other state programs, the need for SSI might change. Say you have a disability— attention deficit hyperactivity disorders are a big one. If you are in an area with really good vocational rehabilitation programs and special education programs, you may not need SSI relative to somebody that lives in a rural setting that doesn’t have any of those supports. That’s part of the explanation of why SSI varies so much across states. And I note that to just add on to Mike’s point that there isn’t an easy explanation for the decrease in participation because the explanation really varies across all these different areas.

For either of you, let’s talk about the study—what you were looking for and what you found.

Levere: There’s been this decline and while there’s been a variation in terms of where that decline has existed, all the data suggests there may be people out there who might benefit from having SSI but who don’t currently get it. And so, we wanted to understand more about that population, because there’s been analyses done of other programs that show something like 20% of people who are eligible for their Earned Income Tax credit don’t file and claim it. So, how many people are eligible for SSI but don’t get it?

What we did is we looked at Medicaid data and we looked at Medicaid claims, which give us a sense of what is the healthcare status of a lot of children, especially low-income children, to see what are the characteristics of people who are on SSI and therefore, how many people have similar characteristics but are not on SSI? What are the people who have similar healthcare claims where we can then say that person is likely eligible? And what we ultimately do is estimate for each child Medicaid recipient, what is the probability that they are eligible for SSI. And we then characterize, based on where we define the threshold of what is potentially eligible. If the threshold is 40% probability, it looks like there would be about 110,000 more children across the country who would be eligible for SSI. If the threshold were 20%. I think that number is about 650,000.

David, did you want to add anything to that?

Wittenburg: Mike described it well, but I’m going to stay with the cross-state variation theme and dig into that. One of the bad ideas I had early on in the paper was, let’s figure out if we can combine models across states. And that didn’t work at all for reasons that are totally predictable. SSI is so varied across states, as is Medicaid. And coming back to the overall point of how SSI operates in this complex ecosystem that heavily influences the outcomes, even if Congress made a change to specific SSI policy, it may not influence participation in ways that are predictable. This is something we’ve seen historically over time, and I think that the findings in our paper just underscore that.

Levere: Dave is making a really important point, which is that one of the things we’ve done—it’s not in the working paper version but we’re starting to go through the publication process and add it in— is we identified the typical claims behavior of somebody who has ADHD, or who has cerebral palsy. And we then estimated for the same person across each of the 32 states that were able to do this analysis, what’s the probability that they would be eligible? And we see very wide variation in terms of that and there’s not a clear pattern. And so that says that there’s a lot more going on in the background, that there are these big policy level variations across states that’s really driving this variation in SSI participation.

So, Mike, just following up on that thought, it, it sounds like it’s very difficult to come up with strategies to try to figure out ways to drive participation.

It’s a really tricky question and one SSA has been thinking about because they’re really interested in making sure that people who are eligible are able to get the benefits that they are entitled to. I cited at the beginning a few of the factors that may play a role. There could be stigma. There could be people that don’t know about the program. There might be administrative hurdles where people know about the program, but the application process is viewed as really burdensome.

One thing that we think that is really beneficial about this paper is it highlights that there are so many potentially eligible kids who are already participating in a social safety net program. And so, with data-driven approaches, you can imagine where there’s data sharing between the Social Security Administration and the Centers for Medicare and Medicaid Services that allows them to identify these exact people and reach out to them. You could work with doctors to help make that connection. You could also streamline the SSA application process where all of the medical data that you need is just pulled in directly from Medicaid data so that it’s easier for the person, It’s easier for doctors, it’s easier for SSA.

David, did you want to add to that?

Wittenburg:  Mike said it well, but I would make three points about trying to increase outreach to prospective eligible participants. Point number one is to just reinforce what Mike said, which is if the application process is complicated, simplify that process using an electronic form—I call it TurboTax in the SSI process. If you go online and you try to apply for SSI, they’re paper forms and you pretty much need a lawyer to go through them. They’re dense.

Point two is the outreach itself. And to reinforce some of Mike’s points, if you’re thinking about general outreach, and there’s this mass variation across states, general outreach strategies are kind of difficult. If your goal is to increase SSI participation, targeted outreach matters. And that’s what I think is part of the strength of the work Mike and I have done here, because it looks at different modeling and says, if you wanted to go increase SSI participation do you want to send a general news blast? Or do you want to start thinking about where people are who are potentially eligible who could benefit from the program?

The third thing that I think creates challenges is the interaction with other programs. And I think one of the conceptual problems that has plagued policy around this issue, is there will be a whiplash with SSI. We’re currently in a state where SSI is dropping, and so there’s a big push to increase it, but if it starts to increase, based on history, there will be concerns that people are getting it who shouldn’t be. The point that I’m trying to make here is that there’s no established ideal for what SSI participation should be. I can’t tell you what SSI should look like in the state of Pennsylvania or the state of Mississippi or the state of Oregon. They have wildly different participation rates. And part of that is those ideals may vary depending on the states. Like SSI might be really needed in some areas and in other areas it might not be as needed because there are these really strong infrastructures to support youth and families.

To try to tie all three points together, people with disabilities have to operate in a fragmented system that is all over the place. Everything is based on the original conditions of these programs, which makes coordination potentially challenging, but it is possible to increase SSI participation, I think, through targeted outreach.


David Wittenburg is a senior fellow at Mathematica.



Michael Levere is an assistant professor of economics at Colgate University and a senior researcher at Mathematica.

« Back to Spotlight Exclusives