Skip to main content

Advanced Search

Advanced Search

Current Filters

Filter your query

Publication Types



The Dose


What Happened When One State Made Having a Job a Requirement for Medicaid

What Happened When One State Made Having a Job a Requirement for Medicaid
  • A “work requirements” program in Arkansas to improve health and increase employment led to thousands of people losing health insurance and other unintended consequences.

  • In Arkansas, the first state to implement Medicaid work requirements, more than 17,000 people lost health coverage, and few people gained employment. Listen to this episode of The Dose to learn more.

Many poor Americans get health coverage through Medicaid. Last year, with encouragement from the Trump administration, some states decided to change their Medicaid programs: for people to be eligible, they would need to show they were working or meeting some other requirement.

For the newest episode of The Dose, Benjamin Sommers, M.D., a professor of health policy and economics at the Harvard School of Public Health, talks to host Shanoor Seervai about what happened in Arkansas, the first state to implement Medicaid work requirements.

While the state’s program was intended to promote employment and better health, it instead led to more than 17,000 people losing their health coverage in just three months — and no significant increase in the number of people with jobs.

Listen to learn more, and then subscribe to The Dose wherever you get your podcasts.

Should having a job be a requirement to get health coverage through Medicaid? Tell us what you think – send an email to [email protected].


SHANOOR SEERVAI: Hi, everyone. Welcome back to The Dose. I hope you had a great summer. We’ve been planning and have a ton of great episodes for you this fall, so stay tuned. Today my guest is Ben Sommers, a professor of health policy and economics at the Harvard School of Public Health. Ben’s research focuses on Medicaid, the government program for poor Americans. That’s around 75 million, so one in five people who get their health care through Medicaid. Last year, some states decided to try and change the criteria for getting health coverage through Medicaid. They wanted people to show they had jobs or meet some other requirement to get their health coverage. Arkansas was the first state to implement what are called “work requirements,” and Ben and his team of researchers wanted to know if this program was working. Ben, welcome to the show.

BEN SOMMERS: Thanks so much for having me.

SHANOOR SEERVAI: All right. To get us started, tell me in a little more detail, what are work requirements?

BEN SOMMERS: So the term that the federal government uses for these is community engagement requirements, and the idea is that in addition to meeting the usual criteria for Medicaid eligibility based on your income or other family factors, the program now, in states that are using these requirements, mandates that individuals either are working, and that’s the most common way to meet the requirements — typically 20 hours a week. Or they fall into several other categories like doing community service, they’re in school, they’re caring for a family member at home, or they qualify for an exemption such as a disability that keeps them from working. And the basic idea here is that if people are meeting these requirements, they can keep their coverage, and if not, they have a certain amount of time to get in compliance, otherwise they lose their Medicaid coverage.

SHANOOR SEERVAI: Got it. And so from your research in Arkansas, what did you find?

BEN SOMMERS: What we found was that in the first six months after implementation, there was a significant increase in the number of people who had become uninsured. And in Arkansas they focused the program on 30-to-49-year-olds. So when we looked at that age group in the state among low-income adults, we find this increase in the uninsured rate from 10.5 percent to 14.5 percent who had no health insurance. And meanwhile we see a decline in the number of people who report that they were getting Medicaid from the state of Arkansas. When we looked at employment to see, well, okay, so some people lost coverage but maybe more people went and get jobs, we don’t see any significant change either in the number of people working, the hours they’re working, or in other activities like community service or job search.

SHANOOR SEERVAI: So basically, what this means is that 17,000 people lost health care in a three-month period, but nobody did what the policy hoped it would make them do, which is start working or show they were doing something else.

BEN SOMMERS: So that 17,000 number comes from the state’s official statistics where they were tracking how many people were being removed from Medicaid based on either not satisfying the requirements, or in many cases just not reporting anything at all. So between October and December 2018, about 17,000 or 18,000 people were removed from Medicaid.

But what wasn’t clear was, well, what happened to them? Advocates of the program said, “Well, maybe a lot of them got jobs and found health insurance through work and that’s a good thing. They’re more self-sufficient; we will spend less taxpayer dollars.” What we found was there wasn’t a significant increase in the number of people working, we don’t see a significant increase in the number of people with private coverage, we only see this loss of Medicaid and rise in the uninsured.

Now you said, no one was working. We can’t say no one. Is it possible that a couple people got jobs? Of course. What we don’t see though, in the aggregate, is evidence that there was a significant increase in employment. So there’s no big surge of people in this income range flocking to get new jobs or new activities like community service or school.

SHANOOR SEERVAI: But if the case was that people were working or met the other criteria, why did they lose coverage, then? What happened between them doing what they were supposed to be doing and still not getting their health coverage?

BEN SOMMERS: Yeah, well, that’s a really important question. So in addition to asking people about their coverage and their employment, we also did a deep dive into all the other ways they might be satisfying their requirement. We asked them about things like job search and job training, community service, childcare, schooling, disability status. So we had a lot of information about people that let us determine for ourselves, would this person be meeting the requirements that the state now has? And what we found was that in Arkansas, actually about 97 percent of the people in this age range who were supposed to be, you know, meeting their requirement, already were doing that, even before the policy took effect, because most of them were either working or had a disability, and of the remaining folks, they were in school or parent of a small — young child or family member who needed extra help at home.

So it was really only about three percent of the sample that could have even changed their behavior in a way that the policy was intending. So then you ask the natural question, which is why did so many people lose coverage? It turns out that this comes back to this question of reporting. It’s not enough to be doing these things. The state has to know that you’re doing it, and if you didn’t make any reporting or if you didn’t do it every month, sometimes people would end up losing coverage because the state simply didn’t know. And what we found in our survey in some follow-up questions is that a lot of people in the state were very confused about this policy.

Only two out of three people in the age range in Medicaid who were subject to the policy had ever even heard of it. So one out of three said, “No, I’ve heard nothing about this policy.” Not surprisingly, if you haven’t heard about it you’re pretty unlikely to be reporting anything to the state. And even of the people who knew about the policy and knew they were supposed to be reporting to the state, only about half were. There were a lot of reasons people told us that they weren’t making that regular reporting to the state, and most of those situations would lead to the loss of coverage, even if you were doing the things you were supposed to be doing otherwise.

SHANOOR SEERVAI: Could you give me examples of what those reasons were?

BEN SOMMERS: Sure. So we asked people what’s the main reason that you’re not reporting if you were required to, and the top answer we heard from people was actually, “Well, I don’t think I’m going to meet the requirements.” Now that’s a reasonable explanation of why you wouldn’t go through the hassle of logging into this website and reporting information, but the wrinkle is we know from the other questions we asked them that they were meeting the requirements. So this kind of comes back to this issue of confusion.

So people said, “Well, either I’m not working or I’m not working enough, so I’m just not going to bother doing this.” But you know, maybe they told us earlier, “But actually I have a disability,” or you know, “I’m caring for my elderly mother who’s ill.” And so we knew that, based on these other questions, that essentially all of the people who said, “Well, I didn’t report because I didn’t think I’d be eligible,” actually should have reported and would have gotten to keep their coverage. Other explanations people offered that they — about one in three said, “Well, I don’t have good internet access.” Up until very late in 2018, the only way to report this information to the state was through the state website, and Arkansas has the lowest rates of internet access of any state in the U.S.

SHANOOR SEERVAI: That’s really shocking to me, that in a place where such few people have regular access to the internet that the state would implement something that requires internet access, and if this is correct, people have to report every month. So that means that this isn’t just a one-time thing. It’s a repeated burden of having to go back, find a way to access the internet, and say you’re doing what you’ve always been doing.

BEN SOMMERS: The state heard those complaints and about midway through December added a telephone option. But that was already after more than 10,000 people had lost coverage. There also were limited hours so the website would shut down every night, which also, if you’re working irregular hours, even if you have internet access, you just may not be available to submit your information. And the monthly requirement was also a factor that in focus groups has come up. Not in our work, but in some of the other interviews that have been done, people said, “Well, I did report, but I didn’t realize I had to do it every month.” And so I told them in the first month, “Yeah, I’m working, and since I didn’t change jobs I thought that was enough.”

So red tape is certainly a big part of this story, though I don’t think it’s the only part of the story because there were other ways that Arkansas actually was trying to make life easier for beneficiaries to meet their requirements. For instance, two-thirds of the total number of people subject to the new policy actually were exempt from reporting because the state knew from other information that they were already working or that they had a disability. And so in that case, these are people who never had to go through that website process.

SHANOOR SEERVAI: Now the people who the state had already had data for, where did the state get this data and then how did they let these people know they were okay on reporting?

BEN SOMMERS: So the sources can include other income data, for instance through quarterly wage data or state income taxes. If people had applied for other programs in the states, for instance food stamps and the SNAP program. So if they had information from these other sources either about employment or about meeting other aspects of community engagement requirements, or established that they had a disability, the state said you don’t need to report. Now, the state was regularly trying to communicate with beneficiaries — they were sending out letters, they had a variety of outreach strategies.

Now the good news is that if you’re exempt from the requirements it actually doesn’t matter whether you know about it, because essentially that’s all going on behind the scenes. The problem is that a lot of people who weren’t exempt didn’t understand or didn’t know about the requirements. A lot of this is pretty universal to the Medicaid population in the U.S. Medicaid is designed as a program that covers people when their incomes are low. People’s income moves up and down, so people come in and out of the Medicaid program. People move. Not everybody reads their mail. I’m sure all of us are guilty of throwing out a form letter in the mail that we said, “Ah, that looks like spam, I’m not going to read it.”

SHANOOR SEERVAI: Yeah, for sure.

BEN SOMMERS: A lot of us don’t answer phone calls if we don’t know who’s calling. So it’s not straightforward to do this sort of outreach.

SHANOOR SEERVAI: Yeah, I want to dig a little bit deeper into the population — the type of people who are on Medicaid. These are people who don’t make a lot of money. They may be between jobs, they may have multiple part-time jobs, they may not speak English well, they may not have finished high school. It’s probably likely that they haven’t been to college, and it sounds like this reporting requirement was actually quite complicated. I can’t see how the state thought that it was going to be straightforward to get people to comply.

BEN SOMMERS: You know, I think it’s always important for me as a policy researcher, is that I don’t try to read between the lines as to, you know, what were the policymakers thinking on this or that. I take them at their word that they thought these policy changes would lead to more employment, more self-reliance, and would get people into private coverage. But you raised a bunch of important issues. And so for instance in our sample if we look at the low-income adults in Arkansas who participated in our survey, just a little less than 20 percent hadn’t graduated high school. Another 45 percent had graduated high school but had never gone to college.

So education is absolutely kind of an important feature to take into account in thinking of this population. And perhaps not surprisingly, but concerning, is that we found one of the strong predictors of whether people knew anything about this policy was their education. So people with less educational attainment were more likely to say, “I’ve never heard of this.” So that’s absolutely a key factor. Some of the other issues you raised, language — not everyone in Medicaid and certainly not everyone in the U.S. in general speaks English as a first language. People may be fluent in English but not particularly confident about reading technical materials.

One of the things that I kind of noticed in the process of doing this study is we have a team of people who study health policy and health care for a living, this is what we do every day, and still we constantly had to go back and look at the materials again and talk to people in the state — advocates, consumer advocacy organizations — to say, “Are we getting the details right?” Because it’s complicated. And again this is our job. So you can imagine for someone who doesn’t have a Ph.D. in health policy, for someone who doesn’t spend their career studying this, this is one of a dozen things that they have to take care of on a day-by-day basis just to keep things moving in terms of having money to pay their bills, keeping their family covered with health insurance.

SHANOOR SEERVAI: I want to go back to what you said about taking policymakers at their word. It’s important to do that. So they must have had a reason that they wanted to roll out this program. What was the justification?

BEN SOMMERS: I think the most common ones are that we don’t want to promote a culture of dependency. Policymakers that voiced this perspective say sometimes people need an extra boost. This risk of losing coverage will incentivize them to go get a job, their income will go up, they’ll get coverage through work, and that’ll lift them out of poverty. And that’s the phrase that we heard from the director of Medicaid in the federal government, and several of the state officials who supported it.

They also point to the fact that if you get people into private coverage then there’s less drain on public programs and public financing, so this will save taxpayers’ dollars. And there’s also an argument that feeds into this, which is there are some critics of Medicaid who feel that it’s just not a very good program, and that it’s not good health care, and that’s a reason to try to get people out of the program and into other types of insurance. Now, the first couple points there — lifting people out of poverty, getting them to other jobs — that was all speculative. You know, no one had tried this before, and now we have a little more evidence from our study and some of the other qualitative work, that at least at first pass, this didn’t succeed in that goal. We didn’t get more people to work, we didn’t get more people shifting to job-based coverage.

On the other comment about, “Well, Medicaid coverage is no good,” that actually we have a lot of evidence on. And the argument that Medicaid coverage is low quality I think is fairly disproven by this body of research, which shows that when people get Medicaid they’re much better able to afford their care, they get more preventive care and chronic disease care, we see improvements in health outcomes ranging from conditions like an appendicitis and other surgical emergencies to heart disease and even in some studies, survival. So the notion that Medicaid coverage isn’t any good is not an evidence-based one.

SHANOOR SEERVAI: I want to dig in a little more to the first justification you brought up for why policymakers want to implement this program, and it’s this idea that people should be self-reliant, and I think it says something about the way that Americans think and feel about health care, they see it as something that you must earn. And it raises the question of, well, who deserves it? And the people who deserve it are either people who are working, otherwise engaged, or of course with the Medicare population you’ve worked your whole life and so at this age you’ve earned it. But what we’re seeing as we’re in the thick of the Democratic primary season for the 2020 election, there are candidates who believe that everyone should have health care, and it shouldn’t be contingent on deserving.

BEN SOMMERS: Yeah, I think that that’s a longstanding kind of central debate in health care in the U.S. And it’s really ultimately about values and an ethical view of health care, that there are some Americans who view this as a commodity like any other, and you can get what you can afford, and then there are others who say, “No, health care is unique and it’s something that we are really entitled to as a right, and the government needs to make sure that everyone can have it.” And I think that that certainly feeds into the issue of work requirements.

It feeds more broadly into the question of Medicaid expansion under the Affordable Care Act. There’s still more than a dozen states that have chosen not to expand Medicaid, and that leaves somewhere between two and three million low-income Americans who simply have no way to get health insurance because the state hasn’t expanded. So work requirements are somewhere in between those extremes. If you look at the numbers — for instance in our study, we’re tracking Arkansas because they were the first one to do the work requirements, but coincidentally, we had been following Arkansas for several years before this because it was one of the states in the South that expanded Medicaid early on. And this is a state that had a lot of uninsured, low-income people.

And what we find in our survey over time, where we’re looking at adults under 65 who have incomes below the Medicaid cutoff, we found that the uninsured rate started at around 40 percent in 2013, so really high. That dropped to about 10 percent by 2016, so a huge reduction in the number of people who were uninsured. So this big coverage expansion in Arkansas, which was essentially all due to the Affordable Care Act and the Medicaid expansion — so then along comes work requirements and we see that number start to tick back up. But for context here, the number goes from around 10 percent to about a little less than 15 percent.

So this is real, I mean this is 17,000 or 18,000 people losing Medicaid, many of whom are not gaining other insurance, but in perspective, this compares to the 40 percent who were uninsured before the Medicaid expansion.

SHANOOR SEERVAI: That makes me think of another justification that I’ve heard of for why policymakers want to implement work requirements, and that is that states which have chosen not to expand Medicaid up until this point may actually expand their programs if they’re allowed to impose work requirements.

BEN SOMMERS: So Virginia was the first state that passed a Medicaid expansion bill with a bipartisan compromise that said, “We’ll expand, but we’re also going to do so under the requirement that we put in a proposal to do this work requirement policy.” So the governor there, a Democrat, said, “We want to expand.” The legislature led by Republicans said, “Only with work requirements,” and then that’s the current reality.

So there’s the ideological question about what you believe should or shouldn’t happen as a prerequisite for health care. There’s the policy argument of, well, is this going to work? Is it going to do what it said, are we going to get more people working, are we going to get people into private coverage? But then there’s also kind of a strategic or a tactical political discussion, which is even if you oppose work requirements both on ideological and policy grounds, is it a compromise you’re willing to live with if it gets Medicaid expansion passed in your state? And that’s something we see supporters of expansion grappling with in several of the states that haven’t yet expanded but are considering doing so. And I hope our evidence can shed some light as to what the real world consequences are going to be. But ultimately a lot of this this is about a political calculation of what advocates think they can get done with or without a compromise.

SHANOOR SEERVAI: What do you think, Ben, about the idea that maybe it’s worth swallowing the bitter pill of work requirements if it leads to expansion? Let’s say from the evidence in Arkansas.

BEN SOMMERS: This will be a partial cop-out, which is that I’m a policy researcher and I can tell you more about the evidence than I can tell you that I have any special perspective on how we value these trade-offs. I think it’s certainly the case that if you look at Arkansas and, say, compare it to a state like Texas which has never expanded, low-income adults in Arkansas with Medicaid expansion and work requirements are still in the aggregate much better off in terms of affording health care than in a state that’s never expanded at all. But I think there are thoughtful critics of work requirements that say, “This is fundamentally eroding what the program is supposed to do.”

SHANOOR SEERVAI: And one of those thoughtful critics is a federal judge who has stopped the program from running in Arkansas, am I correct?

BEN SOMMERS: That’s right. So this is Judge Boasberg, a federal court judge who heard the cases about Arkansas. Kentucky, which was actually the first state to have its work requirements proposal approved, but the judge blocked that from ever taking effect shortly before it was supposed to be implemented in 2018. So Arkansas got to go ahead with its program for that first round before the judge then came back and blocked continuing implementation. And then most recently, New Hampshire had started to implement its work requirements program in this year, and the judge put that one on hold as well. And in all three of the cases, the judge’s ruling essentially said, “Medicaid is not designed by Congress to be a work program. It’s not designed to help people become more self-reliant or improve their health or their subjective well-being. It’s designed to help people pay their bills. And if you kick people out of the program because they’re not working, you are undermining the legislative intent, what Congress meant to do when it created the Medicaid program and when it expanded it under the Affordable Care Act.”

So as long as that ruling holds, it’s going to be hard for states to go forward with any program like this where there is a risk of coverage loss. Now of course that ruling is being appealed; it could be overturned. And it’s also possible down the road that Congress comes back and decides, “Well, no, we actually do like the idea of work requirements. We’re going to pass a law that clarifies this is an okay use of Medicaid policy dollars and state flexibility.”

So I think in either case, we still want to know based on the evidence we have how the policy was working.

SHANOOR SEERVAI: Right. And so three states that we’ve already discussed, Arkansas, Kentucky, and New Hampshire, tried to implement their work requirements programs and they’ve been stopped from doing that. But how many other states are interested in adding this to their Medicaid program?

BEN SOMMERS: So, a bunch of states actually, more than 10 have applied or are in the process of applying to the federal government for permission to do this. Not all of them have been accepted. But there is a lot of interest in a bunch of these states, some which have expanded, some which haven’t, and they’re all taking slightly different tacks on it. Some have longer hours that are required, some are using more lenient reporting where you only have to report every few months. What remains to be seen is, if they do take effect, whether these subtle differences make a big difference or not in the outcomes. Part of what we found in Arkansas was it wasn’t the details of 80 versus 70 or 90 hours of work a month. It was simply that people didn’t know about it and weren’t reporting. Because like we started out with the finding we were talking about earlier, if 95 percent or 97 percent of people are already doing these things, it’s not the details of what qualifies and what doesn’t, it’s how easy the system makes it for you to keep your coverage and how much red tape is there.

SHANOOR SEERVAI: Before we wrap up, I want to talk about these 17,000 or 18,000 people and what you hope to track as you continue your research in the state and what could happen to them now they don’t have coverage.

BEN SOMMERS: Right, so the original design of the Arkansas policy was that you would lose your coverage if you didn’t report, and it was until the end of the calendar year. Once the calendar turned over to January 2019, all of those individuals who had been disenrolled from the program in theory could reapply, and as long as their income hadn’t changed, they could enroll. The state has pointed out that only about 10 percent or 11 percent at the time of oral arguments had actually come back.

Most of those people in early 2019 had not signed up again for Medicaid. And they argued that this was evidence that people probably had found jobs and had found new coverage. That’s certainly one interpretation. Combining that with our data, what you’d say is, “Well, those are people who hadn’t gotten new jobs or new coverage by the end of 2018 but maybe over the following couple of months they were able to.” But another interpretation is that, well, half of the people were having trouble reporting, and one out of three didn’t even know about the policy. So it may be that people just didn’t understand that they were eligible to come back in, or hadn’t been able to complete that process.

Right now we can speculate only. It could be a good sign; it could also be a very bad sign.

SHANOOR SEERVAI: Are there any other final takeaways that you have from the research you’ve done so far?

BEN SOMMERS: I mean, this is one state — Arkansas does not resemble every other state in the U.S. They had particular features of the way they were doing this that might not be the same in other states, and we only have six months of evidence on this because, you know, the last round of disenrollment was the end of 2018 and then the courts blocked it a couple months later.

So we can’t say that this is the definitive result that’s going to apply to every other state or long term in Arkansas. What we can tell you is that, to our knowledge, this is the first real quantitative evidence we have, kind of an independent assessment of how the program’s working. And at the very least, I think it should be raising some caution and some concerns about whether this policy’s going to be able to accomplish its stated goals and whether the potentially unintended consequences of it are going to outweigh that hoped-for benefit.

SHANOOR SEERVAI: And I guess one of the underlying things that has sort of come out through this conversation is that, of course there’s this idea of work requirements, and we’ve gone into detail on whether or not those are a good thing or a bad thing, but it sounds like reporting, even if other states choose to do it differently, will always prove to be a challenge. Whenever you add red tape, bureaucracy, a form that people have to fill out, to their lives, you make them more complicated.

BEN SOMMERS: That’s right. And to the extent that states have always had to reverify people’s income eligibility and other factors over time, we know from a lot of research, well before the Affordable Care Act and well before work requirements, that people move in and out of Medicaid, and often do so unintentionally because they have been unable to navigate the bureaucracy. And I’m a primary care doctor when I’m not — I do policy research and I see patients. And I see this in my clinic.

I see a patient who’ll come in who I haven’t seen in a while and I’ll say, you know, “What happened, I haven’t seen you in several months,” and they said, “Well, I went to the pharmacy last week and they told me that my insurance expired. I hadn’t realized that. And so I hadn’t been able to get an appointment,” or, “I hadn’t been able to fill my medications.” And this is happening regularly in every state in the country because Medicaid is, by design, this kind of patchwork approach where it’s not guaranteeing you coverage long term. So because of that, this is always a challenge, and work requirements seems to have layered an additional level of complexity and red tape, and in Arkansas, at least in that first year, we are seeing the harms that that can cause.

SHANOOR SEERVAI: All right, well, we’ll keep watching for what happens in Arkansas and other states, and I just want to thank you for joining me on the show today.

BEN SOMMERS: My pleasure.

Show Links:

Guest Bio: Benjamin Sommers
Arkansas’s Medicaid Work Requirements Contributed to Higher Uninsured Rate and No Change in Employment

Illustration by Rose Wong


Publication Details



Shanoor Seervai, Former Researcher, Writer, and Lead Podcast Producer


Shanoor Seervai, “What Happened When One State Made Having a Job a Requirement for Medicaid,” Sep. 6, 2019, in The Dose, produced by Joshua Tallman and Shanoor Seervai, podcast, MP3 audio, 28:31.