From 2013 to 2017, Arkansas’s Medicaid expansion helped increase the number of insured Arkansans by 223,000. Then in June 2018, Arkansas became the first state to implement a program, Arkansas Works, to terminate Medicaid coverage for expansion enrollees who fail to report having worked at least 80 hours per month for three or more months. This unprecedented policy, authorized by the Centers for Medicare and Medicaid Services (CMS) under a Medicaid Section 1115 waiver demonstration, is likely to reverse Arkansas’ significant progress in expanding coverage. And prior research has shown that work requirements do not improve people’s health or cause substantial or lasting increases in earnings.

According to our analysis (see below), work requirements would cause the number of Arkansans who lose Medicaid over a year to roughly double. We project that, over a 12-month period, 30,700 to 48,300 adult beneficiaries (19% to 30% of the target population) would lose coverage as a result of not meeting work requirements, out of about 160,000 adults subject to the requirements next year (see exhibit). Arkansas mandates that beneficiaries both complete and report work activities monthly unless they are exempt; our projected losses include those not meeting either component of the requirement. Some additional beneficiaries would lose Medicaid coverage for unrelated reasons, such as because they move, fail to reapply, or have changes in income or family structure.

These are conservative estimates. Additional losses could occur because thousands of adults may be discouraged from applying because of the burdens of the new policy. Analyses of prior work requirements in welfare showed that much of the caseload losses occurred because the policies depressed applications.

Fluctuating Work Patterns Put Low-Income Adults at Risk

Some people who initially comply with work requirements become noncompliant later, particularly because of fluctuations in how much they work. Employment can be volatile among those with low incomes because of issues like seasonal work and inconsistent work schedules. In addition, exemptions for full-time students, medical frailty, or other reasons end after varying lengths of time if people don’t renew them. Those who do not meet the requirement for three months within a year lose coverage. Among those initially compliant, the risk of not meeting the work and reporting requirement in each successive month is 5.6 percent.

We present low and high estimates of projected Medicaid coverage losses (see exhibit). The low scenario assumes losses stabilize after the fifth month; that is, those who are employed or exempt after the first few months maintain that status for the remainder of the year. The high scenario, which seems more plausible, assumes that the losses continue to mount because some (5.6% per month) who met the requirements in previous months fail to do so in subsequent months. Data showed that 58 percent of those noncompliant in the initial month were noncompliant in the next two months and lost coverage; the remaining 42 percent were able to comply for at least some subsequent months or left for other reasons. We also estimate that 42,400 would lose Medicaid for reasons unrelated to work requirements, based on an exit rate of 3.5 percent per month. Therefore, the work requirements would roughly double the number people losing Medicaid over 12 months, compared with losses that would occur in the absence of the work requirements. This result is comparable to recent estimates of the effects of work requirements for Kentucky.

Although Arkansas is initially applying work requirements to expansion beneficiaries with incomes below 100 percent of the federal poverty level, CMS permitted the state to broaden the requirements to those with incomes up to 138 percent of poverty. If the state does so, even more people would lose coverage.


In Arkansas, Medicaid work requirements will lead 19 percent to 30 percent of the target population — or 30,700 to 48,300 adults — to lose coverage over a year. Moreover, fundamental questions exist about whether CMS properly permitted the approval of work-requirement demonstration projects. The first work requirement approved by CMS, for Kentucky, was halted by a federal court order and a court challenge regarding approval of Arkansas’ project has been filed. Depending on the outcome of the new case, Arkansas might need to suspend its work requirements.

This analysis shows how the loss of insurance coverage can mount over time and eventually lead thousands of impoverished Arkansans to lose Medicaid. While the majority of adult beneficiaries work, many are unable to do so consistently over time and could lose their health coverage as a result. Loss of coverage not only will reduce their access to care, it will create financial hardships for safety-net facilities, such as community health centers that rely on Medicaid revenue. The evidence from Arkansas indicates that work requirements will lead thousands to lose Medicaid coverage, without improving their earnings or their health.


We developed a simple survival-type model based on monthly data reported by the Arkansas Department of Human Services during the phase-in of Medicaid work requirements for expansion beneficiaries ages 30 to 49 from June to September 2018. Our model begins with a cohort of the estimated 160,000 adults ages 19 to 49 who will be subject to work requirements when it is fully phased-in. Each subsequent month, some exit because of noncompliance with work requirements or for other administrative reasons. This enables us to estimate caseload losses over 12 months as shown in the exhibit.

In the first month of phase-in, of the 25,815 Medicaid adults who had to meet work requirements (or have an exemption), 7,464 adults, or 29 percent, did not comply, whether because of work or reporting. Not complying for three months leads to termination for the rest of the calendar year. After three months, 4,853 were terminated because they were noncompliant. This indicates that among the 7,464 who did not report work or an exemption the first month, 58.3 percent were terminated for noncompliance while 41.7 percent became compliant or withdrew from Medicaid for other reasons. (Another 4,019 individuals were terminated for not meeting the work requirement in October.) Thus, in the first simulated month, 29 percent are noncompliant (i.e., do not report working 80 hours or more or being exempt for reasons such as medical frailty, schooling, or being in a drug treatment program), and 58.3 percent of those will go on to lose coverage.

After the initial month, data from Arkansas indicated that some (5.6%) of those who were compliant earlier became noncompliant in the second month. Our high estimate indicates that this risk of becoming noncompliant will continue at this rate throughout the year, while the low estimate assumes that those who meet or have an exemption from the requirement for the first two months will continue to meet the requirement for the remaining 10 months. As above, we estimate that 58.3 percent of those who become noncompliant will lose coverage because of the work requirement. All estimates assume that 3.5 percent of the population leaves Medicaid for other reasons each month, based on reported declines in the target population for reasons unrelated to compliance with the work requirement each month.

In the first four months of the program, work requirements had a substantial and consistent impact on beneficiaries in the target population. For example, accounting for closures because of work requirements, the overall rate of those not meeting the requirement remained between 27 percent and 29 percent in each of the four months.