What This Paper Covers

Medicaid finances health care coverage and services for more than 75 million Americans. Historically, coverage was targeted primarily to low-income children, pregnant women, the elderly, and people with disabilities. Over the last decade, however, Medicaid enrollment among able-bodied, working-age adults has surged due to the Affordable Care Act’s expanded eligibility rules, as well as lax enforcement of eligibility rules by states. Moreover, federal funding creates incentives for states to favor services for able-bodied, working-age adults over the most vulnerable Americans, shifting the program’s focus away from traditional enrollees and leading to rapid cost growth.

Considering these realities, adding a work requirement for able-bodied, working-age Medicaid enrollees without young dependents could improve the program’s fiscal sustainability while also re-orienting Medicaid access back to the most vulnerable patients. A common critique of work requirements is that such rules may worsen population health by causing people to lose health insurance and delay or forego needed medical care. With nationwide work requirements for certain Medicaid enrollees slated to take effect in 2027, it is important to consider the potential impacts of this policy.

This study represents the first direct empirical evaluation of the effects of Medicaid work requirements on access to care and health. We focus on Arkansas, which enforced work requirements in 2018 and 2019.

What We Found

Because Arkansas’s work requirements lasted for nearly a year and impacted a substantial number of people, the state provides a unique opportunity to study the actual impacts of these policies on access to care and health. Using detailed hospital records data, we compared rates of preventable hospitalization and preventable emergency department visits (e.g., cases of uncontrolled diabetes or asthma) between adults in Arkansas subject to the work requirements and adults exempt from the rules due to age cutoffs. We find no statistically discernable effects of work requirements on access to care or health.

Why It Matters

Our findings suggest that work requirements in Medicaid do not harm population health, at least in the short-run. This is consistent with evidence from randomized trials that insurance coverage has little or no effect on physical health outcomes. Our results also align with a large literature documenting that work itself is associated with improved health. Work requirements may be an effective tool for encouraging upward economic mobility while prioritizing health resources for the neediest Americans.

Medicaid finances health care coverage for more than one in five Americans, including 40 percent of children.¹ However, as Paragon has demonstrated extensively, the program is overdue for reform. One major problem—exacerbated by the Affordable Care Act—is that states have weak incentives to ensure proper enrollment.² This has led to nearly $1.1 trillion in improper payments over the past decade.³ Another problem is that states manipulate Medicaid’s federal funding formula through provider taxes and other financing gimmicks, which minimize their actual state contributions and inflate federal costs.⁴

Medicaid was intended to help vulnerable populations, such as individuals with disabilities, children, and pregnant women. But its funding structure now discriminates against these populations. Federal incentives have driven states to favor care for able-bodied, working-age Medicaid enrollees over the neediest Americans. Considering these realities, adding a work requirement for able-bodied, working-age Medicaid enrollees without young dependents could improve the program’s fiscal sustainability while also re-orienting Medicaid access back to the most vulnerable patients.

However, a common critique to requiring work and community engagement is that such requirements will worsen population health by causing people to lose health insurance.⁵ For example, the Center for American Progress recently claimed that a Medicaid work requirement could result in 15,400 deaths annually.⁶ However, the association between insurance coverage and health is tenuous at best,⁷ with randomized trials suggesting no causal link between insurance coverage and health.⁸ Moreover, a large literature suggests that work itself is associated with improved health.⁹ Yet the potential health impact of Medicaid work requirements has not been directly studied.

Arkansas provides a useful real-world example to examine the potential health impacts of Medicaid work requirements. In 2018, Arkansas enacted a requirement that certain Medicaid enrollees participate in community engagement, defined as employment for at least 80 hours per month or participation in other activities such as volunteer work or enrollment in a job training or educational program.¹⁰ Exemptions were granted for medically frail individuals and pregnant women. The work requirements went into effect on June 1, 2018, for persons 30-49 years old and were subsequently extended on January 1, 2019, to individuals ages 19-29. The work requirements for both age groups were rescinded on March 27, 2019, after a federal court ruling in Gresham v. Azar.¹¹

Because Arkansas’s work requirements lasted for nearly a year and impacted a substantial number of people, the state provides a unique opportunity to study the actual impacts of these policies on health. As a first step toward doing so, we examined trends in preventable hospital admissions and emergency department visits in Arkansas using the State Inpatient Databases and the State Emergency Department Databases. Both datasets are maintained by the Healthcare Cost and Utilization Project,¹² which is part of the Agency for Healthcare Research and Quality (AHRQ). In addition to maintaining these datasets, AHRQ has also developed a methodology to identify preventable emergency room visits and hospital admissions.¹³ This methodology uses diagnostic codes in the data to identify emergency department visits or hospital admissions that could have been prevented with better access to primary care. For example, admissions for diabetic ketoacidosis or asthma exacerbations are classified as potentially preventable admissions.¹⁴

These metrics provide important insights into the effects of work requirements. If work requirements lead to large-scale coverage loss and a decline in primary care utilization, more people may delay or forgo routine procedures and treatments, resulting in an increase in medical emergencies.

Figure 1 shows the monthly rates of preventable hospital admissions and emergency department visits in Arkansas between January 1, 2018, and February 29, 2020. We chose February 2020 as the cutoff month because the first case of COVID-19 was reported in Arkansas in March 2020,¹⁵ and COVID-19 had significant impacts on both hospital admissions and emergency department visits.¹⁶ Preventable emergency department visits are defined as visits meeting the criteria for AHRQ measures PQE01 (Visits for non-Traumatic Dental Conditions), PQE02 (Visits for Chronic Ambulatory Care Sensitive Conditions), PQE03 (Visits for Acute Ambulatory Care Sensitive Conditions), or PQE04 (Visits for Asthma).¹⁷ Preventable hospital admissions are defined as admissions meeting the criteria for PQI90, a composite measure that includes admissions for several conditions such as complications of diabetes and heart failure.¹⁸

Figure 1 shows preventable admissions and emergency department visits for persons ages 19-29, for whom the requirements took effect on January 1, 2019. Figure 2 shows preventable admissions and emergency department visits for persons ages 30-49, for whom Arkansas’s work requirement took effect on June 1, 2018. The requirement ended for both groups in April 2019. If work requirements had worsened access to care, we would expect to see an increase in medical emergencies during the time the requirements were in effect for both groups. But as Figures 1 and 2 show, there is no clear increase. Indeed, the rate of emergency department visits was somewhat lower during the period where the work requirements were in effect (-4 visits/100,000 people/month; 95% CI -19 to 11; p=0.18), as was the rate of hospital admissions (-1 visit/100,000 people per month; 95% CI -8 to 6; p=0.48), although neither difference was statistically significant.

This simple comparison is a useful starting point, but it may not capture the causal effect of work requirements on health outcomes. For example, it does not adjust for possible factors such as seasonal changes in visits and admissions (e.g., more people may seek care during flu season) as well as unobservable factors that might drive visits and admissions (e.g., the population potentially becoming sicker over time). A more accurate approach would exploit the fact that the work requirements applied to different age groups at different points in time. As noted above, the work requirements impacted 30- to 49-year-olds between June 2018 and April 2019, but people age 50 and older were exempt from the requirement. Thus, adults over 50 serve as a natural comparison group for adults under age 50. Specifically, trends can be compared between people just on one side of this cutoff (e.g., 47-49 years old) to people on the other side of this cutoff (e.g., 50-52 years old). This approach would first examine changes in avoidable admissions among people 47-49 years old before, during, and after the work requirements were in effect. It would then compare these changes to changes in the 50- to 52-year-old group, who were never impacted by the work requirements, using this latter group as a control for unobservable factors (e.g., seasonal changes). This approach, known as difference-in-differences,¹⁹ is often used to study the impacts of public policies, such as changes in health programs and minimum wage laws.²⁰

The basic idea of the difference-in-differences approach is shown in Figures 3 and 4, which plots preventable hospital admissions (Figure 3) and emergency department visits (Figure 4) for 47- to 49-year-olds and 50- to 52-year-olds. If work requirements had an impact on preventable admissions, we would expect to see a marked increase in visits/admissions among 47- to 49-year-olds during the time frame the work requirements were in effect relative to any observed change in 50- to 52-year-olds observed during this time period. For both measures, however, we fail to see this relative increase, suggesting that the work requirements had no effect in these outcomes.

We formalized this difference-in-differences analysis through a linear regression using the following age groups: 16-18, 19-21, 27-29, 30-32, 47-49, and 50-52. Each of these groups was chosen as they are age groups located right around the age cutoffs for the work requirements, which impacted 30-to 49-year-olds from June 2018 to December 2019, and 19- to 49-year-olds from January 2019 to April 2019. Our results show that work requirements were associated with a statistically insignificant 0.2 percent increase in the rate of preventable hospital admissions (95% CI -15 to 18; p=0.95) and a statistically insignificant 2 percent increase in the rate of preventable emergency department visits (95% CI -3 to 8; p=0.35).

Our results of null health effects of the work requirements hold when we consider an alternative definition of the treatment and control groups based on two-year age groups (e.g., 19- to 20-year-olds) rather than the three-year age groups we use in our main analysis.²¹ In addition, one concern may be that even if the work requirements did not increase the total number of preventable admissions or visits, they might increase the number of preventable admissions or visits from uninsured individuals, as people formerly covered by Medicaid may have become uninsured. However, our results found that this was not the case, as work requirements were associated with a statistically insignificant 7 percent increase in preventable emergency department visits (95% CI -7 to 23; p=0.24) and a 10 percent decrease in preventable admissions (95% CI -32 to 19; p=0.41) among uninsured individuals.

Arkansas instituted Medicaid work requirements on some working-age, able-bodied adults for roughly 10 months from mid-2018 to early-2019. We found no statistically significant effects on health and access to care during this period, as measured by admissions for preventable emergency department visits and hospital admissions, for the age group subject to the work requirements. In addition, our analysis suggests that there were no increases among uninsured individuals for preventable emergency department visits or hospital admissions.

Although Arkansas provides a real-world example of work requirements, we note that an analysis focused on Arkansas alone has some limitations. First, by definition, the analysis is of the impact of work requirements as enacted by that state, which may differ from how other states might enact similar requirements, as well as from current plans to extend the requirements more broadly. Second, because the work requirements were in effect for slightly less than a year, our analysis captures only the short-term impacts of work requirements. Third, our sample sizes may limit our ability to detect small changes in population health. However, on the whole, our analysis suggests that work requirements are unlikely to cause large decreases in population health and that the most likely scenario is that Arkansas’s work requirements had minimal impacts on population health.

Finally, we did not consider other possible outcomes. For example, the Arkansas work requirements have been criticized for imposing significant logistical burdens on Medicaid enrollees.²² Although we did not examine this issue, it is worth noting that technology has advanced markedly (e.g., the growth in artificial intelligence) since these requirements were implemented in 2018 and that currently there are a variety of independent vendors and external apps that could more easily verify compliance with work requirements.

There are many reasons why the work requirements may have had little, if any, impact on access to care and ultimately on health. First, previous research has shown that the relationship between health insurance and health is generally weak.²³ While insurance has some impact on health care utilization, people are generally less price-sensitive when it comes to health care compared to other goods. Second, even if the work requirements caused people to lose insurance, those who lost insurance may have tended to be healthier to begin with. Indeed, as previously noted, the work requirements included exemptions for medically frail individuals, and the lack of overall impact on population health suggests that the work requirements may have been successful in preserving health care access for vulnerable patients, who may have been more likely to either meet the requirements or find other insurance sources.

Overall, Arkansas provides a useful case study on the potential health impact of Medicaid work requirements, and the lesson to be learned is that their effects on health may be insignificant, at least in the short-term period.