Abstract
- Issue: The Centers for Medicare and Medicaid Services approved Medicaid work requirement demonstration projects in four states, and other states also have applied. However, the future of these projects has been clouded by legal and policy challenges.
- Goal: To assess whether state Medicaid work requirement projects are designed for success in promoting employment among unemployed Medicaid beneficiaries.
- Methods: To examine the design of new work requirement projects, we reviewed the evidence, analyzed the overlap of Medicaid and Supplemental Nutrition Assistance Program (SNAP) work requirements, and convened a roundtable of seven experts who have research or implementation experience with work programs for Medicaid and public assistance recipients.
- Findings and Conclusion: Mandatory work programs would be less effective and efficient than well-administered voluntary programs. Far more people will be subject to Medicaid work requirements than are currently subject to them in SNAP. This surge could overwhelm the limited resources of existing employment training and support programs. Medicaid demonstration projects contribute almost no additional funding to train the unemployed or provide necessary social supports. Medicaid work requirement programs are not well designed to help people get jobs or improve health and are more likely to lead to a loss of health insurance coverage.
Background
In April 2018, President Donald Trump ordered several federal agencies to enforce existing work requirements for Medicaid and develop new work programs.1 Work requirements have been in effect in the Temporary Assistance for Needy Families (TANF) program and Supplemental Nutrition Assistance Program (SNAP) for many years. This is the first time they have been applied to the Medicaid program. In January 2018, the Centers for Medicare and Medicaid Services (CMS) solicited state proposals for Medicaid work requirements under Section 1115 demonstration project authority, with the stated goal of improving health by increasing beneficiaries’ employment.2 CMS rapidly approved four state projects (Arkansas, Indiana, Kentucky, and New Hampshire); several additional state proposals have been submitted or are in planning. Architects of these proposals expect that the requirements will help low-income unemployed adults gain employment and reduce dependency. Work requirements address the concern that Medicaid discourages adults from working,3 although evidence from the Oregon Health Insurance Experiment indicates that Medicaid does not significantly harm employment status.4
In June 2018, a federal court struck down CMS approval of Kentucky’s project, preventing its implementation. The decision is likely to be appealed.5 CMS then opened a new round of public comments on Kentucky’s project; most comments opposed it.6
Arkansas implemented its mandatory work project in June 2018; it continues despite a court challenge.7 Affected beneficiaries who do not report that they worked or were exempt for three months lose coverage for the rest of the calendar year. Among the first 25,800 beneficiaries who have been subject to the requirement for three months, more than 4,500 lost Medicaid coverage.8 Those who lost coverage may have been unaware of the new requirements and could have had work or an exemption that they did not convey to the state.9 More recent analyses indicate that over 12 months, this could cause 31,000 to 48,000 Arkansans to lose Medicaid.10
This brief addresses the design and administration of Medicaid work requirement programs and their potential to increase employment. It focuses on two areas: whether mandatory work requirement programs are an effective way to boost employment, and if the current employment and training infrastructure can support a major influx of Medicaid beneficiaries newly subject to work requirements. To examine these issues, we convened a roundtable of experts with both research and implementation experience with work programs for Medicaid and public assistance recipients; see How We Conducted This Study.
Findings
Under CMS’ “community engagement” policy, states may require adult Medicaid beneficiaries to work or be engaged in work-related activities, unless they are elderly, disabled, or pregnant. States are permitted to let work-related activities like job search, training, or volunteering count for part of the requirement; Arkansas, for example, allows 39 hours of job search activities to count toward the 80-hour monthly work requirement.11 Those who do not comply will lose Medicaid coverage and could be barred from reentry for some period. While CMS encouraged states to help beneficiaries gain job skills, it does not allow Medicaid funds to be used for job training or related support services, such as transportation or child care. CMS instead recommended that Medicaid dovetail with existing work programs, like TANF or SNAP. Exhibit 1 compares federal Medicaid, SNAP, and TANF work requirements.
The basic theory behind Medicaid work requirements — conditioning benefits on employment — has been tested before and has not been shown to improve long-term economic well-being. A recent White House report stated that research shows that welfare-to-work programs led to improved employment and earnings.12 But research suggests that successful programs included components of job training, education, and earnings supplements.13 They require intensive resources to help develop job skills, but such resources are not required by federal policies. Moreover, independent analyses of the same studies cited by the White House found that while welfare-to-work programs may help expedite employment in the short run, they do not meaningfully improve beneficiaries’ long-term employment or earnings14 or improve their health.15
Early analyses have examined the potentially harmful effects of Medicaid work requirements, particularly the loss of health insurance coverage and access.16 Early data from Arkansas, the only state that has implemented Medicaid work requirements thus far, demonstrate the validity of those analyses.18 A key example was Indiana’s Gateway to Work program, originally a voluntary program. Between March 2015 and January 2016, Indiana sent more than 300,000 letters to Medicaid recipients informing them about the Gateway to Work program, but only 551 work orientation visits took place.19
However, evidence from Montana indicates that voluntary programs can succeed in promoting work without a threat to Medicaid coverage. The difference may be found in design and implementation. Montana adopted a successful voluntary program (HELP-Link) for newly eligible Medicaid beneficiaries. Recruitment was incorporated into enrollment processes; new enrollees are asked to complete an assessment of employment status and barriers to work. Staff select beneficiaries for outreach based on their circumstances. They contact them by telephone to suggest a personalized assessment and make an appointment to develop an individual employment plan. The state provides services such as job search, credit assistance and on-the-job training where appropriate.20
During the first six months, among the 94,000 Montana residents newly enrolled in Medicaid, this targeted, personalized approach led 22,000 to receive job services from the state and 2,500 to receive one-on-one employment training services, according to Jessica Rhoades, who oversees HELP-Link. Among those who completed training, 91 percent found work.21 These levels are particularly noteworthy given that four-fifths of Montanans who receive Medicaid already live in working households. Montana has experienced a 9 percent increase in employment among the nondisabled Medicaid population and a 6 percent increase in employment among Medicaid enrollees with disabilities.22
In contrast, CMS does not require Section 1115 work requirement states to notify beneficiaries about work services in person or by telephone or to focus resources on those most likely to be helped. For example, although Arkansas has taken steps to inform beneficiaries of the new requirements, many continue to be unaware of the changes.23 One telephone outreach effort attempted to reach beneficiaries who likely needed to report their work effort; only about 20 percent were successfully contacted.24
Mandatory programs are an inefficient use of limited resources.25 Rather than targeting resources more appropriately, everyone is treated equally: those who need only a little help (because they are educated and have job skills and transportation) and those who need more help (because they have serious employment barriers and very limited skills or experience). The administrative resources needed to track participation for all beneficiaries are substantial; case workers often spend significant time tracking recipients’ work-activity hours for compliance with SNAP and TANF requirements.26 Montana’s more carefully designed and implemented voluntary program appears to be more successful in helping the unemployed find jobs. The voluntary program allows limited resources to be spent on helping people address barriers to work, rather than on tracking compliance. For example, existing workforce programs can be used for tuition support for community college classes.27
Finally, Medicaid work requirements might not actually lead people to look harder for work. A recent study found that among Kansans potentially affected by a work requirement, only 11 percent said it would have any effect on their job search.28 Nearly all Medicaid beneficiaries who could work were either already employed or seeking employment.29
Medicaid work requirements are not resourced to help people find employment; however, they are likely to lead to substantial loss of health insurance coverage. When work requirements have been imposed on SNAP or TANF participants, many quickly lost benefits.30 Adverse publicity could deter eligible people from even applying to Medicaid, which has occurred with TANF participants in the past.31
In Arkansas, a substantial fraction of the initial group affected by work requirements has already lost coverage. Arkansas is phasing in work requirements for those ages 30 to 49 in 2018 and those ages 19 to 29 in 2019. In each of the first three months, more than one-quarter of the targeted population did not comply with the requirement.32 In September, 17 percent (4,353) of the initial 25,815 people required to meet or demonstrate an exemption from the work requirement reached three months of noncompliance and lost coverage. As more people are targeted by the project, the number who lose coverage will swell. Around 167,000 will be required to meet or demonstrate an exemption by mid-2019.33
More People Potentially Targeted by Work Requirements
CMS policy greatly expands the target population of low-income adults who could be required to demonstrate that they either comply with or are exempt from work requirements for one or more benefits that they receive. Medicaid work requirements would apply to a much larger target population than SNAP or TANF requirements for several reasons. Medicaid has more enrollees than SNAP, and while CMS permits Medicaid to include adults with children, SNAP excludes parents. CMS lets Medicaid include a broader age range, up to age 64. People who have not met requirements are most at risk. But those who are working or potentially exempt also could lose coverage if they fail to report this information.
Exhibit 2 illustrates the number of adults who are in the target population for Medicaid and/or SNAP work requirements, as delineated by federal guidelines, in the four states currently approved by CMS (Arkansas, Indiana, Kentucky, and New Hampshire). In Medicaid, these are nonpregnant, nondisabled adults ages 19 to 64. The SNAP requirement applies to childless, nondisabled adults ages 19 to 49. There are 512,000 people who potentially are subject to SNAP work requirements: 273,000 enrolled in both Medicaid and SNAP, and 239,000 who participate in SNAP alone. Some 1.55 million adults potentially are newly subject to work requirements for Medicaid, three times more than the number who already are subject to SNAP work requirements. This surge in volume is likely to strain the capacity of the employment and training infrastructure.
Our analyses describe those who could be required to work under federal criteria, since states may establish broader exemptions in Medicaid or may seek or end SNAP waivers within the framework of federal rules. After adjustments for existing state criteria, the number of adults newly subject to Medicaid work requirements still exceeds the number subject to SNAP work requirements.34
Too Few Resources for Effective Employment and Training Services
Despite the falling unemployment rate, Medicaid beneficiaries who currently are unemployed may need assistance in finding employment or overcoming barriers to work. An estimated 74 percent of Kentuckians and 78 percent of Arkansans who are not currently working and will be required to meet Medicaid work requirements face several challenges. They either lack a high school diploma, do not have access to either a vehicle or internet in their household, or are affected by their own or a family member’s serious health limitation.35 Other potential barriers include caring for dependents, older age, or a criminal record. Furthermore, unlike SNAP work requirements, Medicaid work requirements would not be suspended when unemployment is high.
Employment and training services have limited resources to assist people in addressing these and other barriers to job search and employment. Existing resources will be stretched over a much larger pool of people in states that implement Medicaid work requirements. Russell Sykes, a former program human services administrator in New York, noted:
Funding is critical to conduct proper assessment and assign clients to appropriate activities such as education and sector-based training. . . . Funds are critical to actual job placement and helping clients navigating problems both before and after getting a job. . . . Expanding work requirements in other programs will be very resource dependent, if the goal is truly to get clients into jobs instead of simply imposing sanctions.36
As CMS prohibits Medicaid funds from being used for training or support services, it is reasonable to question where these resources will come from.
The cost of successful employment and training services varies, but clearly is much higher than currently available resources allow. Some people with prior job skills, a good education, and no major job-hunting barriers may need little help. Those dealing with major skills deficits or serious barriers may need far more help, including supportive services. The cost of TANF work programs in a median state is $414 per month per participant.37 A study found that well-tested job skill programs cost from $7,400 to $14,200 per participant.38 Although hundreds of thousands of Medicaid recipients could be subject to work requirements, no substantial investment has been made to help them find work.
Without sufficient funding, states are unable to provide adequate services for unemployed beneficiaries. Offering little else but low-intensity services, such as job search, is unlikely to be successful. A recent assessment of employment and training, conducted for SNAP, concluded that providing a large mandatory population with low-touch services such as job search is unlikely to increase employment very much.39 Successful strategies include individualized assessments, addressing barriers to employment, and helping participants gain skills and experience; all are more expensive to implement than low-touch services.
Employment and training programs are primarily financed through limited federal funds, with relatively little additional state funding. Three main sources of employment and training funding for public assistance clients are TANF, SNAP’s Employment and Training program, and Workforce Innovation and Opportunities Act (WIOA) programs. Each has very limited funding and already is overburdened.
TANF can support employment and training, child care and other services, and cash assistance. However, TANF funding has not increased since 1996, and only 9 percent is allocated for employment services, with another 3 percent for supportive services such as transportation or behavioral services.40 SNAP served 21 million households in 2017. Only $438 million was available for its employment and training program, or about $20 per household.41 WIOA programs support training referrals, career counseling, job listings, and similar employment-related services. However, WIOA services address the needs of diverse target groups, and resources are spread thin. Funding for 2019 WIOA job training grants is 22 percent lower than in 2010 after adjusting for inflation.42
The dearth of funding available for employment services inevitably limits the ability of Medicaid work requirements to offer serious help to beneficiaries. States pursuing Medicaid work requirements have allocated little funding to meet the new demands for services. Kentucky planned to use $5 million in state funds for Medicaid recipients, as well as $3.5 million drawn from other federal workforce funding for employment services. However, local workforce administrators noted that much more was needed.43 The $8.5 million in employment service funding is equal to about $4 per affected person per month. In comparison, because CMS permits Medicaid funds to be used for data systems, a large portion of the $374 million Kentucky has allocated to implement work requirements will be spent on technology to track compliance.44 Similarly, Arkansas budgeted $6.8 million to upgrade computer systems to track compliance, but less than $1 million for other functions, mostly provided by federal matching funds.45 Early analyses of Arkansas’s work requirement program show that job training programs were lacking and that few Medicaid enrollees subject to the new rules were getting help in seeking employment.46
Conclusions
A fundamental question is whether work requirements are designed to help beneficiaries improve their opportunities or are structured to sanction beneficiaries in a bid to lower Medicaid program caseloads. Current Medicaid work requirement programs, poorly designed and resourced, are unlikely to help many people find work or become healthier. Instead, the initiatives are more likely to reduce Medicaid rolls and leave more people without health insurance coverage. To address employment barriers in a meaningful way requires careful planning, design, implementation, and funding. All these elements are currently lacking in mandatory Medicaid work requirement programs.
Rather than allocating available resources to mandatory work requirement administration, funding might be better spent on initiatives to improve health, such as reducing opioid or tobacco usage, or on well-designed voluntary employment opportunities. Mandatory work requirements are more likely to reduce beneficiaries’ insurance coverage and access to health care than improve their earnings or health.
How We Conducted This Study
Analyses of Overlap of Medicaid and SNAP Work Requirements
The analyses shown in Exhibit 2 are based on the 2016 American Community Survey, focusing on Medicaid and SNAP participants ages 19 to 64. Those who were disabled or pregnant were excluded based on receipt of Supplemental Security Income or having a child in the past year. The estimates were done for both the four approved Medicaid states (Arkansas, Indiana, Kentucky, and New Hampshire) and the overall United States. Results are similar for the four states and the nation, although the ratio of Medicaid to SNAP participants is somewhat larger in the four approved states. This is because SNAP eligibility is established at the federal level, all four currently approved states chose to expand Medicaid, while many other states did not. The analyses also found that roughly one-half of SNAP and Medicaid recipients already work 20 hours or more per week and may already meet work requirements. However, many could still lose coverage if their work hours fluctuate or they are unable to document their work status.
The actual number of adults subject to SNAP requirements and state-level Medicaid work requirements will depend on several factors. These include whether SNAP requirements have been waived in some areas, and on additional Medicaid criteria adopted by states (for example, exclusions for parents with dependent children, more restrictive age limits, or education exemptions). Recent analyses by Yale University researchers comprehensively applied state work requirement criteria. An estimated 814,000 adults are in the target populations subject to work requirements in Arkansas, Indiana, Kentucky, and New Hampshire, according to research based on 2014 Survey on Income and Program Participation data.47
Expert Roundtable
On May 30, 2018, we convened a panel of experts to help us examine implementation issues related to work requirements and employment and training programs. The experts have extensive experience implementing and/or studying human services or health policy, allowing us to translate experience gained in work programs in SNAP and TANF to the Medicaid context. We integrated the lessons learned from this meeting with our review of the scholarly and grey literature on work requirements in Medicaid and social service programs.
Experts who participated in the roundtable were:
- Heather Hahn (Urban Institute)
- MaryBeth Musumeci (Henry J. Kaiser Family Foundation)
- LaDonna Pavetti (Center on Budget and Policy Priorities)
- Jessica Rhoades (Health Policy Advisor to Gov. Steve Bullock, Montana)
- Sara Rosenbaum (George Washington University)
- Judith Solomon (Center on Budget and Policy Priorities)
- Russell Sykes (American Public Human Services Association)
In addition, Rachel Nuzum of the Commonwealth Fund participated in the meeting. Project staff from George Washington University included: Leighton Ku, Erin Brantley, Brian Bruen, Erika Steinmetz, and Drishti Pillai.
Acknowledgments
We gratefully acknowledge contributions of our expert roundtable members and other knowledgeable experts such as Jenny Kenney of the Urban Institute, Jessica Greene of Baruch College, and Hannah Katch of the Center on Budget and Policy Priorities. We also thank Sara Collins, Chris Hollander, and their colleagues at the Commonwealth Fund for support and guidance.