Research Insights Medicaid and Personal Responsibility Waivers: Opportunities and Challenges in Evaluating Potential Impacts Introduction Until recently, however, work requirements were not permitted Initially, after the Supreme Court made the Affordable Care Act in Medicaid, and cost-sharing provisions were tightly limited. But (ACA) Medicaid expansions optional to the states, only about half since 2017, with interest in expansion persisting in many initially chose to take advantage of the opportunity. But given the opportu- opt-out states and a conservative Republican administration in the nity to improve the health of state populations, and to take advan- White House, waiver restrictions have been substantially relaxed by tage of generous increases in federal matching funds, many more the Centers for Medicare and Medicaid Services (CMS). have since followed suit. By December of 2018, the total number of expanding states had risen to 37. Through Medicaid’s waiver On September 7, 2018, AcademyHealth convened a conference of process, the states have also enjoyed increased flexibility in how about 30 policy analysts and public officials to review and dis- they can manage the program, which some have used to promote cuss research related to the new Section 1115 expansion waivers what are collectively referred to as “personal responsibility” policies, and to identify the most pressing research needs going forward. such as increased cost-sharing, work requirements, and incentives What follows is a synthesis of the presentations and discussions at to encourage healthy behavior. the meeting, along with relevant background from the gray and peer-reviewed literature. The discussion was off the record, so this Medicaid and other safety-net programs have a long history with brief will review the comments largely in paraphrase, and with- similar initiatives. Work requirements were an essential ingredient out attribution. A blog post describing the meeting discussion is of welfare reform in the 1990s, and pre-ACA waivers under Section also available at https://www.academyhealth.org/blog/2018-09/ 1115 of the Medicaid statute allowed experiments with small premi- experts-examine-evidence-medicaid-and-personal-responsibility- ums and other cost sharing features that were otherwise prohibited requirements. under the original legislation in 1965. Genesis of this Brief: This brief is based on a meeting of policymakers and researchers that took place in Washington, D.C., on September 7, 2018. AcademyHealth convened the meeting as part of its Research Insights Project. Funding for the conference was made possible by Grant No. 2R13HS018888-07 from the U.S. Agency for Healthcare Research and Quality (AHRQ). The views expressed in written conference materials or publications and by speakers and moderators do not necessarily reflect the official policies of the U.S. Department of Health and Human Services; nor does mention of trade names, commercial practices, or organizations imply endorsement by the U.S. Government. The Research Insights Project convenes invitational meetings, holds webinars, and produces reports and issue briefs to foster discussion of existing, relevant research evidence among policy audiences that need to implement health reform and develop new policy. Additional information and publications may be found on the project’s website at http://www.academyhealth.org/about/programs/research-insights. Medicaid and Personal Responsibility Waivers: Opportunities and Challenges in Evaluating Potential Impacts Background ciaries’ transition to private coverage. Not all Medicaid rules are Letters from CMS to state officials in early 2017 charged that exist- waived. Premiums and other cost sharing can’t exceed five percent ing Medicaid rules were “rigid and outdated,” and that the expan- of family income. Some groups can’t be charged: poor and near- sion provisions threatened to divert the program from its core poor children, pregnant women below 150 percent of the federal mission and drive up state and federal spending.1 CMS promised poverty level, the medically frail, and some other documentable a new era, empowering the states with increased freedom to tailor hardship cases. Income-based sliding scales must be used. But program designs to meet the various states’ unique needs. The letter participants at the meeting emphasized also that the reasons for emphasized new efforts to increase employment among enrollees, including premiums in many waiver proposals was that “getting to incentivize more prudent use of resources by consumers of care, yes” in many state negotiations would have been impossible without and more fully align Medicaid with private insurance, particularly them. “We can’t ignore the politics,” said one presenter. by increasing the use of cost sharing, high-deductible coverage, and savings accounts. The 1115 proposals vary on multiple dimensions: premium amounts, enrollee income levels, grace periods, lockout provisions By October of 2018, five states had received at least initial approval – which prevent enrollees who miss payments or other require- from CMS for their 1115 expansion waiver proposals, and one, in ments from re-enrolling for a specified period of time. Some Arkansas, had gone into effect. Implementation of the first approved states allow premium discounts for healthy behavior or receiving waiver, in Kentucky, had been blocked in federal court. Approval was preventive services. A cap of 2 percent of annual income on these pending in ten other states. All varied widely in design details, pro- obligations is common, as are 60-day grace periods. Non-payment gram goals, and the political circumstances that brought them forth.2 penalties such as lockouts tend to be stiffer for income groups above 100 percent of the federal poverty level (FPL). The variability Observers noted substantial challenges in implementing the waiver of state policies will make it difficult to evaluate the net impact of proposals. They would entail modification of eligibility procedures, premiums on enrollment, health outcomes, and state budgets. new systems to document compliance, outreach efforts to inform potential enrollees of program changes, interfaces with other state There is, however, extensive experience with premiums and other and federal program, staff expansions and training, new appeal pro- forms of cost sharing in Medicaid prior to and outside of 1115 expan- cedures, and more. Upgrading and retooling information technol- sion waivers. By 2018, 30 states charged premiums or enrollment fees ogy (IT) capacity would be needed in all these efforts.3 for children in Medicaid and the Children’s Health Insurance Program (CHIP). While Medicaid prohibits premiums for those with incomes The CMS guidance letters explicitly cautioned the states about below 150 percent of poverty, eight CHIP programs have waivers to the need to ensure compliance with Medicaid’s overarching legal charge premiums or enrollment fees for children from 133 to 150 per- framework. But in June, a federal district court blocked implemen- cent of poverty. Additionally, copays and coinsurance may be charged tation of Kentucky’s CMS-approved waiver, which entailed a work for some services to patients between 100 and 150 percent of the FPL. requirement, on the grounds that it was inconsistent with Medic- Above 150 percent also, there is no limit for non-emergency use of the aid’s statutory purpose of providing medical assistance to the state’s Emergency Department up to Medicaid’s overall limit of 5 percent of citizens. Shortly thereafter, Arkansas released estimates that nearly income for those above 150 percent of poverty.4 20 percent of otherwise eligible low-income people would lose coverage for failing to document their 80-hour-a-month obligation Thus there is a considerable evidence base for evaluating the po- or prove their exempt status. tential impact of premiums on the new Medicaid expansion waiver programs. Systematic reviews of the research literature on these ef- In roughly the order followed at the meeting, this brief will ad- fects show a consistent downward impact on enrollment when pre- dress premiums, health savings accounts (HSAs) and other forms miums are added or increased. The effects are larger on low-income of cost sharing, work requirements, and healthy behavior incen- groups, although the magnitude of effects varies across studies. tives in Medicaid; and finally cross-cutting issues of implementa- tion and evaluation. Premiums in public programs are also associated with increases in uninsurance and private insurance enrollment, although the Premiums magnitude of these effects varies across studies. To date, little The policy rationale for premiums, in both CMS guidance and state evidence is available on the impacts of premiums for adults who are waiver proposals, is to offset state spending on services, increase eligible for Medicaid expansion coverage under the Affordable Care enrollees’ cost consciousness, and help bring Medicaid into closer Act above and below the federal poverty level. Studies have found alignment with private insurance – hopefully smoothing benefi- that premiums have a negative effect on access to care, but down- 2 Medicaid and Personal Responsibility Waivers: Opportunities and Challenges in Evaluating Potential Impacts stream health effects are much more difficult to assess and have yet In the first iteration of the Healthy Indiana Plan (HIP), starting in to be adequately documented. Overall, findings on the negative 2008, Medicaid expansion enrollees were required to contribute five effect of premiums on enrollment in Medicaid are consistent with percent of their income to POWER accounts to pay for up to the the landmark RAND Health Insurance Experiment (HIE) of the first $2,500 in care received in a year. Twenty percent of the contri- 1970s. Also in line with the HIE, a relatively recent study found that butions were subsidized by a state cigarette tax.7 Subsequently the “the premium requirement itself, more so than the specific dollar state split the plan, with different choices for enrollees in different amount, discourages enrollment.”5 income groups. Those above 100 percent of the FPL were still re- quired to contribute to their POWER account. Two levels of cover- Administrative costs associated with new premium requirements age were offered – HIP Basic and HIP Plus, with the latter including will be another important concern for state officials and will need vision and hearing benefits and no point-of-service copays. Those careful monitoring and analysis. States can expect to realize savings in HIP Plus who miss payments are dropped to the lower level or from anticipated disenrollments. But new information systems will disenrolled entirely for specified periods. be needed to process payments, whether online, with credit cards, or in cash. Achieving robust enrollment will require investments in The complexity of the Indiana program created not only admin- outreach and education. New procedures will be needed for non- istrative challenges for the state, but barriers for potential Indiana payment enforcement and associated appeals processes. Similarly, enrollees. In one study, for example, 39 percent of eligible beneficia- exemption applications have to be managed, as well as appropriate ries had not heard of POWER accounts, and just 36 percent were due-process appeals processes. making the required contributions. So nearly two-thirds of expan- sion enrollees were at risk of losing benefits or coverage because of The net financial effects of premium revenue, disenrollment, and inadequate communications.8 administrative costs will bear close watching. States have estimated that net administrative costs of expansion waivers, including other Given that many low-wage jobs are temporary or have irregular features such as work requirements and HSAs, may in some cases hours, workers’ incomes fluctuate, causing frequent changes in run into hundreds of millions of dollars.6 eligibility status. So Indiana workers may often shuttle between HIP Basic, HIP Plus, lockouts, uninsurance, and private coverage in or HSAs and other forms of cost sharing out of state Marketplace plans. The impact of this churn in cover- In addition to premiums, there are a variety of cost-sharing age status on access to care and health outcomes needs attention, mechanisms among the new waiver-seeking states, including conference participants agreed. those modeled on HSAs. States experimenting with HSAs tout their value in part as a pathway to familiarizing Medicaid enroll- High administrative costs and low participation rates swamped ees with how private insurance works. But the accounts, untaxed savings in another HSA experiment in Arkansas, and the state shut in private markets, were designed principally for use in conjunc- the program down in 2016. The Healthy Michigan plan (HMP) has tion with high-deductible coverage plans that appeal primarily to had better success engaging enrollees with its savings accounts. As higher-income consumers. elsewhere, Michigan’s HSA is embedded in a spectrum of associ- ated cost-sharing mechanisms. Among HMP’s more than 600,000 In Medicaid, the savings accounts are designed to help patients meet enrollees, those with incomes between 100 and 133 percent pay point-of-service copays, co-insurance, and/or deductibles. They may 2 percent of their income to enroll in one of about a dozen exist- also represent a way to create incentives for healthy behaviors, such ing, established Medicaid managed care organizations in the state, as completing health risk assessments (HRAs), in return for account although available choices vary significantly by region. contributions. A further policy rationale is to promote consumer price sensitivity, although the effectiveness of this strategy is undermined Their payments are credited to their MI Health Account, in by a pervasive lack of transparency in service pricing. But for low- amounts calculated by income data and family characteristics. income Medicaid enrollees, shopping may not matter when expensive Copays are not collected at the point of service. Rather, an aver- specialty services are out of reach at any price. age monthly co-pay is calculated for the beneficiary on a quarterly basis. As part of a CMS-mandated evaluation, a survey of more The accounts entail monthly contributions by enrollees, although than 4,000 enrollees reported that 68 percent of respondents said requirements may also be met with employer or state contributions. they received a statement and 88 percent reviewed it carefully, and Indiana pioneered the use of HSAs in Medicaid with its Personal that it helped them to be aware of the costs of care. Similar shares Wellness and Accountability (POWER) accounts, as part of a pre- of respondents agreed that the modest amounts that they paid were ACA 1115 coverage expansion waiver. fair and affordable. 3 Medicaid and Personal Responsibility Waivers: Opportunities and Challenges in Evaluating Potential Impacts On the other hand, most (85.7 percent) believed or were unsure Similar requirements have been common since the passage of if they would be disenrolled from HMP for not paying their bill. welfare reform in 1996, for example in the Temporary Assistance Over one-quarter (28.1 percent) of respondents were aware that for Needy Families (TANF) program and the Supplemental Nutri- they could get a reduction in the amount they have to pay if they tion Assistance Program (SNAP). In both cases, various outcomes completed a health risk assessment, engaged in specified healthy of work requirement programs have been analyzed, so some of behaviors, or received preventive services.9 Best practices identified their general effects are understood. Evaluations of the TANF in the AcademyHealth discussion included frequent and clear com- welfare-to-work experience have shown modest positive effects on munications between providers and enrollees, use of grace periods employment that may last up to about five years, along with some and other measures to minimize disruptions of access. gains in employer-sponsored health coverage.10 Reductions in the receipt of welfare benefits tend to be substantial, so the net effect The stated goal of some of the expansion waiver provisions is to on family income appears to be negligible. Effects have varied increase enrollees’ familiarity with private insurance, an environment widely across states. in which cost sharing has been increasing steadily for a decade or more. A few states have been approved to offer premium assistance In a discussion of labor market factors, participants agreed that the to Medicaid enrollees who join a qualified health plan in their state concept of personal responsibility needed to be reconciled with the Marketplace or from their employer. In general, co-pay limits have reality that enrollees have no control over demand for labor, but drifted upward under expansion waivers, but some states have tried to such provisions would penalize enrollees when employment oppor- be strategic with them, waiving cost sharing for preventive services, for tunities are limited. Currently, with five work requirement waivers example, or increasing them for over-used or low-value services. at least initially approved and more pending, the labor market en- vironment appears to be favorable, with 50-year lows in unemploy- Many questions remained about the use of premiums, copays, ment reported in September 2018. But the national average masks HSAs and other cost-sharing mechanisms for participants in the local variation, particularly in poor communities where receipt of September meeting. Some surveys suggest that some low-income Medicaid is concentrated. And many of the jobs contributing to the enrollees would be willing to pay at least a little more to keep their current trend are of the low-wage, temporary, and part-time variety coverage, but it is unclear whether this signals a rational economic that may be difficult to fit with the typical 1115 requirement of 80 calculation about the value of care. The net effect of complicated documented hours per month.11 waiver programs on state budgets is also uncertain, since some en- rollee contributions come in small increments, which entail dispro- Over the first summer of the Arkansas Works 1115 waiver program, portionately high administrative costs. Premiums and enrollment more than 4,350 of 26,000 Medicaid enrollees subject to the new fees tend to reduce enrollment, but the strength of this effect varies. requirements failed to meet them and lost coverage. In Arkansas, as elsewhere, many eligible enrollees are employed, but often in tem- Discussants suggested that state policymakers sometimes seem to porary or part-time jobs, and many others are exempt for various be unaware of disconnections between the design details of their reasons.12 Arkansas requires monthly documentation of compli- 1115 programs and the professed goals of expanding coverage while ance to be filed online, although about a fourth of those affected are increasing personal responsibility, as when administrative costs estimated to lack Internet access, with much higher rates in some exceed savings, or some forms of cost sharing produce sharp drops localities.13 Members of a federal oversight panel expressed alarm.14 in enrollment. The role of research in pinpointing such disconnects and clarifying choices for policymakers was an explicit takeaway In Kentucky, the first state to receive CMS approval for a work re- from the discussion. quirement, a federal judge blocked implementation of the proposal after state officials estimated that 95,000 low-income people would Work requirements lose their Medicaid coverage if the state’s plan went into effect. The The biggest and most controversial change in expansion waivers judge found that federal officials who approved the plan had not after January 2017 has been CMS approval of several state proposals seriously considered its potential impact on the state’s provision of to include a work requirement for Medicaid enrollees, a feature that health services for eligible residents as Medicaid law requires. all previous administrations had rejected. At least five states want work conditions for the new category of enrollees, and a few others Work requirements in Indiana, New Hampshire, and Wisconsin want them for their traditional populations. Most proposals require have been approved and are due to start in 2019. Ten more states a documented 80 hours a month of employment, training, or job have work requirement proposals pending, including six in non- seeking. Some enrollees are exempt due to disability, care-giving, or expansion states. The Arkansas plan has also been challenged in other priorities. court, with a focus on the adequacy of the state’s online system 4 Medicaid and Personal Responsibility Waivers: Opportunities and Challenges in Evaluating Potential Impacts for processing enrollees’ monthly compliance information. As of Healthy behavior incentives September 2018, the Arkansas application also lacked an approved Healthy behavior incentives are a feature of several expansion evaluation plan as required by law.15 waiver proposals that fit under the rubric of personal responsibil- ity. They have a long history in workplace wellness programs and Crucially, Medicaid does not offer dollars to support employment a more recent history in Medicaid. But changing behavior is a efforts, for such needs as childcare and job training, in contrast to notoriously difficult endeavor, and evidence about the effectiveness work requirement programs in TANF and SNAP. State resources for of these efforts in saving money and promoting health is mixed. funding such supports vary, but are often quite limited. Programs vary widely, data collection is not standardized, and outcomes may take many years to show up. Attributing causality to Estimates of the overall cost of implementation of work require- any single factor is inevitably confounded by multiple co-factors. ments vary, but are substantial: Clear-cut evidence of the effectiveness and cost-effectiveness of healthy behavior incentives is limited to tightly targeted programs • Alaska, $79 million over six years with measurable outcomes built in. • Kentucky, $186 million • Michigan, $15 million to $30 million annually Research on workplace wellness has yielded many useful insights into behavioral psychology and economics. Especially after the advent of • Minnesota, $163 million in 2021 self-funded employee health coverage in the 1970s and 1980s, employ- • Ohio, $378 million over five years ers saw an opportunity to reduce their health care spending by pro- • Pennsylvania, $600 million moting healthy behavior and preventive care. They offered incentives for workers to enlist in weight-loss and smoking-cessation programs; • Tennessee, $34 million a year to receive preventive services such as mammograms, and cholesterol • Virginia, a range from $200 million a year with “high touch” and blood pressure screenings; and to monitor and manage chronic case management, and $7 million without it.16 conditions like diabetes and hypertension with regular testing and The states, of course, can expect that some of their increased admin- provider visits. Incentives might include cash payments, reductions in istrative costs will be offset by declines in enrollment and thus also employees’ premium shares, perks like gym memberships or parking in spending on benefits. The net effect of these budgetary impacts privileges, or simply tokens of recognition. then needs to be measured against the potential harms to the af- fected population and local providers and to factor in other offsetting The effectiveness of these different incentives was found to vary government spending related to rising uncompensated care burdens. widely. Cash rewards – the larger the better – worked better than The level of risk in turn depends on the details of each states’ plans and in-kind perks. Quick disbursement – monthly, say, rather than policies, and details of its implementation practices. quarterly or annually – improved effectiveness. Ease of use, as with electronic communication tools, made a difference. Overall, educa- Enforcement of penalties for non-compliance and procedures for tion, engagement, and communication, along with clarity and sim- establishing exemptions were cited at the AcademyHealth meeting plicity, emerged as essential. Several factors distinguish workplace as particularly critical factors for determining enrollment effects. IT capacity is another. Economic stress, low education levels and programs from Medicaid, including that the workplace constitutes a literacy, limited transportation and internet access, and cultural contained environment, providing multiple channels and opportu- barriers often confront the communities that rely on Medicaid, nities for managers to facilitate program participation. leaving them vulnerable to coverage losses where strict disenroll- ment and lock out policies and practices are implemented. Trans- Prior to the ACA, CMS had approved at least a dozen healthy portation, documentation standards, lockout policies, and grace behavior programs, dating back as far as 2006. The Medicaid Incen- periods will also interact to affect net enrollment changes. tives for Prevention of Chronic Disease program gave ten states a total of $85 million, starting in 2011. Six programs targeted smok- The states’ experiences with supported welfare-to-work programs, ing, another six diabetes, five obesity, four hypertension, and three primarily for mothers, may have only limited value as a template high cholesterol. All the programs had rigorous evaluation com- for Medicaid’s complicated requirements, the conferees agreed. “It’s ponents incorporated in their design. Overall results were mixed. a really different world,” said one. Concerns of special importance Some increased use of preventive services was found in three to the research community are the design, rigor, and data require- diabetes programs. In three smoking projects, calls to quit lines and ments of the states’ evaluation plans. Political considerations have cessation counseling increased. Minimal effects on body weight figured prominently in the waiver process, highlighting decision were observed, and few changes in hospital use were seen. Three makers’ need for relevant and reliable research on program out- programs registered net savings, but two others showed losses. comes. 5 Medicaid and Personal Responsibility Waivers: Opportunities and Challenges in Evaluating Potential Impacts Seven states incorporated healthy behavior components in their Should otherwise eligible recipients be locked out of coverage for 1115 expansion waivers, including Kentucky’s suspended program. six months or a year because they can’t find a pay stub, or lack Use of negative incentives increased. In Iowa, for example, enroll- Internet access, or have poor reading skills? Further, because of ees could be charged premiums for dental benefits if they failed to the concentration of poverty in neighborhood pockets, aggregate complete an HRA. A lack of awareness about the programs among coverage gaps are likely to have community-wide effects on local providers and beneficiaries was found in several instances. In economies and social conditions. The negative effect of reduced Michigan, fewer than 30 percent of eligible enrollees knew that they coverage on provider capacity has also been documented. Com- could receive premium and copay reductions for completing an munity health centers are particularly vulnerable. Routine program HRA in a 2016 survey.17 In Kentucky’s proposal, an extensive range evaluations are unlikely to measure these broader impacts. of healthy behaviors could be rewarded with cash contributions to an enrollee account for dental or vision care. But the accounts Procedures necessary for managing the programs create many chal- could be debited for non-emergency ED use or nonpayment of lenges. Notice and appeal processes must meet statutory standards. premiums. In programs that have been evaluated thus far, program Even without federal help, states must provide beneficiary supports complexity tended to depress participation. that satisfy the Americans with Disabilities Act. Exempt popula- tions must be protected. As noted previously, IT investments will An important insight from prior experience is the value of efforts be substantial and entail design challenges for interfaces with other to inform and engage beneficiaries, which implies a need for “high state agencies, the state Marketplace and its plans, and actual state touch” approaches that necessarily entail higher costs. The effective- residents seeking coverage. As much or more of a burden will be ness of larger rewards also puts upward pressure on costs, highlight- staffing needs to implement the new version of Medicaid, with ing the problem of capturing long-term health benefits and savings some waiver provisions also being extended to previously covered in research evaluations. Much remains to be learned about the trad- populations in several states.18 eoffs between positive and negative incentives – carrots and sticks. Informing and involving providers in healthy behavior programs As in welfare-to-work programs, case management is essential to appears to be a largely unexplored frontier, especially for those with successful transitions. Between hiring and training new workers, fewer Medicaid patients; and incentives for provider engagement Ohio expects to spend $378 million over five years to support case are lacking. Improvements could be made in program design to management. Tennessee’s budget for these needs is $22 million in better the chances of obtaining meaningful evaluation results. the first year of its proposed work requirement. There, as elsewhere, budget pressures are often reducing case management to bare bones Monitoring and evaluation levels. Moreover, many state health and human service departments Crosscutting issues addressed at the meeting included two levels are already understaffed and face high vacancy rates, so that new of thinking about evaluation. One was its importance as a tool for responsibilities are likely to detract from ongoing projects, such as focusing attention on the details of policy and program execu- developing improved payment systems to promote integration and tion, with the need for vigorous data collection as a corollary and better care. Work previously underway on simplifying and stream- an expectation that the states have the will and the ability to make lining enrollment processes will be fundamentally disrupted. CMS mid-course corrections when indicated. A broader standard for resources may also be stretched. weighing outcomes was also considered, which was to judge results on macro effects, or how well the programs meet Medicaid’s overall Evaluation challenges are also testing the states’ mettle, from statutory goal. formulating the right questions to capturing program performance to gathering the data needed to answer them. Provisions must be In the latter respect, concerns were expressed about potential harms made for special populations and the most vulnerable. Those seek- to beneficiaries from the cumulative and longitudinal effects of the ing exemptions from work requirements are the most likely to face coverage disruptions that various provisions of the waivers seem obstacles. Comparison groups have to be constructed to measure likely to entail, on the basis of the research at hand. Coverage inter- program impacts and may require tracking outcomes for popula- ruptions multiply as incomes fluctuate, documentation problems tions in non-expansion states. Care and coverage outcomes for arise, life situations change, and new policies ripple through the those who lose eligibility need to be tracked. Adequate evaluations state infrastructure. Gaps in coverage are likely to be compounded are difficult and expensive to conduct. over time, compromising care and outcomes. 6 Medicaid and Personal Responsibility Waivers: Opportunities and Challenges in Evaluating Potential Impacts Conclusion 5. L Dague, The effect of Medicaid premiums on enrollment: a regression discon- tinuity approach, Journal of Health Economics 37 (2014): 1-12, at https://ccf. The new waiver programs have not been in operation long enoug georgetown.edu/wp-content/uploads/2012/03/Dague-Premiums.pdf for much to be known yet about their outcomes and impacts, but 6. Wagner and Soloman, Complex waivers. eventually evaluations will have a critical role to play. Some confer- 7. S Freedman, L Richardson, and K Simon, Learning from waiver states: coverage ence participants emphasized that these programs operate under effects under Indiana’s HIP Medicaid expansion, Health Affairs (Millwood) 37:6, CMS’s demonstration authority, which means that benefits and June 2018: 936-950. harms from their implementation have to be monitored and mea- 8. BD Sommers et al., New approaches in Medicaid: work requirements, health sav- ings accounts, and health care access, Health Affairs (Millwood) 37:7, July 2018: sured. Evidence presented and discussed at the conference shows 1099-1108. that both kinds of outcomes are possible. 9. SD Goold and J Kullgren, Report on the 2016 healthy Michigan voices enrollee survey, University of Michigan Institute for Healthcare Policy and Innovation, Jan. 17, 2018, at https://www.michigan.gov/documents/mdhhs/2016_Healthy_ The shape of Medicaid’s future depends on how well these tasks are Michigan_Voices_Enrollee_Survey_-_Report__Appendices_1.17.18_fi- performed. nal_618161_7.pdf 10.A Weil and J Holohan, Health insurance, welfare, and work, Brookings Institu- About the author tion, 2001, at https://www.brookings.edu/research/health-insurance-welfare-and- work/ Rob Cunningham is an independent writer in Washington, D.C. 11.B Casselman, Jobless rate sinks again, the lowest in 50 years, New York Times, Oct. 6, 2018. Endnotes 12.A Gangopadhyaya et al., Under Medicaid work requirements, limited internet 1. Letter from Secretary Price and CMS Administrator Verma to the Nation’s access in Arkansas may put coverage at risk, Urban Institute, Oct. 29,2018, at Governors. March 14, 2017. Available: https://www.hhs.gov/sites/default/files/ https://www.urban.org/urban-wire/under-medicaid-work-requirements-limited- sec-price-admin-verma-ltr.pdf internet-access-arkansas-may-put-coverage-risk 2. J Wishner et al., Medicaid expansion, the private option, and personal responsi- 13.A Goldstein, More than 4,300 Arkansas residents lose Medicaid under work bility requirements, Urban Institute, May 27, 2015, at https://www.rwjf.org/en/ requirements, Washington Post, Sept. 18, 2018. library/research/2015/05/medicaid-expansion--the-private-option--and-person- al-responsibil.html 14.R Pear, Advisors sound an alarm after thousands are dropped from Medicaid, New York Times, Sept. 15, 2018. 3. J Wagner and J Soloman, States’ complex Medicaid waivers will create costly bureaucracy and harm eligible beneficiaries, Center on Budget and Policy Priori- 15.S Rosenbaum, Medicaid work demonstration legal developments, Health ties, 23 May 28, 2018, at https://www.cbpp.org/research/health/states-complex- Affairs blog, Sept. 14, 2018, at https://www.healthaffairs.org/do/10.1377/ medicaid-waivers-will-create-costly-bureaucracy-and-harm-eligible hblog20180914.426396/full/ 4. S Artiga, P Ubri, J Zur, The effects of premiums and cost sharing on low-income 16.Wagner, Complex waivers. populations: updated review of research findings, Kaiser Family Foundation, 17.Goold, Survey. June 2017, at https://www.kff.org/medicaid/issue-brief/the-effects-of-premiums- 18.For details, see Approved and pending eligibility and enrollment restrictions and-cost-sharing-on-low-income-populations-updated-review-of-research- (table), Kaiser Family Foundation, Sept. 28, 2018, at http://files.kff.org/attach- findings/ ment/Approved-and-Pending-Eligibility-and-Enrollment-Restrictions 7