Automatic Enrollment in Health Insurance: A Pathway to Increased Coverage for People with Low Income

Abstract

• Issue: The number of uninsured Americans remains stubbornly high, and many Americans do not obtain the coverage for which they are eligible — even when insurance is free.
• Goal: To outline a system for automatically enrolling people with low incomes in health coverage and then model the impact on coverage and spending at the federal and state levels.
• Methods: The Urban Institute’s Health Insurance Policy Simulation Model was used to analyze alternative auto-enrollment approaches. These include identifying uninsured individuals who are tax filers or recipients of the Supplemental Nutrition Assistance Program (SNAP), unemployment insurance, or Social Security and enrolling people who are eligible for free coverage.
• Key Findings and Conclusions: We show results for the nation and for three states (California, Georgia, and Michigan). If all states adopted an auto-enrollment policy for those with incomes at or below 150 percent of the federal poverty level, 4.3 million uninsured people would be identified and enrolled. An additional 1.8 million would be “deemed” covered, either auto-enrolled through provider contact or contingently covered and thus protected from huge medical bills. Provider spending on uncompensated care would fall 32 percent, while federal spending would increase by $30.3 billion and state spending would increase by $7.7 billion per year.

Introduction

Of the over 26 million uninsured people in the United States, more than half (52%) are eligible for subsidized coverage through Medicaid or the Affordable Care Act’s (ACA) marketplaces but do not enroll. Low-income individuals account for a large share of these uninsured. Under current law, people with income below 150 percent of the federal poverty level (FPL) make up 43 percent of the uninsured, and those below 200 percent of FPL make up 56 percent (for a single person in 2024, these income levels translate to $21,870 and $29,160, respectively). Enrolling people who are eligible for zero-premium coverage through Medicaid or the ACA marketplaces could reduce the number of uninsured substantially. In principle, many of these individuals not facing premiums could be auto-enrolled. Auto-enrollment could be enacted nationwide through federal law or by individual states, either with or without federal changes facilitating state action.

In previous work, we looked at the design and impacts of policies that would extend auto-enrollment for the entire U.S. population.¹ As a second option, we looked at expanding auto-enrollment to those who participate in the Supplemental Nutrition Assistance Program (SNAP) or the Temporary Assistance for Needy Families (TANF) program. Auto-enrollment for the entire population would be a huge and complex policy initiative. In this report, we focus on auto-enrollment only of people eligible for zero-premium coverage through Medicaid or the ACA marketplace. We do not anticipate this proposal would apply to those excluded from program eligibility for immigration reasons, but it could be if eligibility restrictions were eased.

How Auto-Enrollment Might Work

The Inflation Reduction Act of 2022 eliminated marketplace premiums for eligible people with incomes below 150 percent of FPL. As a result, all individuals at that income level who live in states that have expanded Medicaid can now receive zero-premium coverage, unless they have an “affordable” offer from an employer or are excluded because of immigration status.² If the remaining 10 states expanded Medicaid, or if eligible people with incomes below 100 percent of FPL in nonexpansion states were permitted to have subsidized coverage in the marketplace, then most legal residents with incomes below 150 percent of FPL could be covered with no premiums.³ Zero-premium coverage could also be extended to higher incomes, either through further enhancement of federal premium tax credits or by states adopting a Basic Health Program or adding state premium subsidies.⁴

In our model, individuals would be auto-enrolled in Medicaid or the ACA marketplaces depending on their eligibility. A designated administrative agency would identify and enroll individuals after confirming their income and citizenship eligibility using available third-party data. A state’s health insurance marketplace could take on this responsibility since these entities already can determine eligibility for both marketplace coverage and Medicaid. Or a state Medicaid agency or a new agency could oversee this process.

We estimate the impact of two auto-enrollment pathways:

1. Enrolling individuals identified as uninsured and eligible for zero-premium coverage based on information from other interactions with government agencies. For this analysis we consider interactions like tax filing and receipt of SNAP, Social Security, and unemployment insurance benefits, which provide key information needed for eligibility determinations and enrollment like name and other identifying data, address and other contact information, family members, and income. Individuals must also typically provide immigration status to qualify for Medicaid and marketplace coverage. The designated administrative agency could verify these eligibility criteria using the Federal Data Services Hub and other third-party data sources. The agency would then confirm coverage status, enroll eligible individuals, and give them an opportunity to opt out or switch plans.
2. “Deeming” individuals below designated income levels to be covered even if they did not enroll. If a low-income person went to a hospital or other provider for care, the provider could, following a basic initial screen for immigration and income eligibility, contact the designated administrative agency to enroll the patient, and would then be paid for the services provided. The designated agency would verify eligibility (based on income, citizenship, and other criteria) using third-party data sources like the Federal Data Services Hub and then enroll the patient in Medicaid or a low-cost marketplace plan. Also, at any point in the year, an eligible individual could contact the agency to enroll. Either way, the coverage would reimburse services back to the start of the year, like an enhanced version of Medicaid retroactive coverage. Those who are deemed eligible but do not seek care would be contingently covered and thus protected from the financial risk of uninsurance.

We further expect that auto-enrolled people would be rescreened as their enrollment nears an end and reenrolled if they are still eligible. Individuals could opt out, but this seems unlikely given the lack of premiums. Continuous coverage for a year, which was nationwide Medicaid policy during the COVID-19 public health emergency and continues in some states, could be one possible option.

Policy Considerations for Auto-Enrollment

Successful auto-enrollment would require implementation of at least five policies:

• Lowering the employer coverage firewall. Under the ACA, people with an offer of employer coverage deemed “affordable” are ineligible for marketplace subsidies (but not for Medicaid). These individuals lack access to zero-premium plans, so auto-enrollment would not apply to them. As a remedy, the firewall could be waived for workers with incomes low enough to qualify for zero-premium coverage.
• Creating an aggressive, ongoing, and well-funded education and enrollment assistance campaign. The policy would need to create awareness that all individuals below 150 percent (or 200 percent) of FPL are effectively insured through Medicaid or marketplace plans, and that even those who did not actively choose a plan are covered. A major administrative effort would be needed to enroll individuals, either those who seek active enrollment or those who are referred by a provider.
• Eliminating the clawback of advance premium tax credits for the population provided free coverage. Currently, recipients must agree to receive those credits in advance and to reconcile them when they file their income taxes. Reconciliation could be waived for auto-enrollees and others who are eligible based on their incomes.
• Offering a special enrollment rule for marketplace coverage. Marketplace enrollment is normally permitted only during the year-end open enrollment period and designated special enrollment periods for circumstances like loss of coverage or new eligibility for subsidies. A special enrollment rule could be established to allow people who are eligible based on their income to enroll in a plan at any time.
• If auto-enrollment were to be adopted nationwide, the Medicaid gap in nonexpansion states would need to be filled by expanding Medicaid or by providing zero-premium marketplace coverage. Currently, adults with incomes below 100 percent of FPL are largely ineligible for subsidized health care in the 10 nonexpansion states, so auto-enrollment in those states would not benefit many people without this change.

These policies could be enacted nationwide through federal law or by individual states, either with or without federal changes facilitating state action. (See box, “Federal and State Legislative Approaches to Auto-Enrollment,” for a discussion of these options.)

The availability of a public option (a government-sponsored plan intended to increase competition in the marketplace) would also make auto-enrollment easier, though it is not necessary, and we do not assume one here. Such a plan could potentially charge premiums below those of other insurers, which would lower the costs of the proposal. It could act as the default plan for auto-enrollees who are not eligible for Medicaid. Without a public option, individuals in the marketplace would have to be enrolled into one of the lowest-cost marketplace plans. There would be issues related to assignment of individuals to plans, plan capacity to serve additional enrollees, and adequacy of plans’ provider networks. These issues are solvable but could be avoided with a public option in place.

The Potential of Auto-Enrollment to Offset Costs

An auto-enrollment policy has the potential to improve the average health risk and reduce the average premium in the marketplace by bringing in people with low demand for health care services, compared to existing enrollees. By auto-enrolling people who did not sign up for coverage on their own, average costs and premiums could fall. Some of these individuals have no medical spending in a given year. For example, in 2021 more than 15 percent of the nonelderly population had no medical spending.⁵ By auto-enrolling at least some of these individuals, their low costs to the system could offset some of the costs of existing enrollees.

We anticipate that the number of people not actively enrolling in insurance coverage would decrease over time. Many people who are directly enrolled after use of services would actively reenroll. Personal experience, education campaigns, and knowledge disseminated through news outlets and social contacts would teach people the advantages of active early enrollment to receive no-cost coverage. These efforts could encourage individuals who would have to pay premiums to enroll as well, and fewer people might delay enrollment over time.

Four Options for Auto-Enrollment

We modeled four reform options that build off the Medicaid expansion baseline, including estimates for the 10 nonexpansion states, assuming they would expand before initiating auto-enrollment (see Appendix 1 for coverage and spending effects for those states).⁶ For all the options, we assume the firewall and advance premium tax credit reconciliation would be eliminated for everyone in the auto-enrollment income range, that a special enrollment rule would be provided for the same group, and a robust outreach campaign would be in place. We present national results under the assumption that all states adopt auto-enrollment and highlight results for selected states, as this policy could be an option for states seeking to reduce uninsurance. The four reform options are:

1. Auto-enroll eligible uninsured people with incomes below 150 percent of FPL if they are identified through tax filings or receipt of SNAP, Social Security, or unemployment insurance benefits. This group would have zero-premium coverage, and the employer coverage firewall would be removed for people with incomes below 150 percent of FPL.
2. Same as option 1 but deem all people with incomes below 150 percent of FPL and who are eligible for zero-premium coverage as insured even if they are not identified through program participation and auto-enrolled. Thus, when an individual visits a provider for services, the provider would receive payment if they refer the patient to the designated agency for enrollment. Any services received earlier in the year also would be paid. Other eligible individuals would be financially protected even if they have not actively enrolled.
3. Extend zero-premium coverage to 200 percent of FPL. As in option 1, people identified through tax filings or program participation would be auto-enrolled, and the employer coverage firewall would be removed below 200 percent of FPL.
4. Same as option 3 but deem all people with incomes below 200 percent of FPL and who are eligible for zero-premium coverage as insured even if they are not identified through program participation and auto-enrolled.

Details on data used and modeling methods are presented in Appendix 2.

Eligibility for Subsidies and Auto-Enrollment

Before gauging the effects of various auto-enrollment policies, we first estimated how expanding Medicaid in every state and eliminating the employer coverage firewall would affect the number of uninsured people eligible for free coverage. With Medicaid expanded in every state, 13.4 million uninsured people would be eligible for Medicaid or ACA subsidies (Appendix Table 3.1). Of these:

• 6.0 million have incomes below 150 percent of FPL and are eligible for coverage with no premiums
• 1.3 million more have incomes between 150 percent and 200 percent of FPL and would be eligible for zero-premium coverage under reform options 3 and 4
• 6.2 million have incomes above 200 percent of FPL, so would not be eligible for zero-premium coverage or auto-enrollment under these reforms.

Eliminating the employer coverage firewall would enable 1.2 million people with incomes below 150 percent of FPL and 6.5 million people with income between 150 percent and 200 percent of FPL to become newly eligible for subsidies (Appendix Table 3.2). Most people who would gain eligibility are currently covered by employer-sponsored insurance (ESI), but some currently pay full premium for nongroup coverage, and 1.1 million are currently uninsured and could be auto-enrolled under reform. In addition, some of the 6.4 million people covered by ESI who would gain subsidy eligibility under reform options 3 and 4 would likely drop ESI for zero-premium coverage.

Tax filing or government program participation could help identify the eligible uninsured. Below 150 percent, tax filing reaches 41 percent of the uninsured, and other programs extend the reach to 71 percent. Tax filing reaches 90 percent of the uninsured with incomes between 150 percent and 200 percent of FPL; other programs bring the total reached to 93 percent for that group (Exhibit 1).

People who are eligible for zero-premium coverage through Medicaid or the marketplace but who are not identified by program participation can be “deemed” insured. If they have nontrivial spending (assumed for modeling purposes to be $250 in uncompensated care) with a provider, we assume they will be enrolled through that provider. If they have only small costs or no costs at a provider, we assume they will remain essentially uninsured but will be financially protected if they get sick (which we call “contingent coverage”).⁷ Over time, the number of deemed-insured people could fall if people who became enrolled through provider contact were automatically reenrolled or if they actively continued coverage. On the other hand, some people might lose coverage for the same reasons they do today, such as problems with paperwork, moving, or changes in income or living situation.

Coverage and Cost Impact of Auto-Enrollment Policy Options

We include the coverage and cost effects not just for people identified through program participation and enrolled in coverage, but also for those who are deemed covered in reform options 2 and 4. Spending for the deemed population is a combination of: 1) spending for those who see a provider, use a significant amount of care, enroll in a plan, and then have all the benefits offered by that plan; and 2) spending for the contingently covered, who use little health services and do not meet our threshold of at least $250, and are never actively enrolled. Our cost estimates are constructed as if all people meeting our spending threshold are insured for the full year.

The contingently covered would be covered in case of significant need for care, but their out-of-pocket costs and spending for uncompensated care on their behalf are estimated as if they were completely uninsured (so are unchanged from the baseline). Existing literature shows that people who are eligible for Medicaid but not enrolled use fewer services than the insured.⁸ However, some people who believe they are uninsured delay care, resulting in more expensive care later on. The sum of spending for both groups represents the expected cost to federal and state government of committing to cover all those deemed insured.

We show results first for our auto-enrollment policies implemented nationwide and then for selected states, starting from the Medicaid expansion baseline.

Auto-Enrollment Results Nationally

Exhibits 2 and 3 show results for a nationwide auto-enrollment policy. This represents a scenario where either the federal government instituted auto-enrollment nationwide or each state adopted its auto-enrollment policy. The results illustrate the potential reach of an auto-enrollment policy.

With current federal subsidies providing zero-premium coverage to people with income below 150 percent of FPL, a national policy that auto-enrolls people identified by program participation or tax filing would reduce uninsurance by 4.3 million people (Exhibit 2). Most of these are enrolled in Medicaid, with a much smaller share gaining subsidized nongroup coverage. Counting those who would be deemed insured, 6.1 million uninsured people would be reached. Even with this reform, significant uninsurance persists. Of the 24 million uninsured in the Medicaid expansion baseline, 15 million have income above 150 percent of FPL and are out of reach of the reform; the remaining uninsured below 150 percent of FPL are ineligible for assistance due to immigration status.

For a policy where all states expand zero-premium coverage and auto-enrollment to 200 percent of FPL, the number of uninsured would fall by about 6.2 million, with most people covered by marketplace plans. If people deemed enrolled were counted, nearly 2 million more individuals would be covered. Those remaining uninsured under the reform have incomes too high to qualify for subsidies or are ineligible because of immigration status.

Eliminating the employer coverage firewall would let some people drop ESI for cheaper marketplace coverage . Under auto-enrollment to 150 percent of FPL, about 0.1 million people would leave ESI, while under auto-enrollment to 200 percent of FPL about 0.8 million would leave ESI (Appendix Table 3.3). The greater impact above 150 percent of FPL stems from available coverage; Medicaid (which would cover most of the below 150 percent of poverty group) has no firewall, while the marketplace (which covers the 150 percent to 200 percent group) has a firewall.

Policies that extend auto-enrollment would increase government health spending, but households and employers would spend less and providers’ costs for care of the uninsured would fall (Exhibit 3). Depending on the income limit of auto-enrollment and whether people not enrolled through program participation are deemed insured:

• Federal health spending would increase by between $27.2 billion and $42.1 billion, as more people receive marketplace subsidies and Medicaid benefits.
• State health spending would increase by $7.0 billion to $11.7 billion. Most of the new spending for people with income below 150 percent of FPL would be through Medicaid, while the additional spending for a policy that extends to people with income between 150 percent and 200 percent of poverty would largely be in marketplace subsidies.
• Households would save between $5.3 billion and $14.9 billion.
• Employers would spend between $0.7 billion and $3.6 billion less on premiums, because some employees would drop ESI for less expensive marketplace coverage.
• Providers’ cost for uncompensated care would fall by 26 percent to 37 percent, because fewer people would be uninsured.

Auto-Enrollment Results for Selected States

The results above would apply if the federal government and all states make auto-enrollment a priority; however, given current political dynamics, only a portion of states might adopt it. We assess the impact for three states: California, Georgia, and Michigan. (Results for all states are provided in Appendix Tables 3.5 through 3.9.)

California

If California adopted auto-enrollment for people eligible for zero-premium coverage (income less than 150 percent of FPL), about one-quarter of the state’s current uninsured population would be covered or protected from large health costs. Extending zero-premium coverage and auto-enrollment to 200 percent of FPL would protect about one-third of this population (Exhibit 4).

Under auto-enrollment to 150 percent of FPL and including spending for people deemed covered, federal and state spending would increase by $4.0 billion, but spending by households and employers would fall by $0.9 billion. If auto-enrollment were extended to 200 percent of FPL, government spending would rise by $5.6 billion, offset by $2.3 billion lower employer and household spending. Consistent with national results, provider spending on uncompensated care would drop dramatically in either case (Exhibit 5; data in Appendix 3).

Georgia

For Georgia to adopt an auto-enrollment policy, it would first need to expand Medicaid, as noted earlier. Medicaid expansion would reduce the number of uninsured by 317,000 (Appendix 1, Exhibit 1). Depending on the auto-enrollment income limit, Georgia would reduce the number of uninsured by between 23 percent and 32 percent relative to the Medicaid expansion baseline (Exhibit 6).

Federal and state spending would rise by $1.3 billion (combined) with auto-enrollment to 150 percent of FPL, including spending for those deemed; with a reform to 200 percent, that spending would rise by $1.8 million. In each case, household and employer spending, and the cost of uncompensated care, would fall (Exhibit 7; data in Appendix 3).

Michigan

Of these three states, auto-enrollment would have the largest impact on the uninsured in Michigan, reducing the number of uninsured by 38 percent with auto-enrollment to 150 percent of FPL and 46 percent with a reform that extends to 200 percent of FPL (Exhibit 8).

Reducing the number of uninsured in Michigan would require a relatively modest increase (between $161 million and $263 million depending on reform) in state spending (Exhibit 9; data in Appendix 3). Federal spending would increase by $0.9 billion to $1.3 billion. Households and employers would spend less, and, as in other states, the drop in uncompensated care spending would be dramatic — about 50 percent under the total enrollment assumption.

Discussion

Expanding coverage and reducing the number of uninsured has been a primary goal of the Affordable Care Act and several subsequent pieces of legislation, such as the Inflation Reduction Act (IRA). At this point, nearly all Americans — with the exception of people in poverty in the 10 states that have not expanded Medicaid and those excluded by immigration status — have a pathway to coverage through employers, Medicare, Medicaid, military programs, or the marketplaces. Substantial efforts have been made to make coverage affordable, but many Americans still do not take up available coverage for any of several reasons, including complex applications, lack of knowledge, and remaining financial burdens. Lack of insurance can have adverse effects on population health and also on the health system that must provide free care. Auto-enrollment of the entire U.S. population has been proposed to expand coverage, but it is a complicated undertaking. Somewhat more feasible is developing a system for auto-enrolling people with lower incomes who have higher rates of uninsurance and are often eligible for coverage without a premium. Auto-enrolling by identifying people with a zero-premium coverage offer under current subsidies (to 150 percent of FPL) increases coverage by 4.3 million, reducing the number of uninsured by 18 percent. Another 1.8 million people would be deemed covered. Of those, 0.5 million would be auto-enrolled through provider contact, while another 1.2 million would be contingently covered and thus protected from huge bills.

If all states added subsidies so people up to 200 percent of FPL had zero-premium coverage, 6.2 million people would gain coverage and another 1.9 million would be deemed insured. Of those, 0.6 million would eventually be enrolled following provider contact, and another 1.3 million would be contingently covered.

Expanding coverage in this way is not extremely costly for states, because the federal government fully finances marketplace subsidies and pays for half or more of Medicaid costs. In addition to covering many uninsured people, auto-enrollment would provide financial protection for providers and perhaps eliminate the need for programs that now support uncompensated care. It also would help with issues faced when low-income individuals transition between jobs or have bouts of unemployment. Some form of coverage would be assured. It would have helped with the problems associated with the loss of coverage during COVID because low-income individuals would be auto-enrolled.

Auto-enrollment would be a major administrative effort. Either the federal government or a state would have to be highly motivated to pursue it. The initial rollout would be a significant undertaking that would require considerable outreach and education to make sure that beneficiaries and providers understand the implications of the policy. It would include the administrative effort of identifying individuals through tax filing, receipt of SNAP, Social Security, or unemployment insurance, and the challenge of working with providers to let them know that low-income people who show up for care without insurance will often be eligible and deemed covered. This requires developing a system in which providers can communicate easily with an eligibility office that would then enroll people in the appropriate coverage. In exchange, the providers would then get paid for the services they provide.

That said, the challenges do not seem insurmountable. And given that free coverage has not proven sufficient to get people covered, some form of auto-enrollment may be a natural next option for policymakers to consider.