Auto-Enrollment Into Individual Market Health Insurance Coverage
|
 For a print-ready PDF of this page, click here.
|
Auto-enrollment among small employers (<50 employees)
Prior to the enactment of the ACA, small employers offering health insurance coverage to their employees had an incentive to maximize the number of employees participating in their health plans. In particular, many states and nearly all insurers had some type of participation requirements that an employer had to meet in order to be issued a policy. These participation requirements were intended to reduce the adverse selection that would occur if only workers with higher health costs enrolled in coverage. Where allowed, some insurers varied premium rates by participation levels. Insurers would require wage and tax forms to ensure that only bona fide employees were being insured as well as to verify participation requirements. All of these procedures were done to better match the risk being assumed by the insurer to the premium rate being charged.
The ACA eliminated small employer incentives to maximize participation rates by requiring that insurers enroll all small employers applying for coverage during the annual open enrollment period, even if they do not meet traditional participation requirements. ACA small group premiums can’t vary by participation levels; premiums can vary only by certain group characteristics: age, area, tobacco use, and benefit plan. The ACA risk adjustment program transfers payments across insurers within the small group market to reflect differences in risk that aren’t reflected in premiums, including the variation in risk caused by different participation levels.
In 2017, more than one-third of firms with 3-49 employees used auto-enrollment, higher than the rate for larger firms (see Table 1). One factor likely contributing to this higher rate is that small employers generally offer fewer health plan options than larger employers, thus making auto-enrollment easier to implement.
Table 1. Auto-Enrollment by Firm Size, Among Firms Offering Health Benefits, 2017
| Firm Size | Percentage Using Auto-Enrollment |
| 3-49 Employees | 35% |
| 50-199 Employees | 13% |
| 200-999 Employees | 8% |
| 1,000-4,999 Employees | 11% |
| 5,000+ Employees | 18% |
| All Firms | 31% |
Among larger employers offering health benefits, the use of auto-enrollment generally increases by employer size, but industry can be even more important than an employer’s size in determining likelihood of an employer engaging in auto-enrollment5. For instance, employers in the technology, utility, and finance industries are more likely to auto-enroll employees into health coverage than employers in the retail and hospitality industries, or those with large seasonal workforces. The “default” plan is most commonly a low benefit option, typically a high-deductible health plan with an account feature. Opt-out opportunities are provided and employers generally do not require proof of coverage to opt out of the employer’s health plan. Due to the increasing cost of health coverage, many employers that have historically done auto-enrollment have moved to require active elections each year.
The ACA initially included a requirement for employers with more than 200 employees to automatically enroll new employees into one of its health plans. Adequate notice to employees was also required, as was the opportunity for employees to opt out of any coverage in which they were automatically enrolled. No final regulations or guidance were released and the provision was repealed in 2015 prior to becoming effective.
Barriers to further expansion of auto-enrollment among employers include high administrative costs, the difficulty of determining alternative coverage sources, and the greater complexity when coverage extends to spouses and dependent children or when multiple plans are offered. Industries with high opt-out rates would face the administrative costs of initially enrolling employees and setting up payroll deduction mechanisms, as well as the costs of reversing those mechanisms for those who opt out. Administrative costs would also be higher in industries with high turnover rates. If auto-enrollment is implemented without a corresponding affordability test, many new hires may end up with significant financial commitments, potentially leading to higher opt-out rates. Health plans typically cover employees and their spouses and dependent children, but any auto-enrollment default likely would be for employees only because the employer may not know of the presence of a spouse or dependent children or their access to coverage. There are also duplicate coverage issues associated with auto-enrollment, such as access to coverage elsewhere via a spouse or other coverage source.
Communication to the employee is critical in an auto-enrollment environment. The default plan and the payroll deduction must be clearly communicated. If the default plan is not comprehensive in coverage or uses a network in which an employee’s provider does not participate, employees may face unexpected out-of-pocket expenses. Opt-out provisions have to be clearly stated. Currently, employees enroll during an open enrollment period that is prior to the effective date of the coverage. Auto-enrolled employees may not be able to change plans after the enrollment period ends, so it is important they get information regarding their plans and any payroll deductions prior to that.
Medicaid and Medicare
In 2016, two states introduced auto-enrollment programs for portions of their Medicaid populations. Louisiana began using data from the Supplemental Nutritional Assistance Program (SNAP) to determine income eligibility for Medicaid and to enroll those eligible6. South Carolina began using auto-enrollment for a demonstration program for Medicare-Medicaid dual eligibles. Individuals age 65 and older who are eligible for both Medicare and Medicaid and have not already selected an integrated plan among those offered are assigned one using an algorithm to identify the plan that best meets their needs7. Results of the programs in Louisiana and South Carolina have not been published to date.
Medicare uses auto-enrollment for certain individuals. Individuals already receiving Social Security or Railroad Retirement Benefits (RRB) at least four months before being eligible for Medicare are automatically enrolled in both premium-free Part A and Part B, which requires a premium. People who are automatically enrolled have the choice of whether they want to keep or opt out of Part B coverage. Individuals who are not receiving Social Security or RRB benefits are not automatically enrolled. The Medicare Part D prescription drug program offers a low-income subsidy program that provides premium and cost-sharing subsidies to eligible enrollees. Dual-eligible beneficiaries and certain other low-income beneficiaries are automatically enrolled in a zero-premium Part D plan if they haven’t already joined a plan.
The Centers for Medicare and Medicaid Services (CMS) introduced an auto-enrollment program allowing Medicare Advantage organizations (MAOs) to offer seamless conversion for their commercial and Medicaid enrollees into Medicare Advantage (MA) plans upon reaching Medicare eligibility. Approved MAOs would identify eligible aged and disabled individuals 90 days prior to Medicare eligibility, inform individuals of conversion enrollment 60 days prior to the MA effective date, and allow individuals to opt out before coverage begins. Twenty-nine MAOs received approval and over 15,000 newly eligible beneficiaries were enrolled for the 2015 plan year8. In October 2016, however, CMS responded to concerns about the program from beneficiaries, providers, and advocacy groups by suspending new approvals in order to further review the program9.
Takeaways From Current Auto Enrollment Programs for Health Insurance
Experience from current auto-enrollment programs suggests several conditions are needed to facilitate its implementation. These include:
The availability of information to identify eligible individuals. Employers are able to identify and enroll their employees, although they may not have spouse or dependent children information or information on whether employees have coverage from another source. State and federal governments can access public program data to identify eligible individuals.
The ability to assign individuals to appropriate plans. The enrolling entity needs to be able to assign individuals into a plan. Assignment is straightforward when only one plan is offered, but gets more complicated when more plan choices are available. Employers can choose one of their lower-cost options for their auto-enrollment default. More vulnerable populations may require a more complicated process, such as the algorithm used in South Carolina’s dual-eligible program plan assignment, to better ensure they are enrolled in an appropriate plan. To the extent it is available, it may be appropriate for such algorithms to incorporate information on age, income, existing provider relationships, specific medical needs, and plan enrollment history. Processes also need to be set up for individuals so they can opt out or change plans.
A method to collect necessary premiums. Under an employer plan, any required premium contributions can be deducted from the employee’s paycheck. Under Medicare, any required premiums can be deducted from a beneficiary’s Social Security benefits. The availability of zero-premium plans, such as under the Medicare Part D low-income subsidy program, eliminates the need to collect premiums.
Reasonable administrative burden. Identifying eligible individuals, assigning them to appropriate plans, collecting any required premiums, and allowing for opt-outs and plan changes can be administratively complex and costly. These burdens can be higher in populations that experience a lot of turnover, for instance employers in certain industries, and in populations with higher opt-out rates.
Implementing Auto-Enrollment in the Individual Market
Identifying Uninsured IndividualsAs noted above, auto-enrollment programs work best when information is available to identify potential enrollees. For the individual market, there is not an existing data source for identifying individuals without other coverage. Even if a data source were available, it would likely need to be updated fairly frequently due to the residual and transitional nature of the individual market.
One option proposed is to use tax filing data. The IRS requires individuals to report their health insurance coverage for the tax filing year. The health insurance coverage information could be used to identify uninsured individuals who could be eligible for auto-enrollment. However, tax data only show coverage status during the prior year. It would not necessarily reflect coverage status at the time of auto–enrollment, which could be during the next open enrollment period. At that point, the coverage information would be at least a year old and an additional step could be necessary to ensure that individuals lacking coverage are offered coverage for the next year. Using tax filing data could be more effective if open enrollment were to coincide with the end of the tax filing season. No information would be available for people who don’t file tax returns.
Income information from tax filings is currently used to determine eligibility for ACA premium tax credits. Some auto-enrollment proposals would specifically target uninsured individuals who would be eligible for a zero-premium plan due to premium subsidies. This approach will be discussed in more detail below.
Another approach would be to tie coverage to other programs—for instance, to auto-enroll individuals upon entering an educational program, obtaining a driver’s license or passport, or obtaining a loan. Such methods may not capture a large number of eligible enrollees, may disadvantage financially vulnerable consumers, and coverage and subsidy status information may not be available. Although coverage information would be available when people receive health services, signing up people at the point of medical service, such as at a hospital, would result in the worst form of adverse selection.
The most comprehensive method would be to have one entity responsible for tracking the insured status of the entire population. The entity would need to create and maintain a database of the entire population and each individual’s insurance status. All insurers, self-insured employers (perhaps through third-party administrators), Medicare, Medicaid, and any other state and federal health insurance programs would need to report all members (including spouses and dependent children) covered by their plans to this entity, preferably on a monthly basis. This information could be used to determine coverage information for each person in the database. Unless the database includes the entire U.S. population, using for instance a near-universal source such as Social Security records, all uninsured people would not be captured in the data. Such a comprehensive database would be very difficult and expensive to set up and maintain. In addition, there could be data privacy and cybersecurity concerns.
Instead of focusing on the entire population, a less comprehensive data collection method would be to require insuring entities to report information to a central source on individuals who are losing coverage. This could include, for instance, individuals who are losing coverage because they are leaving a job or are losing dependent coverage upon turning age 26. Auto-enrollment efforts could concentrate on this population. To be most effective, however, it would need to be determined whether people losing coverage had already gotten new coverage.
Plan Assignment and Premium Collection
Methods for assigning identified uninsured individuals into health plans would need to be developed. These could include randomly assigning individuals to plans with premiums below a certain threshold. Once individuals are assigned to a plan, the insurer would be responsible for collecting any premium owed. Premium subsidies would be collected from the government and any additional premium would have to be collected from the insured. Unlike employers, which can deduct premiums from employees’ paychecks, collecting premiums directly from individuals can be more challenging. Insurers would need to communicate premium requirements to the individuals, but would not have a way to ensure those payments are made. If uninsured dependents are auto-assigned to child-only policies, the insurer would have to determine who has financial responsibility for these dependents so they could be billed for the coverage. Individuals declining to pay any additional premium would have their coverage terminated retroactively. This increases administrative costs, and claims costs may already have been paid but were not covered by premiums. If healthier people are more likely to opt out and higher-cost people retain coverage, auto-enrollment could worsen the risk pool rather than improve it. Enrolling individuals into zero-premium plans, as discussed below, would reduce administrative concerns and would increase the likelihood that auto-enrollment leads to an improved risk pool.
Focusing auto-enrollment on young adults no longer eligible for dependent coverage has been suggested. One such approach would be to auto-enroll individuals age 27 to 30 into catastrophic plans using a tax credit (currently, premium tax credits can’t be used toward catastrophic plans). The young adults coming off dependent coverage would need to be identified in order to accomplish the auto-enrollment and to determine whether other coverage is available and whether they are eligible for a tax credit. Incorporating more young adults into the ACA market could help improve the risk pool. Under current ACA rules, however, catastrophic coverage is risk-adjusted separately from the metal plans (i.e., platinum, gold, silver, bronze), meaning pricing for insurers could be more complicated and metal level plans wouldn’t necessarily see premium reductions. A benefit of this approach is that it could familiarize young adults with insurance coverage and increase the likelihood that they will continue to purchase coverage in the future.
Auto-Enrollment Into Zero-Premium Plans
Because collecting premiums from auto-enrolled individuals can be difficult, current auto-enrollment programs are typically limited to those with zero-premium options or when the entity can withhold the premium from a payment to the individual. One way to avoid this problem in an individual market auto-enrollment mechanism would be to apply it only to people who receive a high enough premium subsidy to pay the entire premium. For instance, because the termination of cost-sharing reduction (CSR) payments to insurers increased premiums, and therefore premium tax credits, the Kaiser Family Foundation estimated that more than 4 million subsidy-eligible uninsured could purchase a zero-premium bronze plan 201810. Finding an effective method of enrolling these individuals into coverage would likely improve the risk pool and put downward pressure on premiums. Re-imposing an individual mandate financial penalty, at either the federal or state level, and directing that penalty toward the purchase of a health insurance plan would increase the number of individuals who could purchase a zero-premium plan11. Fewer individuals would be eligible for zero-premium bronze plans if silver premiums were lower, for instance if the federal government resumes paying plans for CSRs.
Under this method, the auto-enrollment system could use IRS insurance coverage information to determine who is uninsured and IRS or state tax income information to determine whether the uninsured person qualifies for a premium subsidy. Currently, IRS data are used to determine eligibility for ACA premium tax credits, which are available for individuals with household incomes between 100 and 400 percent of the federal poverty line (FPL). Tax credits are based on the premium of the second-lowest silver plan, which varies by rating area and age. Tax credits phase down with income and are not available for individuals above 400 percent of FPL. If the current year’s income is significantly different, the individual may be asked to repay some or all of the tax credit.
Individuals with lower incomes may be able to be assigned to zero-premium plans, but it is less likely that individuals with higher incomes could be. The availability of zero-premium bronze plans depends on the difference in cost between the second-lowest silver premium and the lowest bronze premium and may not be available in all rating areas. Individuals with incomes between 100 and 250 percent FPL are eligible for cost-sharing reductions, but only if they enroll in a silver plan. As a result, some individuals with low incomes would have lower total premium and out-of-pocket costs by enrolling in a silver plan rather than a zero-premium bronze plan with higher cost-sharing requirements.
Other proposals would replace the current premium subsidy structure with a flat premium tax credit or an age-based flat premium tax credit. A flat tax credit would be simpler to administer but could result in the tax credit being able to purchase differing plan designs for individuals depending on their age. Under current rating rules, premiums may vary by a 3:1 ratio between ages 21 and 64, with the slope of the premiums dictated by federal (and sometimes state) regulation. The flat tax credit could also vary by age, but unless it varies by age with exactly the same slope as the premium curve, the credits could be used to purchase different plan designs for individuals depending on their age. A flat tax credit would also pay for different benefit plans by geographic area, because premiums vary by geographic area and state. If insurers have to develop and maintain many plans in order to have plans that can be purchased with tax credits at every age/rating area, this will add to administrative expenses.
Once identified, individuals could be enrolled in coverage with premiums at or below the tax credit. The amount of premium tax credit required to purchase the lowest available premium varies by geographic area and age (see Table 2). If the tax credit is not enough to purchase a bronze plan, then the plan could be designed with variable cost-sharing so that the premium would equal the available subsidy. This could require higher deductibles and a higher maximum out-of-pocket limitation than currently allowed (the 2018 out-of-pocket maximum is $7,350). For instance, one study found that plans would need to have very low actuarial value (AV), with some deductibles over $20,000 per person, in order for older adults to be covered by a $3,000 tax credit12. If the premium subsidies are not sufficiently generous, the insured may be unable to afford the required cost-sharing.
Table 2. Lowest Available Bronze Premiums at Ages 27 and 62, 2018, Selected Cities
| 2018 Lowest Available Bronze Premium | Age 27 | Age 62 |
| Pittsburgh, Pa. | $2,388 | $6,546 |
| Nashville, Tenn. | $3,456 | $9,474 |
| Omaha, Neb. | $5,232 | $14,343 |
Source: American Academy of Actuaries Individual and Small Group Markets Committee calculations based on the Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation (ASPE), “Health Plan Choice and Premiums in the 2018 Federal Health Insurance Exchange,” ASPE Research Brief, October 30, 2017
Similar to how Medicare randomly assigns certain beneficiaries eligible for the Part D low-income subsidy to zero-premium plans, individuals could be randomly assigned among a set of plans provided by issuers with premiums at or below the tax credit. Rules to determine what insurance plan to assign people to would have to be developed in a way that would not create anti-selection against any particular insurer and might also need to incorporate enrollee medical needs. Insurers offer different plan designs and networks and have different cost structures. For a specific price point corresponding to the tax credit, the plans available from different insurers will have different cost-sharing structures and different networks. Having standardized plans would reduce the differences in plan offerings, but insurers would still have different premiums for similar plan designs due to different network and cost structure differences. With random assignment, it could be difficult to ensure that individuals in similar circumstances are enrolled into plans that are of similar value or that individuals are enrolled in the plans that best meet their needs.
Auto-enrolled individuals would need to be contacted to make them aware of their coverage, and to inform them of their ability to opt out and their responsibility to notify the insurer if they get other coverage such as employer- or government-based programs. The insured may potentially have to pay back the value of the tax credit to the government at tax time if they do not notify their insurer to cancel coverage when obtaining employer or other government coverage, or of an increase in income in the case of income-related tax credits. It will be critical to inform these individuals which plan they have been assigned to and where to locate the network directory. Individuals may be assigned to plans that do not include their existing providers (this may be less of an issue for previously uninsured individuals if they didn’t have a regular source of care). There may need to be a “window” between this notification and the final enrollment to allow individuals to switch insurers or plans so as to get into plans that better meet their needs.
Facilitated Enrollment
Rather than directly auto-enrolling eligible individuals into coverage, a system could be put in place that facilitates enrollment. For instance, insurance navigators could reach out to individuals identified as potentially being uninsured and eligible for premium subsidies. These navigators could work with the individuals to confirm their coverage status and tax credit eligibility, provide information on available insurance choices, and enroll them in a plan. Although this approach would be resource-intensive and would add administrative cost, it could reduce the complexities and uncertainty regarding whether an individual is still uninsured, enrolling them into a plan that meets their needs, setting up opt-out mechanisms, and collecting required premiums.
A key to an effective auto-enrollment program for the individual market is for enrollment to increase insurance participation rates among those who are healthy. If only those with higher health costs are targeted through auto-enrollment (such as enrolling individuals when they receive health services), or if healthy individuals have higher opt-out rates, then it is less likely the risk pool will improve.
Similar to how Medicare randomly assigns certain beneficiaries eligible for the Part D low-income subsidy to zero-premium plans, individuals could be randomly assigned among a set of plans provided by issuers with premiums at or below the tax credit. Rules to determine what insurance plan to assign people to would have to be developed in a way that would not create anti-selection against any particular insurer and might also need to incorporate enrollee medical needs. Insurers offer different plan designs and networks and have different cost structures. For a specific price point corresponding to the tax credit, the plans available from different insurers will have different cost-sharing structures and different networks. Having standardized plans would reduce the differences in plan offerings, but insurers would still have different premiums for similar plan designs due to different network and cost structure differences. With random assignment, it could be difficult to ensure that individuals in similar circumstances are enrolled into plans that are of similar value or that individuals are enrolled in the plans that best meet their needs.
Auto-enrolled individuals would need to be contacted to make them aware of their coverage, and to inform them of their ability to opt out and their responsibility to notify the insurer if they get other coverage such as employer- or government-based programs. The insured may potentially have to pay back the value of the tax credit to the government at tax time if they do not notify their insurer to cancel coverage when obtaining employer or other government coverage, or of an increase in income in the case of income-related tax credits. It will be critical to inform these individuals which plan they have been assigned to and where to locate the network directory. Individuals may be assigned to plans that do not include their existing providers (this may be less of an issue for previously uninsured individuals if they didn’t have a regular source of care). There may need to be a “window” between this notification and the final enrollment to allow individuals to switch insurers or plans so as to get into plans that better meet their needs.
Facilitated Enrollment
Rather than directly auto-enrolling eligible individuals into coverage, a system could be put in place that facilitates enrollment. For instance, insurance navigators could reach out to individuals identified as potentially being uninsured and eligible for premium subsidies. These navigators could work with the individuals to confirm their coverage status and tax credit eligibility, provide information on available insurance choices, and enroll them in a plan. Although this approach would be resource-intensive and would add administrative cost, it could reduce the complexities and uncertainty regarding whether an individual is still uninsured, enrolling them into a plan that meets their needs, setting up opt-out mechanisms, and collecting required premiums.
Summary
Auto-enrolling uninsured individuals into individual market coverage has the potential to help improve the risk pool and put downward pressure on premiums. However, there are significant challenges to making auto-enrollment work in the individual market. There is not an existing framework or comprehensive data source to identify individuals (and their spouses and dependent children) eligible for coverage who are not eligible for coverage elsewhere. In addition, because there is not an easy way to automatically collect individual market premiums, such as withholding from a check, auto-enrollment is likely to be more effective if individuals can be enrolled into coverage that is no additional cost to them. This involves calculating the premium subsidy for the individual or family and identifying coverage that can be purchased with the available subsidy.A key to an effective auto-enrollment program for the individual market is for enrollment to increase insurance participation rates among those who are healthy. If only those with higher health costs are targeted through auto-enrollment (such as enrolling individuals when they receive health services), or if healthy individuals have higher opt-out rates, then it is less likely the risk pool will improve.
| 1. |
American Academy of Actuaries, An Evaluation of the Individual Health Insurance Market and Implications of Potential Changes, January 2017. |
| 2. |
See, for example, Brigitte C. Madrian and Dennis Shea, “The Power of Suggestion: Inertia in 401(k) Participation and Savings Behavior,” The Quarterly Journal of Economics, November 2001, 116(4): pp.1149-87. |
| 3. |
Ibid. |
| 4. |
The Kaiser Family Foundation and Health Research & Education Trust, Employer Health Benefits: 2017 Annual Survey (Figure 3.10), 2017. |
| 5. |
Information regarding auto-enrollment among medium and large employers was gathered through informal discussions with employee benefit consultants. |
| 6. |
Louisiana Department of Health, “Louisiana Receives Approval for Unique Strategy to Enroll SNAP Beneficiaries in Expanded Medicaid Coverage,” June 1, 2016; Medicaid Expansion Annual Report 2016/2017, June 30, 2017. |
| 7. |
South Carolina Department of Health and Human Services, “Healthy Connections Prime Passive Enrollment Scheduled to Begin April 2016; Seniors in South Carolina now have a new health care option,” Jan. 22, 2016. |
| 8. |
Centers for Medicare and Medicaid Services, “Seamless Conversion Enrollment—Policy and Data of Approved Medicare Advantage Organizations,” Oct. 21, 2016. |
| 9. |
Centers for Medicare and Medicaid Services, “Seamless Enrollment of Individuals upon Initial Eligibility for Medicare,” Oct. 21, 2016. |
| 10. |
Matthew Rae, Larry Levitt, and Ashley Semanskee, “How Many of the Uninsured Can Purchase a Marketplace Plan for Less Than Their Shared Responsibility Penalty?” Kaiser Family Foundation issue brief, November 2017. |
| 11. |
After being automatically enrolled, an individual would not be subject to any financial penalty for that plan year, and therefore might be less likely to be eligible for a zero-premium plan the following year. |
| 12. | Linda Blumberg, “What Can Consumers Purchase with the Age-Related Tax Credits in the Empowering Patients First Bill?” Urban Institute, March 2017. This study examined tax credits proposed under the Empowering Patients First bill: $1,200 for people ages 18-34; $2,100 for people ages 35 to 49; $3,000 for people ages 50 and older; and $900 per child up to age 18. The study also assumed that allowable age rating variation would expand from 3:1 to 5:1. |
Share
