Increasing Insurance Choices In The Affordable Care Act Marketplaces, 2018–21

The Affordable Care Act (ACA) relies on insurers to offer health plans in the individual health insurance Marketplaces. Since the ACA’s implementation, levels of Marketplace competition have varied, reaching a nadir in 2018. We examined the characteristics of counties that experienced changes in insurers’ participation in the ACA Marketplaces from 2016 to 2021. Using data from the Kaiser Family Foundation and other sources, we found that 1,968 counties (accounting for 66 percent of the US population younger than age sixty-five) have more insurers in 2021 than in 2018, whereas only twelve counties (comprising 0.4 percent of the US nonelderly population) have fewer insurers. The number of counties with monopolist Marketplace insurers declined from 1,616 in 2018 to 294 in 2021. Recent Marketplace insurer gains were more likely in counties that lost insurers from 2016 to 2018 or had a monopolist insurer in 2018. Increased competition may lead to lower gross premiums in the ACA Marketplaces. Given the Biden administration’s support for the ACA Marketplaces, it appears likely that the ACA individual health insurance market will be stable and profitable for the next several years.

The Affordable Care Act’s (ACA’s) individual health insurance Marketplaces rely on the willingness of private insurers to sell regulated insurance products across the entire country. 1 Insurers’ decisions to participate have two parts: They must first decide whether they will participate at all, and then they must decide in which geographic areas they will offer plans. These choices have varied substantially over time, and overall Marketplace insurer participation reached a nadir in 2018. 2 Uncertainty guring the 2017 and 2018 rate-setting cycles over the ACA’s future and federal policy changes were both implicated in insurer exits.

Insurers’ Marketplace participation has been an ongoing concern since the passage of the ACA, and previous research concluded that counties with limited insurer participation experienced faster gross premium increases from 2014 to 2018. 3 Insurer exits also have been associated with subsequent reductions in Marketplace enrollment at the county level. 4,5 Insurers returned to some counties served by the federal Marketplace in 2020 and 2021. 6 However, little is known about the characteristics of the counties that gained insurers after the 2018 plan year. In this article we assess the extent to which counties gained Marketplace insurers during 2018–21, the characteristics of these counties, and the types of insurers that entered new service areas or expanded their Marketplace offerings. This information may help guide state and federal policy on issues such as reinsurance and state-based public options.

Study Data And Methods

We obtained annual county-level data on insurers’ Marketplace participation from the Kaiser Family Foundation’s Insurer Participation on the ACA Marketplaces database for 2016–21, 2 which is based on insurers’ state rate filings and a review of information on HealthCare.gov. We merged county-level data with data on community-level characteristics from the Health Resources and Services Administration’s Area Health Resources Files, 7 data on the partisan composition of state legislatures and governors’ offices from the National Conference of State Legislatures, 8 data on average medical loss ratios for all individual insurance plans by state from the Centers for Medicare and Medicaid Services’ Medical Loss Ratio Data and System Resources, 9 and Kaiser Family Foundation data on state Medicaid expansion 10 and the creation of state-based Marketplaces. 11 As a supplementary analysis, we also extracted data on insurer and plan characteristics for 2018 and 2021 from the Robert Wood Johnson Foundation’s Health Insurance Exchange (HIX) Compare database. 12 HIX Compare contains information on nearly every ACA Marketplace plan and insurer in all fifty states plus Washington, D.C. Our final data set included 25,080 annual observations from 3,135 counties.

ANALYSIS

Our analysis proceeded in five steps. First, we counted the number of US counties and the percentage of the population younger than age sixty-five with one, two, or three or more insurers by year. Second, we examined mean changes in insurer participation from 2018 to 2021, stratified by the number of insurers that sold on-Marketplace products in each county that had either one or two insurers in 2018. Third, we created maps to show how changes in the number of participating Marketplace insurers varied by geography.

Fourth, we estimated linear probability models to explore the characteristics of counties that either had limited insurer participation in 2021 (defined as two or fewer insurers) or gained Marketplace insurers from 2018 to 2021. We included several covariates that may relate to insurers’ participation on the ACA Marketplaces. 13 These included rurality and population demographics that have been previously associated with the likelihood of insurer entry, such as the percentage of residents who are Black, Hispanic, or ages 45–64; 14 per capita death rates (which may be a proxy for underlying population health); 15 each state’s average medical loss ratios for all individual insurance plans (a general measure of insurer profitability); 16 whether the state expanded Medicaid (as Medicaid expansion been shown to transfer a cohort of people with higher-than-average health care costs out of the ACA Marketplaces and thus reduce Marketplace premiums); 17,18 party control of state legislatures and governors’ offices; and whether the state operated its own Marketplace (an indicator of state-specific political and financial support for ACA Marketplace operations). 13,17,19,20 Rurality was determined using Rural-Urban Continuum Codes from the Department of Agriculture (metropolitan, codes 1–3; nonmetropolitan, codes 4–7; and rural, codes 8–9). 21 State partisan control took on values of Republican, Democrat, or divided, depending on whether a single party controlled both chambers of the state legislature and the governor’s office. We used 2018 values for all characteristics, with the exception of our reinsurance waiver variable, which took on a value of 1 if a state implemented a Section 1332 waiver to implement a reinsurance program with an effective date from 2018 to 2021 and a value of 0 otherwise. Because the scales for these covariates vary widely, continuous variables were Z-scored to aid in interpretability. We also tested whether either having limited insurer participation in 2021 or gaining Marketplace insurers from 2018 to 2021 was related to having a monopoly insurer in 2018 or declining Marketplace participation between 2016 and 2018. Monopoly insurer status was not included in adjusted models because of its high correlation with several other variables. Standard errors were clustered at the state level, and all models were weighted by total county population younger than age sixty-five to produce nationally representative estimates. 13

Last, we described the characteristics of new insurers that entered into the Marketplaces in policy years 2019–21, existing insurers that expanded their Marketplace offerings into new geographic areas, and existing insurers that did not expand those offerings.

LIMITATIONS

This study had several limitations. Our analysis was cross-sectional in nature, and thus our results should be interpreted as associations. State policy and partisanship variables are highly correlated, which complicates differentiation of their independent associations with the outcomes. We were unable to observe enrollment in individual plans or insurers, nor did we observe net premiums paid. The HIX Compare database is presented at the rating area level, which does not map to county boundaries in some states. We were unable to see which plans were offered in specific counties or to observe changes in consumer welfare (the benefits that individuals derive from insurance coverage) or behavior in response to changes in Marketplace offerings and insurer participation. The Area Health Resources Files do include mortality information for specific age groups, but these are often suppressed because of low cell counts (fewer than twenty-five deaths) for sparsely populated counties or younger age groups. We relied on the overall crude death rate instead.

Study Results

Overall, insurers’ participation in the ACA Marketplaces increased from 2018 to 2021, partially reversing the trend in insurer exits from 2016 to 2018 ( exhibit 1 ). In 2021, 1,451 counties (which comprise 75.2 percent of the US population younger than age sixty-five) have access to three or more Marketplace insurers, whereas 1,390 counties (with 21.4 percent of the US nonelderly population) are served by two insurers, and 294 counties (with 3.4 percent of the US nonelderly population) have a single insurer.

Percent of US population younger than age 65, by number of insurers on the Affordable Care Act Marketplace at the county level, 2018–21

SOURCE Authors’ analysis of annual county-level data on insurers’ Marketplace participation from the Kaiser Family Foundation and population data from the Area Health Resources Files, 2018–21.

These gains were nonrandom and disproportionately accrued in counties with limited insurer participation. From 2018 to 2021 counties gained a mean of 1.22 insurers (95% confidence interval: 1.17, 1.26) if they only had one insurer in 2018, compared to a mean gain of 0.68 insurers (95% CI: 0.62, 0.74) in counties with two insurers in 2018 and a mean gain of 0.51 insurers (95% CI: 0.38, 0.64) in counties with three or more insurers in 2018 ( exhibit 2 ).

Mean number of Affordable Care Act Marketplace insurers per county, 2016–21, by insurer participation in 2018

SOURCE Authors’ analysis of annual county-level data on insurers’ Marketplace participation from the Kaiser Family Foundation, 2016–21. NOTE The exhibit displays the mean number of unique corporate parent companies of insurers (which may have multiple Health Insurance Oversight System identification numbers for different licenses for products such as health maintenance organizations and preferred provider organizations available on the Affordable Care Act Marketplaces) in 2016, 2017, 2019, and 2021, stratified by the number of Marketplace insurers that were available in 2018.

County-level changes in the number of insurers participating in the Marketplaces in 2018–21 are shown visually in exhibit 3 . We found that 1,968 counties (which account for 66 percent of US nonelderly population) had more insurers in 2021 than in 2018. The South Atlantic, Midwest, and Pacific Northwest regions observed the largest gains in insurers during this period. Insurer gains were also made in several states that had insurer monopolies in 2018; Alabama remains the only state with a majority of its land area served by one insurer in 2021. Only twelve counties had fewer insurers in 2021 compared with 2018, comprising approximately 0.4 percent of the US nonelderly population. The number of counties with monopolist Marketplace insurers declined from 1,616 (26 percent of the US nonelderly population) in 2018 to 294 (3 percent of the US nonelderly population) in 2021.

Change in the number of insurers offering Affordable Care Act Marketplace insurance policies, by county, 2018–21

SOURCE Authors’ analysis of annual county-level data on insurers’ Marketplace participation from the Kaiser Family Foundation, 2018–21. NOTE The exhibit displays the net change in the number of unique corporate parent companies of insurers (which may have multiple Health Insurance Oversight System identification numbers for different licenses for products such as health maintenance organizations and preferred provider organizations) offering Marketplace plans between 2018 and 2021.

Compared to counties with three or more Marketplace insurers in 2021, counties with limited insurer participation in 2021 were more likely to also have had limited insurer participation in 2018 (100.0 percent versus 36.6 percent) and to have Republican-controlled (50.4 percent versus 46.9 percent) or divided (37.1 percent versus 30.6 percent) state governments, and they were less likely to be in states with state-run Marketplaces (14.8 percent versus 35.8 percent) ( exhibit 4 ). These findings comport with previous research. 13

EXHIBIT 4

Characteristics of counties with limited versus increased participation in Affordable Care Act Marketplaces, 2018–21

SOURCE Authors’ analysis ofannual county-level data on insurers’ Marketplace participation and state Medicaid expansion status from the Kaiser Family Foundation, state partisan control from the National Conference of State Legislatures, medical loss ratios from the Centers for Medicare and Medicaid Services, and demographics from the Area Health Resources Files, 2018–21. NOTES The exhibit shows mean characteristics of counties with and without limited insurer participation (defined as two or fewer participating Marketplace insurers) in 2021 and counties that did and did not have increases in Marketplace insurer participation from 2018 to 2021. Observations from 3,148 counties were weighted by county nonelderly population.

Compared to counties that did not gain Marketplace insurers from 2018 to 2021, counties that gained insurers were more likely to have limited insurer participation in 2018 (58.9 percent versus 39.2 percent), a monopoly Marketplace insurer in 2018 (34.0 percent versus 9.8 percent), or loss of insurers from 2016 to 2018 (88.6 percent versus 80.7 percent) or to be in states with Republican-controlled state governments (56.0 percent versus 31.5 percent), and they were less likely to be in states that had a state-run Marketplace (23.9 percent versus 43.8 percent), expanded Medicaid (54.9 percent versus 87.0 percent), or implemented reinsurance waivers (14.6 percent versus 20.2 percent). Additional results for demographic and rurality variables, as well as p values, are in online appendix exhibit A1. 22

We also examined characteristics of plans offered by three types of ACA Marketplace insurers: insurers that did not participate in ACAMarketplaces in 2018 and entered Marketplaces for the first time in 2019–21, insurers that participated in ACA Marketplaces in 2018 and expanded into new service areas in 2019–21, and insurers that participated in ACA Marketplaces in 2018 and did not expand into new service areas during 2019–21. Subsidiaries of the same parent company were counted as a single insurer. A total of 121 unique insurers participated in any Marketplace in 2021; ten of these were new entrants (offering 1,629 plans), thirty-eight were existing insurers (offering 35,851 plans) that had expanded into additional service areas since 2018, and seventy-three were existing insurers (offering 16,606 plans) that did not expand into new areas. A breakdown of new and existing ACA Marketplace insurers’ 2021 product offerings, including metal levels and plan type, is in appendix exhibit A2. 22 Compared with existing Marketplace insurers that did not expand into new areas, existing Marketplace insurers that expanded into new areas were less likely to be Blue Cross Blue Shield Association members (7.9 percent versus 37.0 percent), were more likely to offer silver plans (78.0 percent versus 67.2 percent), and were consequently more likely to offer cost-sharing reduction plans (58.5 percent versus 50.1 percent). Compared with insurers that did not expand into new areas, insurers that expanded into new areas were more likely to offer health maintenance organization (49.7 percent versus 43.4 percent) and exclusive provider organization (44.1 percent versus 37.1 percent) plans. Compared with all existing insurers, new Marketplace entrants were more likely to be Blue Cross Blue Shield Association members (40.0 percent versus 27.0 percent) and to offer plans that were health maintenance organizations (63.2 percent versus 47.7 percent), multitier (36.2 percent versus 8.3 percent), or child only (4.7 percent versus 0.6 percent), and they were less likely to actively market those plans (86.5 percent versus 95.2 percent) (appendix exhibit A2). 22 The top ten insurance carriers in terms of total unique insurance plans offered on the Marketplaces are in appendix exhibit A3. 22

Discussion

Our results demonstrate that the percentage of people who reside in US counties with only one or two insurers on the ACA Marketplaces has decreased markedly since 2018. Increased insurer participation in 2019, 2020, and 2021 was associated with both a loss of insurer participation from 2016 to 2018 and insurer monopolies in 2018.

Many factors may have led to insurers’ Marketplace exits between 2016 and 2018, including insurers facing insolvency and liquidating because of insufficient risk adjustment or risk corridor payments, 23,24 as well as losses resulting from high claims costs relative to premium revenue. 25 In 2017, as insurers were deciding whether and where to offer plans on the Marketplace for 2018, they faced substantial political and policy uncertainty. Insurers had experienced several years of significant losses on the Marketplaces with only hints of financial improvement, political uncertainty as a new administration with a unified government had promised to repeal and replace the Affordable Care Act, and policy uncertainty as direct payments for cost-sharing reduction subsidies were in litigation. 13,26–29 As a result, the percentage of people who lived in US counties with only one or two Marketplace insurers spiked in 2018.

However, insurers’ practice of silver loading to compensate for the loss of federal government reimbursement for cost-sharing reduction benefits and the failure of any attempt to legislatively repeal or replace the ACA removed some of this financial and policy unpredictability. Silver loading and the lack of Marketplace competition led to insurers having very low average medical loss ratios and high gross per member per month margins for 2018. 30 High levels of profitability resulting from high average gross premiums and concurrent low net premiums (due to increased subsidies) may have led insurers to enter new markets or reenter markets that they had previously exited.

Policy Implications

The ACA Marketplaces still have substantially lower insurer participation in 2021 than during their peak in 2015. 2 Gross premiums have been declining since 2018, partially as a result of increased insurer competition. 6 Increased competition is an unalloyed positive for people who do not receive premium subsidies. 31 However, the ACA’s price-linked subsidy system produces a counterintuitive result: Lower gross premiums may result in concomitant reductions in premium spreads between low-cost and benchmark plans on the Marketplace. This could increase the out-of-pocket expenses paid by some subsidized buyers (those with household incomes of 100–400 percent of the federal poverty level), as these spreads determine the magnitude of the federal premium tax credit subsidy. 32 Provisions in the American Rescue Plan Act of 2021 that increased ACA individual-market insurance subsidies may counteract this trend.

Increased insurer participation in ACA Marketplaces may reduce potential cost savings that subsidized consumers could achieve under state public option programs. Lower gross premium levels resulting from increased insurer participation could keep net premiums flat for subsidized Marketplace enrollees or could lead net premiums to increase. 33 Washington State’s public option program offered medical providers rates that were similar to commercial rates, at 160 percent of the Medicare fee schedule. 34 The public option has not significantly reduced gross or net premiums for the plans sold on the ACA Marketplaces where there is competition, as private insurer premiums converged to a common level before implementation of the public option. 35

The American Rescue Plan Act’s enhanced subsidies for people with incomes of 100–400 percent of the federal poverty level and the new availability of subsidies for people with incomes greater than 400 percent of poverty will increase the number of ACA individual-market health insurance buyers who are insensitive to gross premium level. These changes may also lead to increases in average enrollment duration and decreases in morbidity in the individual market as the enhanced subsidies increase affordability, and they could lead healthier people to enroll in coverage. 36 Given the Biden administration’s support for the ACA Marketplaces, it appears likely that the ACA individual health insurance market will be both stable and profitable for the next several years. 37

Conclusion

Recent trends of increased ACA Marketplace insurer participation in counties with declining participation from 2016 to 2018 could be interpreted as a response to reduced political, policy, and financial risk. “Repeal and replace” was foreclosed by its political defeat. The ACA’s financial structure of price-linked subsidies builds resilience into the program, as this subsidy system guarantees a large customer base that is minimally exposed to changes in premium levels. 38,39 In the next several years political and policy predictability as well as expanded subsidies and increased federally funded advertising and political support in profitable markets could be an attractive environment for insurers to enter new markets and expand their service areas, thus leading to enhanced competition.