Sunday, August 16, 2020

The ACA today: A mid-pandemic assessment

Subscribe to xpostfactoid

High-CSR coverage in Nebraska (top) vs. unsubsidized coverage in Alaska (bottom)

Enrollment in ACA marketplace plans totaled 12.7 million at the end of Open Enrollment 2016 and 11.4 million at the end of OE 2020.  Enrollment as of the end of OE has shrunk in every year since the 2016 peak.

Those top-line fact tend to shape basic perceptions of how the marketplace has fared. But in itself the top line is almost meaningless. The marketplace, and the U.S. healthcare system more generally, have undergone an array of changes, some of which mitigate the impression of modest, steady erosion, and some of which reinforce or even intensify it. Among the OTOHs:
  • Collapse in unsubsidized enrollment.  Enrollment in ACA-compliant health insurance plans by people who did not qualify for ACA premium subsidies stood at 6.7 million in the first quarter of 2016 and at 3.4 million in Q1 2019. * Huge premium increases in 2017 (triggered by a genuine market correction) and 2018 (in response to Republican attempts to repeal the ACA and sabotage the marketplace) appear to have driven half of unsubsidized enrollees out. The ACA's failure to provide affordable insurance to a large percentage of those who don't qualify for subsidies stands out as one of its core failures -- offset though it may be by the law's provision of subsidized insurance (mostly Medicaid) to some 20 million people who would otherwise have remained uninsured.

    I await evidence that two years of basically flat premiums in 2019 and 2020 -- with a third apparently on the way in 2021 -- may have started to reverse the bleeding in unsubsidized enrollment. (Premium reductions in New Jersey in 2019 did achieve this.) So far no such evidence is forthcoming. The Trump administration's fostering of a market for medically underwritten, lightly regulated so-called short-term limited duration plans (now available in most states for a full year term, belying the short-term label) may have forestalled any recovery in unsubsidized ACA-compliant coverage.

  • Subsidized enrollment is up.  CMS pointed this out in an August 2019 report that was focused on the decline in unsubsidized enrollment. The report showed average monthly subsidized enrollment higher in 2018 than in 2016. That trend has continued: as of February 2020, subsidized enrollment in ACA marketplace plans is slightly higher (9.232 million) than in February 2018 (9.230 million).

    How can this be, given that overall marketplace enrollment as of the end of OE fell by 11% from 2016 to 2020, and subsidized enrollment accounts for about 85% of on-exchange enrollment? Well...

  • Retention has improved.  In 2016, the enrollment tally fell 12.6% between the end of Open Enrollment on January 31 and effectuated enrollment as of March, the first month in which all who enrolled during OE owed premiums.  In 2020, enrollment dropped just 6.5% from the end of OE to the first effectuated enrollment snapshot, showing enrollment as of February.** Average monthly enrollment tallies show year-round improved retention in 2018 compared to 2016. In fact enrollment in the ACA marketplace was higher in December 2018 than in December 2016.  Total on-exchange enrollment as of February 2020 (10,673,516)  was just 1.4% off the all-time high recorded in March 2016 (10,828,894). Given elevated "special enrollment period" enrollment during pandemic months***, it's all but certainly at an all-time high for mid-year, or for August, right now (you can check out monthly totals here).

    As I've noted before, one factor that very well may have improved retention from 2018 forward is silver loading -- a practice that began that year, after Trump cut off direct reimbursement of insurers for Cost Sharing Reduction (CSR) subsidies in October 2017. Since CSR is available only with silver plans, after the cutoff insurers were allowed to price the benefit into silver plan premiums only, which created steep discounts in bronze and gold plans in many states and regions. An awful lot of people got free or near-free bronze plans, and a free plan is easy to retain. A smaller but significant number of enrollees obtained gold plans that cost less than silver.  Another possible factor: CMS shortened the Open Enrollment season starting in 2017. Those who enroll in a shorter OE may be more highly motivated and likely to retain their plans.

  • Sabotage has hurt.  The ACA stipulates that state and federal ACA exchanges spend a portion of the user fees collected from participating insurers on enrollment assistance and outreach. In 2016, the Obama administration allocated $63 million to "navigator" programs in the 38 states using the federal exchange, (the 13 states including D.C. that established their own exchanges have their own revenue streams, via the user tax, to fund outreach). Over the next two years, the Trump administration cut the funding by 84%, to $10 million, while also cutting the advertising budget by 90%. ACA enrollment is complex -- too complex -- and low income people, often with uncertain English skills and/or limited internet access -- need skilled and empathetic assistance. Ignorance of the programs remains rife. The social service agencies and veteran enrollment counselors who staff them have shown resilience, ingenuity and commitment in the face of denuded funding, but they operate on a shoestring, and assistance is far below Obama-era levels. Meanwhile, unscrupulous brokers selling the short-term plans promoted by the Trump administration have muddied the waters and left many people with illusory coverage. 

  • Transition to Medicaid. In states that have refused to enact the ACA Medicaid expansion, eligibility for subsidies in the private-plan ACA marketplace begins at an income of 100% of the Federal Poverty Level, as opposed to 138% FPL (the cap for Medicaid eligibility) in expansion states.  As premiums are quite low at 100-138% FPL (2% of income for a benchmark silver plan with strong Cost Sharing Reduction), and as nonexpansion states are mostly poor states, about a third of enrollees in nonexpansion states would be eligible for Medicaid (and not marketplace subsidies) if the state enacted the Medicaid expansion.  Since January 1, 2016, six states have belatedly enacted the expansion, with three more in the pipeline.  Medicaid expansions trigger a multi-year decline in the state's marketplace enrollment, as enrollees with incomes in the 100-138% FPL range transition to Medicaid (which is generally free or very close to free and has zero or close to zero out-of-pocket costs). 

    Enrollment in 2020 in four states that expanded Medicaid between 2016 and 2019 - Montana, Louisiana, Virginia and Maine -- was lower by 285,000 than enrollment in the last year unaffected by Medicaid expansion in each state. Most of that decrease is attributable to expansion. 

  • OOP rising. As medical inflation continues relentlessly to exceed general inflation, the fixed actuarial value of ACA marketplace plans at each metal level leaves enrollees with ever-increasing out-of-pocket costs.  (Actuarial value refers to the percentage of the average enrollee's yearly medical costs a plan is designed to cover, according to a formula created by CMS.) In 2016, the average deductible for a silver plan with no CSR (i.e., a silver plan for enrollees with incomes over 250% FPL) was $3,064. In 2020, it was $4,544. The average out-of-pocket maximum for no-CSR silver was $6,160 in 2016 and $7,735 in 2020.

    For enrollees with incomes below 200% FPL, the strong CSR attached to silver plans provides some insulation from rising costs. Deductibles at the two highest levels of CSR (raising AV to 94% for enrollees with incomes up to 150% FPL and to 87% at 150-200% FPL) haven't changed much - they're about $200 at top-level CSR and $800 at the next level. But average annual OOP maxes have risen from $874 to $1,187 for CSR-94, and from $1,795 to $2,362 for CSR-87.

  • Silver loading mitigates. Silver loading, the inflation of silver premiums when CSR is priced in, has created significant discounts in bronze and gold plans -- mainly in the form of free or near-free bronze plans, which are widely available among enrollees over age 40. Total ACA marketplace enrollment is probably about 5% higher than it would be absent silver loading.

    While silver loading has roughly doubled gold plan enrollment to about 8% of total enrollment, however, gold plan discounts and takeup are far below the levels forecast by those who analyzed and anticipated the effects of Trump's cutoff of direct CSR reimbursement (including CBO). With CSR taken into account, the average actuarial value of silver plans obtained by enrollees exceeds the AV of gold plans, yet in most regions gold plan premiums still exceed those of silver plans. The main culprit, actuaries tell us, is probably CMS's risk adjustment formula, which favors silver plans. (The federal risk adjustment programs compensates plans in which enrollees have higher-than-average medical needs.)

    A marketplace in which gold plans were consistently priced below benchmark silver, or at least within spitting distance of it, would offer significantly better value to enrollees with incomes over 200% FPL.  In short, silver loading has added a much-needed boost to the marketplace's underpowered premium subsidies, but in a haphazard and confusing way that cries out for reformed benefit design.

  • Medicaid matters more. While the ACA marketplace has always sucked up most of the political passion attached to "Obamacare," the expansion has been responsible for the bulk of the reduction in the U.S. uninsured population achieved through the ACA. By early 2017, 12-15 million people had gained Medicaid coverage thanks to the ACA's expanded eligibility criteria, while several million more who were eligible by pre-ACA criteria had been stimulated to enroll. By contrast, about 9 million people are in subsidized ACA marketplace plans, and probably a bit shy of 13 million in all ACA-compliant plans. Medicaid enrollees generally pay no premiums and have zero or near-zero out-of-pocket costs, while ACA enrollees are exposed to thousands and sometimes tens of thousands of dollars in premium and OOP costs.

    By January 2020, total Medicaid enrollment had shrunk from a 2017 peak of 75 million by about 5% to 71 million, probably due in part to a chill on immigrant enrollment created by Trump's changes to the federal public charge rule, which now potentially penalizes the immigration status of those who enroll in Medicaid (though a court has stayed the Trump changes). Also contributing to the drop in Medicaid enrollment: peak employment (prior to the pandemic) and rising incomes. After the economy seized up this spring and summer in the wake of the U.S.'s catastrophic response to the Covid-19 pandemic, Medicaid enrollment has probably already snapped back to peak levels or beyond and should continue to increase.

    By most estimates (1,2,3), Medicaid will likely enroll about twice as many of the newly uninsured as the ACA marketplace. Under the ACA, Medicaid eligibility is based on current monthly income, while subsidy eligibility in the ACA marketplace depends on estimated annual income. For some people who earned substantial income prior to being laid off, marketplace subsidies may be weak to nonexistent, while Medicaid eligibility may be instant. The marketplace has been further sidelined by the CARES Act, which counts the $600 per week supplementary unemployment insurance benefit provided from April through July (and probably to be extended in some form at some point) as income for the purpose of determining ACA subsidies, while exempting it from consideration in Medicaid eligibility.
In its six-plus years of existence, the ACA marketplace has proved both resilient and limited in its ability to reduce uninsurance and under-insurance in the U.S. It has helped more people than it's hurt, but it has hurt several million people for whom it has rendered individual market insurance unaffordable. Its goal of ensuring that anyone who lacks access to employer-sponsored or other insurance can obtain comprehensive insurance in the individual market regardless of their medical needs or history has been undermined by under-subsidization, as the second price quote in the image at the top of this post illustrates. The Medicaid expansion has been more successful and cost-effective in serving the needs of its target population and reducing the ranks of the uninsured. Medicaid could be the basis of a more satisfactory expansion of coverage in the U.S.

Update, 8/17: Stan Dorn of Families USA, an apostle of autoenrollment of the uninsured in insurance that will cost them nothing, is part of a bipartisan team of authors urging three means to raise awareness of ACA options and streamline enrollment. Dorn et al. cite good evidence that most of the uninsured who consider insurance too expensive don't know what's available -- often including Medicaid. 


* The Q1 2019 total in the Kaiser Family Foundation analysis is an estimate, extrapolated from previous quarters.

**The year's first effectuated enrollment tally in 2016 was dated March 31 because Open Enrollment extended to January 31, and those who enrolled after January 15 did not owe their first premium until March. Beginning in 2018, enrollment in 38 states ended on December 15, and all first payments were due by January 1. In 2016, some enrollees would have paid their first premium and dropped out by March, so the comparison is not completely clean. Based on the 2016 drop from March to April of about 100,000, perhaps one percentage point in the comparison might be shaved off.

*** CMS reported that through June 2, Special Enrollment Period enrollments in 2020 (available to those who lose access to other health insurance or have other life changes) were 188,000 higher than 2019 totals to that point.  Covered California, the state-based exchange, which in response to the pandemic created and has extended an emergency SEP open to all uninsured state residents,  reported more than 100,000 excess SEP enrollments from March 20 through July 25. Other state-based marketplaces that opened emergency SEPs in response to the pandemic have reported about 100,000 enrollments through that channel, as charted by Charles Gaba as an aside rather deep in this post


  1. Thanks for an excellent summary.

  2. I've just discovered your blog, and, as I have an interest in the ACA, and some knowledge of the ACA, I find it useful. Including the "ACA today" post.

    (I'm impressed that you're following the details of the silver loading reaction to the Trump sabotage attempt. Few people know about it. I actually added it to the Wikipedia ACA article of last summer, but it got taken out, as did, ultimately, everything I put in, even though no one found any mistakes. See my link below to the Wikipedia ACA article pointing to Problems section, where I had actually placed it in other sections that I wrote, which show up as well from the link.)

    ACA Problems section, full article as well, version of last summer (all of my sections are out as of 7/2/20 when the last remaining section, the "Problems" section was removed by other editors.


    Mainly, this post is to communicate with you to make sure you know of two particular defects in the ACA which are not well publicized.

    One is the "family glitch", which is that the employer-insurance affordability test for families to get a subsidized on-exchange plan when employer insurance is available at some cost-- eligibility uses the cost of the employer insurance for the individual worker ONLY, rather than cost for the whole family.

    The second is an issue of the complexity of the ACA, being split into Medicaids, expanded Medicaid, and on-exchange plans, plus CHIPs, is simply too technically complicated for some states to manage without causing coverage gaps, sometimes that can bankrupt a person.

    The problems will vary from state to state, since Medicaid eligibility determination varies by state, and some states also have their own exchanges.

    Myself, in MA, since the ACA started, I've had about 4 incidents, each of which could have bankrupted me had I been expensively sick at the time. (Most of these in the last 2 years.)

    Twice, mid-year, the MA agencies moved me down to expanded Medicaid from on-exchange plans, on very short notice, like 6 days, and unpredictable to me. (In one case, this was due to mid-year adjustment of FPL cutpoints, moving me and about 1000 other people to expanded Medicaid and losing on-exchange automatically. A second time, the state ignored my own estimated income for the year (which turned out to be correct), and used its estimate. (In both cases, the issue is, if we had been in the hospital at the time, with approved procedures and in-network providers, the sudden change could have invalidated all of that.)

    A separate time, in switching me from expanded Medicaid to on-exchange, the state told me it could not guarantee there would be no gap in coverage. Medicaid is one agency, and the exchange, another, and they do not coordinate, they told me.

    These were all in the last two years.

    Earlier on, in switching me at end of open enrollment up to on-exchange from expanded Medicaid, 3 agents in a row told me there would be a gap of 2 weeks or 6 weeks between coverages. They said "no mistake--that's just how they do it". (In this case, by intense publicity in the press and on facebook, I was able to get things straightened out. But only by these methods.)

    I don't what happens in other states. I imagine some have problems like this, but do not publicize them. (With my own problems, in most cases the state told me they were not going to try and fix the underlying issues.)

    Also note, many states have NOT switched Medicaid eligibility to annual for next year, that is, they did not coordinate it with on-exchange subsidy eligibility. Thus, some income patterns month to month, for people with annual incomes of say 150% FPL resulting in a person being eligible for NEITHER expanded Medicaid nor subsidized on exchange for some months! (Just post-ACA passage, some experts said the synchronizing needed to be done, but apparently some states have not done it.)