xpostfactoid

Thursday, January 09, 2025

ACA fulfills early forecasts

Note: All xpostfactoid subscriptions are now through Substack alone (still free), though I will continue to cross-post on this site. If you're not subscribed, please visit xpostfactoid on Substack and sign up.

CBO inside

As we near the end of the ACA marketplace’s Open Enrollment Period (OEP) for 2025, CMS is out with a new enrollment snapshot showing 23.6 million plan selections. As OEP has another week to go in the 32 states using HealthCare.gov, and up to 23 days more in some state-based marketplaces, Charles Gaba estimates final enrollment at 24.2 million*, up about 13% over OEP 2024 — and just about double the 12.0 million total in OEP 2021. That was the last year before the Democratic Congress and the Biden administration changed the game by radically boosting premium subsidies and expanding eligibility for them as part of the American Rescue Plan Act (ARPA), enacted in March 2021 (and effective immediately during an emergency Special Enrollment Period that had started on February 15).

The ARPA subsidy increases made benchmark silver coverage with strong Cost Sharing Reduction free to enrollees with income up to 150% FPL; removed the 400% FPL cap on subsidy eligibility; and reduced the percentage of income required for the benchmark silver plan at all income brackets in between. Enacted as a pandemic measure lasting only through 2022, the enhanced subsidies were extended through 2025 by the Inflation Reduction Act, enacted in August 2022. If the Republican Congress declines to extend the ARPA subsidy boosts, OEP 2025 will stand as the marketplace’s high-water mark, and millions will drop coverage beginning in 2026 (7.2 million according to the Urban Institute’s estimate).

I have written more than once that the ARPA subsidy boosts brought the ACA within striking distance of fulfilling the mission expressed in its name (affordable coverage for all who lack access to pre-ACA sources of health insurance) and envisioned by its drafters — and by the Congressional Budget Office, e.g., in its final projection prior to enactment, on March 20, 2010. That is almost literally true now. Enrollment in the ACA’s core programs — the Medicaid expansion and the subsidized marketplace — is quite close to the 2010 CBO ten-year projections, albeit six years late (and just shy of four years after ARPA was enacted — when, in a sense, the clock restarted).

Top line, the CBO in 2010 was almost dead-on as to the ACA’s long-term effect on the uninsured rate, forecasting that 92% of the nonelderly U.S. population would be insured ten years in. That wasn’t true in 2019 (pre-pandemic, pre-ARPA), but it’s close to true now. The latest quarterly estimate from the National Health Interview Survey (NHIS), for Q2 2024, pegs the uninsured rate in the under-65 population at 9.1%. In 2010 CBO forecast 23 million uninsured as of 2019; the NHIS pegs the uninsured population (all ages) at 25.3 million in Q2 2024.**

I cannot locate the March 2010 CBO report on the ACA’s likely effects (updates from 2011 forward are readily available), so I’ll have to rely on an old printout. Forgive some old scribblings in the 2015 column.

Some notes as to the forecasts for 2019:

CBO anticipated the marketplace approaching full capacity with 23 million enrollees in 2017, its fourth year of operation, edging up to 24 million in 2018 and remaining at that level in 2019. Current enrollment is…24 million. (Add 1.8 million in the Basic Health Programs established in three states, which CBO did not anticipate — see the first note at bottom.)
CBO projected a net increase of 16 million in Medicaid enrollment from 2010 to 2019. There are a couple of ways to assess this projection. First, as of September 2019, total Medicaid enrollment was up 15.1 million from July-September 2013, the pre-ACA comparison point used by KFF.
Viewed another way, the total number of Medicaid enrollees rendered “newly eligible” by ACA expansion criteria (that is, adults with income up to 138% FPL who would not have been eligible by other criteria) was 16.6 million*** in June 2024 — the last month for which numbers are available. By that point, the “Medicaid unwinding” — resumption of disenrollments in May 2024 after a three-year pandemic-induced moratorium — was mostly over.
While getting the top line for increased Medicaid enrollment more or less right, CBO actually underestimated the ACA’s effects on Medicaid enrollment, as it expected the ACA expansion to be in effect in every state. In June 2012, the Supreme Court rendered the expansion optional for states, and initially only 24 enacted it. As of December 2019, 34 states had enacted the expansion; as of now, ten states, including Florida and Texas, still have not.
The persistence of “nonexpansion states” has inflated marketplace enrollment. In the ten remaining nonexpansion states, 5.8 million marketplace enrollees reported income in the 100-138% FPL range, which would have placed them in Medicaid had all states enacted the expansion.
In light of the points above, based on what it knew/assumed in 2010, CBO overestimated marketplace enrollment and underestimated Medicaid expansion (as about 6 million marketplace enrollees in nonexpansion states “ought” to be in Medicaid). Those misses more or less cancelled each other out as to the top line.
Medicaid enrollment increased by more than 20 million during the three-year moratorium on involuntary disenrollments enacted in March 2020 as part of the Families First Act, a pandemic relief bill. The “unwinding” — resumption of redeterminations and disenrollments — that began in spring 2023 was clumsy and cruel in many states, and too many children lost coverage. When the dust settled, however, enrollment was still 7.7 million higher in September 2024 than in March 2020, the eve of the pandemic, and the uninsurance rate may prove not to have risen. After a year-plus of unwinding, enrollment among those rendered eligible by ACA expansion criteria was up was 5.4 million higher in June 2024 than on the eve of the pandemic in March 2020.
CBO forecast a net drop in employer-sponsored insurance (ESI) of just 3 million over ten years. According the KFF’s annual survey based estimates, enrollment in employer-sponsored insurance was 3 million higher in 2010 (estimated at 157 million) than in 2024 (154 million). The labor force did increase by about 4.4 million from 2019 to 2024, perhaps suggesting a slight further shrinkage in ESI.
CBO also forecast a drop of 5 million from 2010 to 2019 in off-exchange “nongroup” insurance combined with “other” insurance — an opaque catch-all category that includes disability Medicare (about 7 million at present), student health plans, care at correctional facilities, and some other odds and ends. The drop was forecast to occur chiefly in off-exchange nongroup enrollment, and that drop was probably steeper than forecast, driven by a) a sharp rise in unsubsidized premiums in 2017-18, b) a steady fade-out of pre-ACA “grandfathered” and “grandmothered” (don’t ask…) plans, and c) by ARPA’s removal of the income cap on subsidies, which prompted a lot of new enrollment in the 400-600% FPL range. CBO probably underestimated the drop in off-exchange nongroup enrollment. In a 2019 brief, KFF estimated nongroup enrollment in 2011 (pre-ACA) at 10 million. By the first quarter of 2015, per KFF, off-exchange nongroup enrollment in ACA-compliant and noncompliant plans combined had dropped only modestly to 8.8 million, but after steep premium increases in 2017-2018 it plummeted to an estimated 3.3 million by Q1 2019. CBO estimates from June 2024 peg nongroup coverage bought outside the marketplace at a similar 3.1 million in 2024. That looks like a drop of about 7 million in off-exchange nongroup enrollment from the pre-ACA period to 2019 and beyond — perhaps 2 million more than forecast, though a lot of moving parts are involved.

It may seem somewhat dicey to compare 2025 totals to CBO’s 2010 projections for 2019. But I believe the comparison is instructive. The 2010 10-year forecast encompassed just six years of operation of the ACA’s core programs, which kicked off in January 2014 (though a few states started the Medicaid expansion early). Marketplace and Medicaid enrollment stalled, and in fact went into modest reverse, during the Trump years, until the pandemic struck. The ARPA subsidy enhancements, which have been in effect through four OEPs (2022-2025), represented a new beginning — and the premise here is that the enhanced subsidies enabled the marketplace to fulfill the function envisioned at the level envisioned.

ARPA made that fulfillment possible. In late 2009, as the ACA was writhing through multiple iterations in the Senate, it was plain to progressive advocates that the emerging subsidy schedule was inadequate to the marketplace’s purpose. In his 2011 book about the battle to pass the ACA, Richard Kirsch, national campaign manager from 2008-12 for Health Care for America Now (HCAN), an umbrella group formed by unions and progressive nonprofits to advocate for universal health care, took the inadequacy of the subsidies as a given:

In the President’s September address to Congress, the President not only made a concession on the public option. He also said, “the plan I’m proposing will cost around $900 billion over ten years.” Yet $900 billion was not enough money to make health care truly affordable to the uninsured. Why did the President make another, preemptive concession to the bill’s opponents, one that would significantly damage his core goal? An article co-authored by Robert Pear and The New York Times White House correspondent Jackie Calmes summarized the impact nicely: “The number suggests a political and fiscal calculation to avoid the sticker shock of the trillion-dollar threshold. But it probably means that Mr. Obama could fall short of his goal of providing universal coverage for all Americans because the lower cost may force lawmakers to reduce the subsidies needed to help more uninsured individuals and small businesses seeking coverage for employees.”...

While the marketplace roughly met enrollment goals in its launch year, 2014, and enrollment increased substantially in 2015, CMS officials in the later Obama years knew that enrollment would not grow in line with CBO projections — and starting in OEP 2018, team Trump took steps to ensure that enrollment would not grow (e.g., gutting funding for enrollment assistance and outreach, shortening the OEP, and standing up an alternative market of ACA-noncompliant plans). Average monthly enrollment reached a peak in 2016 that would not be passed until the pandemic and resulting job layoffs stimulated off-season enrollment in 2020 (as highlighted in the Average Monthly Enrollment column below). In 2021, ARPA took over, and the rest is the history we’ve been examining.

Sources: Marketplace Open Enrollment Public Use Files and Full-Year and February Effectuated Enrollment tables, available via the 2024 Early Effectuated Enrollment Snapshot.

None of this is to suggest that the marketplace is an unproblematic program (and wait till team Trump 2.0 gets hold of it). I personally think that a market of private plans offered mostly by for-profit insurers paying modified commercial rates to providers and pushed by competition toward very narrow provider networks is a suboptimal way to offer health coverage to those who lack access to other sources of insurance. Out-of-pocket costs in the marketplace are way too high; networks are too narrow; and the proliferation of plan choices in each market, numbering over 100 on average in 2024, is nonproductive and bewildering. All that said, coverage is currently available and affordable to most of those who need it. That wasn’t true until ARPA, and it’s a BFD. It very likely won’t be true in 2026 or any time soon.

- - -

* Another 1.8 million people are enrolled in the Basic Health Programs offered in lieu of the ACA marketplace to low-income enrollees by New York, Minnesota and Oregon (1.6 million of them in New York*). BHPs, an alternative provided to states by the ACA statute, are standardized, Medicaid-like plans with low out-of-pocket costs. New York’s Essential Plan is technically no longer a BHP, as the state extended eligibility to enrollees with income up to 250% FPL in 2024. By statue, BHPs can only serve enrollees with income up to 200% FPL. New York reorganized the Essential Plan, preserving its federal funding, via an ACA Section 1332 “innovation waiver.”

** Since 2019, the population has grown by about 9.3 million, suggesting that if CBO’s estimate for that year had been on target and remained stable until now, another 760,000 people would be uninsured. So the modest gap between the CBO 2010 estimate and the current NHIS estimate of the absolute number of uninsured is roughly in line with the percentage gap.

*** CMS offers two measures of ACA expansion enrollment in Medicaid: those rendered eligible by ACA criteria (“Group VIII”), and those rendered newly eligible by those criteria. The distinction stems from a handful of states independently extending Medicaid eligibility to all adults with income up to a threshold of 100% FPL or higher; in those states, a number of Group VIII enrollees would presumably remain eligible if there were no ACA expansion. In June 2024, there were 20.9 million Group VIII enrollees, 16.7 million of them “newly eligible.”

- - -

Non sequitur: some writing from me in a different mode, for kids.

Tuesday, December 31, 2024

Gateway to a dark age: 2024 in review

Carnival for Muskovites; Lent for the rest of us

All bloggers (excuse the archaism…) get a free end-of-year post reviewing their year’s work, right? Here goes.

First I have to confess that it feels a bit odd to review themes in the relatively mundane world of ACA administration and performance in a year that in the 11th month shaped up as an annus horribilis, with Trump’s reelection. On the healthcare front, the parade of cranks and perverse contrarians nominated by Trump threaten disaster at the CDC, NIH and FDA, while a Republican Congress readies a fresh run at the kind of catastrophic spending cuts that Republicans have fantasized about for decades. My hope is that with the smallest possible majority in the House, Republicans will fail or (to be fair to some chunk of their members) balk at major cuts to Medicaid or major legislative changes to the ACA. I even retain a sliver of hope that the boosts to ACA subsidies enacted in the American Rescue Plan Act, currently funded only through 2025, will be extended, at least in part. But I expect deep wounds to our institutions, in healthcare as everywhere. If damage stops short of outright catastrophe, we can count ourselves lucky.

One major thread in this year’s posting is the concentration of enrollment growth in the ACA marketplace in the post-ARPA years in states that refused to enact the ACA Medicaid expansion. This is not primarily a result of enrollment fraud, as alleged by the Project 2025-adjacent Paragon Institute under the leadership of Trumpist health economist Brian Blase, who is laying the groundwork for major cuts to public health insurance programs. In July I countered Blase’s charges at some length, while acknowledging valid points — e.g. the obvious need for a crackdown on broker fraud (a crackdown begun this summer, per below) and for some regulatory tightening (eliminating the ability of low-income enrollees to make monthly enrollment changes, an option exploited by unscrupulous brokers). Marketplace enrollment growth in OEP 2024 and, it’s now emerging, in 2025, was driven in large part by the Medicaid unwinding (a fact that Blase acknowledges but casts in a nefarious light).

As we brace for Trump 2.0, I hope my 3-part review of Trump 1.0’s administration of the ACA marketplace (along with a more recent parsing of Trump and Vance’s claims that Trump “strengthened” the ACA) may prove useful. Surprise: it wasn’t all bad — although it seems that Trump’s “heroism was quite inadvertent” (to paraphrase Woody Allen’s antagonist in Love and Death) on the main count (see: silver loading).

A rising tide of broker fraud in the ACA marketplace burst into view this year, via excellent reporting from KFFs Julie Appleby, underpinned by a lawsuit alleging fraud perpetrated by a major EDE web-broker and a pair of agencies deploying dozens of downline agencies. In several posts, I delved into the evidence and CMS’s response. Posts included a look at gray-area fraud and sloppy agency practice; a close look at the expanded allegations in an amended complaint from the plaintiffs alleging large-scale fraud; and red flags in a past CMS celebration of rapidly expanding broker participation in the marketplace.

The Biden years were a heady time for ACA watchers — though always shadowed by the threat of a Trump resurgence. Medicaid enrollment, including among those made eligible by the ACA expansion, swelled, a a three-year pandemic-induced moratorium on disenrollment played out — then shrank back in an “unwinding” of that moratorium kicked off in May 2023, leaving a net increase since the eve of the pandemic of about 11%, or 8 million, as of August 2024. The enhanced marketplace premium subsidies implemented with ARPA in March 2021 triggered a near-doubling of enrollment, from 12 million in OEP 2021 to a likely 24-odd million by the end of OEP 2025. The unemployment rate has hovered near 4% for the entirety of Biden’s term, a full employment level not sustained since the 1960s. Not surprisingly, the uninsured rate dropped to an all-time low of 7.9% nationally in 2023 (the last year tracked).

All that enrollment growth is under threat from a new Trump administration and majority-Republican Congress. Medicaid enrollment will be cut by, at the very least, work requirements being readied now in red states. The Republican Congress will try for truly catastrophic further cuts — e.g., reducing the federal match rate for the ACA expansion population; further reducing match rates for “rich” states; and imposing block grant funding or per capita caps on federal Medicaid funding. In the ACA marketplace, the odds are against renewal of the ARPA subsidy boosts. While there is allegedly little appetite among House Republicans to allow the subsidies to revert to pre-ARPA levels, it’s hard to imagine them taking positive action to extend subsidies scheduled to sunset in 2026. Perhaps Democrats will manage to slip extensions into must-past omnibus spending bills.

As Republicans at least gesture toward major cuts, progressives will dust off and update their analyses of the major cuts threatened in 2017 — as the Center for Budget and Policy Priorities, Georgetown’s Center on Health Insurance Reforms and Center for Children and Families, the Center for American Progress, and Charles Gaba are already doing. It will be a weary rematch. As in 2017, fending off catastrophic defunding and major repeal will be the best outcome to be hoped for. Perhaps the more intense danger is on the public health/disease management front, as Trump appointees gear up to disarm our already-inadequate defenses before the next pandemic and roll back decades of progress in vaccination and infectious disease control. Happy New Year!

Thursday, December 26, 2024

Passive auto re-enrollment spikes in the ACA marketplace

Don't put your health insurance renewal on autopilot

Until late this morning, the narrative for the ACA marketplace’s Open Enrollment Period (OEP) for 2025 appeared to be that enrollment was down from 2024 highs. The December 4 enrollment snapshot showed 5,364,197 “active” plan selections nationally, compared to 7,299,900 as of December 6, 2023. That’s an apparent 26.5% drop, which should be discounted by about 5% to account for the two extra enrollment days in last year’s early December snapshot.

The “losses” appeared to be concentrated in HealthCare.gov, the federal marketplace (FFM), as active enrollment in the 20 state-based marketplaces was actually up a bit year-over-year as of early December, even discounting Georgia, which newly launched an SBM for OEP 2025. Taking all states together, new enrollees were down from 1,476,658 on December 6, 2023 to 987,689 on Dec. 4 this year (again, discount for two days).

But mid-OEP year-over-year comparisons are always dicey, and Charles Gaba, for one, has been skeptical as each snapshot since Nov. 1 indicated lagging “active” enrollment — which includes new enrollment and re-enrollment by those who logged on and actively chose a plan, as opposed to those who passively allow auto re-enrollment. Gaba put forward two reasons to doubt that enrollment this year would lag.

First, it seemed likely that much of last year’s large pool of new enrollees (4.2 million in the 32 FFM states) would passively allow auto re-enrollment. New enrollment last year was heavily concentrated at low incomes, and attributable in large part to the Medicaid “unwinding” — that is, the process of re-determining Medicaid enrollees’ eligibility after a three year pandemic-induced paused. Redeterminations began in May 2023 and continued into this summer. Most low-income enrollees pay no premium and may not be fully engaged in the enrollment process.

Second, some half dozen state-based marketplaces have been reporting strong enrollment growth in their own news releases. Also, CMS’s Dec. 4 report showed 11% year-over-year auto re-enrollment growth in the SBMs — up from 3,49,956 last December to 3,785,626 this year, again excluding Georgia’s total.* (The December reports show auto re-enrollment in the SBMs but not in the FFM because SBMs effectuate auto re-enrollment early in OEP or even prior to it, whereas the FFM books re-enrollments in mid-December. More on that below.)

Well, today CMS released new top-line figures, inclusive of auto re-enrollment in the FFM, and the new numbers vindicate Gaba’s prediction. Auto re-enrollment in the FFM almost doubled, from 3,624,950 as of the end of OEP 2024 to 7.4 million as of now for 2025 (auto re-enrollment happens all at once). That increase drove 2025 enrollment in the FFM to 16.6 million, already surpassing last year’s total FFM enrollment of 16.4 million, which included 1.3 million in Georgia (now included in the SBM total, which was not provided in today’s release). As of this time last year, FFM enrollment had reached 15.3 million. Strip out Georgia’s 1.2 million total from the tally for this time last year, and as Gaba points out, FFM enrollment is up by about 18%. Six SBMs that Gaba has tracked show a 23% year-over-year increase.

Massive auto re-enrollment in the federal marketplace is not an unmixed blessing. As I outlined in a prior post focused on differences in auto re-enrollment between the FFM and the SBMs:

Auto re-enrollment can be dangerous, because 1) enrollees’ personal circumstances that affect subsidies — their income and the family members seeking coverage in the exchange — may change; 2) an enrollee’s current plan’s premium may rise in the coming year; and 3) most unpredictably, the benchmark (second cheapest silver) plan against which subsidies are set can change. If the coming year’s benchmark plan has a lower premium than the current year’s, subsidies shrink, since enrollees pay a fixed percentage of income for the benchmark plan. If the enrollee’s premium rises and the benchmark falls, it’s a double whammy.

The problem is particularly acute in the FFM, because HealthCare.gov

sends out a renewal letter, but with no specific information as to subsidy and premium in the coming year for the enrollee’s current plan. Instead, the FFM requires insurers to send renewal letters prior to November 1 (first day of OEP), with an estimate of premium in the current year. But the insurer’s letter, while it provides the plan’s new premium (before subsidy) in the current year and an estimate of what it will cost net of subsidy, bases the subsidy estimate on the prior year’s benchmark.

Auto re-enrollment is less problematic in the SBMs, because the SBMs ensure that enrollees and their agents or brokers have better information as of the start of OEP: generally, an estimate of what their current plan will cost them in the coming year based on the next year’s premiums and benchmarks and assuming no change in the enrollee’s income. Accordingly, auto re-enrollment rates have historically been much higher in the SBMs than in the FFM. In the FFM in 2024, just 30% of renewals were auto re-enrollments, compared to 72% in the SBMs.

So far, about 51% of 2025 re-enrollments in the FFM are passive auto re-enrollments, up from 30% last year. (That percentage may drop a bit, as some auto re-enrollees may change plans before the Jan. 15 end of OEP, with the plan switch effective on Feb. 1).

In the FFM in 2024, 55% of enrollees had income below 150% FPL, entitling most of them** to free benchmark silver coverage, compared to just 16% of enrollees in the SBMs (all of the 2024 SBM states have expanded Medicaid, which cuts out the large pool of marketplace enrollees found in the nonexpansion states). Moreover, enrollment since spring 2022 has been available year-round to enrollees with income below that threshold, and the Medicaid unwinding continued through this summer. In short, there is a huge cohort of low-income enrollees in the FFM, most of whom probably paid no premium in 2024.

Lower income enrollees tend to be lower-information, often with limited English proficiency and/or limited access to or comfort with computers. Since early 2021, when enhanced ACA subsidies made high-AV coverage free for enrollees with income up to 150% FPL, the ranks of agents targeting this population has swelled. A large if hard-to-determine portion of agent-assisted enrollments are now executed by high-volume call centers, which in some cases have engaged in outright fraud and in perhaps a larger number of cases provide cursory service. (There are lots of good agents, but the soaring number of agents registered with HealthCare.gov — 83,000 in 2024, up from 49,000 in 2018, suggests a hypercompetitive market, and allegations of large-scale fraud, by CMS as well as by litigants, suggests entry of a significant number of bad actors). In the high-volume call centers, an enrollee may not have an ongoing relationship with a single agent, but rather connect with a new one at each contact, as in a customer service center.

Seven million-plus auto re-enrollments in the FFM marketplace may portend some rate shock in coming months.

- - -

* The newly launched Georgia Access, Georgia’s SBM, appears to have auto re-enrolled essentially all existing enrollees, as the Dec. 4 snapshot lists 1.2 million auto re-enrollees for state. Georgia enrollment as of the end of OEP 2024 was 1.3 million.

** A small percentage of enrollees with income below 150% FPL may be ineligible for subsidies, e.g., because of an offer of “affordable” insurance from an employer. A somewhat larger percentage of the nearly 400,000 enrollees with income below 100% FPL are ineligible for subsidies. Subsidy eligibility begins at 100% FPL for all except lawfully present noncitizens subject to the federal 5-year bar to Medicaid eligibility (or to longer waiting periods in a few states).

Photo by JESHOOTS

Saturday, December 21, 2024

Paul Krugman asks whether health insurance is purely parasitical; doesn't quite answer

Paul Krugman, noting that most U.S. healthcare spending is filtered through private insurers, takes aim (in his new post-NYT retirement Substack) at private health insurance in the U.S.:

Let me offer a somewhat, but only somewhat, caricatured view of U.S. health care: It’s a system in which taxpayers bear the cost of major medical care, but this taxpayer money flows through private companies that take a cut, spend a lot on administration, and do their best to deny care to people who need it.

His argument as to whether private insurers add value is sketched in lightly, and focused entirely on Medicare Advantage:

What service do private insurers provide in return for the tolls they in effect collect on a largely taxpayer-financed system? Medicare Advantage plans generally offer more extensive coverage than traditional Medicare. But there doesn’t seem to be any clear evidence that this is because private insurers provide efficiency gains the public sector doesn’t. What happens instead is that Medicare Advantage plans appear to be able to game the system sufficiently that they receive more taxpayer funding per enrollee than traditional Medicare spends on recipients in equivalent health…So you could make the case that at this point private health insurance is, in large part, a parasitical racket.

You could make that case, but leading with “you could” is a pretty equivocal way to make it. My take on the evidence (adapted from my comment on Krugman’s post) is below. The first point is a footnote to Krugman’s overview of private insurance’s share of U.S. healthcare spending.

1) Most Medicaid coverage is now also administered through private insurance companies (MCOs). Per KFF, 74% of Medicaid enrollees are in MCOs.

2) According to a KFF literature review, including a review of enrollee satisfaction surveys, Medicare Advantage enrollees are about as satisfied as traditional Medicare enrollees. By all accounts, problems in MA tend to come when you need serious care (especially post-acute care) — and denials appear to be on the increase. According to a Senate investigative report (inspired by reporting from STAT’s Bob Herman and Casey Ross), coverage denials from UnitedHealthcare MA plans for acute care more than doubled between 2020 and 2022 after the company adopted an AI tool (owned by UHC subsidiary NaviHealth, and also used by Humana and CVS Aetna) to assess claims.

3) Insurers "save" healthcare spending by denying not only needed but unneeded care -- and it's difficult to parse out the proportions of each. Surveys indicate that MA plans do encourage people to get screenings and so may prevent more acute care down the road. According to the KFF literature review, “The analyses consistently found that Medicare Advantage enrollees had higher utilization of preventive services and lower utilization of post-acute care and home health services.”

4) Some healthcare economists (e.g., Austin Frakt) find evidence that Medicare Advantage treatment protocols "spill over' into traditional Medicare and so save money there.

5) Private companies (MACs) also administer traditional Medicare, but with a very broad coverage grant and few coverage denials. MACs have no incentive to deny coverage.

6) In countries that deliver universal healthcare through private insurers, such as Germany and Japan, major medical insurers must be nonprofit. Germans can opt out of the nonprofit Statutory Health Insurance system into private plans, as 11% of the population did in 2019. But the government sets provider payment rates for private plans and limits premium increases. In Japan, as in many countries that deliver universal coverage through private plans, supplemental insurance can be for-profit.

7) In the U.S. whatever money insurers save by managing (denying) care is swamped by overpaying providers, which happens because we have no uniform payment rates, so providers divide and conquer. In the case of Medicare Advantage, where providers are paid Medicare rates or less, the savings are negated by payment formulas that overpay MA plans on a capitated basis, via a) a ridiculous risk adjustment system that incentivizes massive gaming, b) over-generous payment benchmarks, and c) bonus payments based on quality ratings that don't measure quality effectively.

8) Private insurers probably do have the capacity or potential to add administrative and care management value. But in the U.S., profit skews the incentives. In fact, it skews incentives for providers as much as for insurers. Almost all participants (e.g., those owned by giant insurers or private equity) are geared toward revenue maximization ("nonprofit" or not).

P.S. My own two-part assessment of Medicare Advantage (from 2022) starts here.

Photo by Arthur Uzoagba

Saturday, November 23, 2024

On the pending expiration of the ACA's ARPA subsidy enhancements

ARPA subsidy boosts going, going...

I’d like here to highlight some factors that will affect how much ACA marketplace enrollment is likely to shrink if Republicans decline to extend the increases to ACA marketplace premium subsidies and eligibility first enacted by the American Rescue Plan Act (ARPA) in March 2021, and later extended through 2025 by the Inflation Reduction Act. I’ll assume no other major changes to the ACA (perhaps a dubious assumption).

The ARPA subsidy boosts made benchmark silver plan coverage free for enrollees with income up to 150% of the Federal Poverty Level (FPL), removed the ACA’s notorious income cap on subsidy eligibility (400% FPL), reduced the percentage of income required for a benchmark silver plan at all income levels between 150-400% FPL, and capped benchmark silver premiums at 8.5% of income for those with income above 400% FPL.

Since the Open Enrollment Period (OEP) for 2021, the last OEP before the ARPA subsidies took effect, ACA marketplace enrollment increased by 79% (9.4 million) through OEP 2024 (see Table 1 below). That enrollment growth was overwhelmingly concentrated at the 100-150% FPL income bracket, where silver coverage with strong Cost Sharing Reduction (CSR) is available for $0 premium, and in the ten states that have refused to date to enact the ACA Medicaid expansion. In those “nonexpansion” states, eligibility for marketplace subsidies begins at 100% FPL, whereas in the expansion states it begins at 138% FPL, the Medicaid eligibility threshold for almost all lawfully present adults.

The Urban Institute estimates that in 2025, the total number of people with subsidized marketplace enrollment (92% of total enrollment) will be 7.2 million higher than it would be without without the ARPA subsidy boosts. One might assume a similar drop in enrollment in years following 2025 if the ARPA enhancements expire. Urban further estimates, “In 2025…household net premiums will be lower by 50 to 100 percent for the lowest income groups under a policy of enhanced PTCs compared with a policy of original PTCs. Net premiums will be lower by about one-quarter for people with higher incomes who receive subsidized Marketplace coverage.” Reverse that for the effects of ARPA expiration.

Without venturing an overall estimate, I want to highlight the concentration of marketplace enrollment (and post-ARPA enrollment growth) at low incomes and the extent to which zero-premium coverage will still be available to low-income enrollees (albeit with lower actuarial value). I assume that Urban’s estimates of premium impact are for benchmark silver premiums, not premiums for plans actually selected, which will change if the ARPA enhancements expire.

Consider the following as to the makeup of current marketplace enrollment:

From OEP 2021 to OEP 2024, enrollment nationwide increased by 79%, from 12.0 million to 21.4 million.
59% of that growth was in the 100-150% FPL income bracket, where benchmark silver coverage was rendered free by ARPA (see Table 1 below).
In 2024, 73% of enrollment in the 100-150% FPL income bracket was in the 10 nonexpansion states (78% if you count North Carolina, which enacted a Medicaid expansion on 12/1/23*). In 2024, 6.9 million enrollees in nonexpansion states (excluding NC) had income in the 100-150% FPL range.
In OEP 2021, enrollment was flat nationwide except for in the nonexpansion states, where it increased 10%. Since OEP 2020, enrollment growth in nonexpansion states has accounted for 74% of enrollment growth nationwide.
Before ARPA passed, the ACA’s income cap on subsidy eligibility was politically fraught, as the ACA’s guaranteed issue and Essential Health Benefits requirements raised premiums, and, after steep premium hikes in 2017 and 2018 ,several million subsidy-ineligible people were priced out of the market. Removing the income cap on subsidies went a long way toward fulfilling the ACA’s promise of “affordable care.” For all that, in 2024, enrollment of those who reported income higher than 400% FPL (i.e., expected premium subsidies) was 1.5 million, about 7% of total on-exchange enrollment. Another 856,000 enrollees did not report income and so paid full price (Table 1).

Keeping in mind the heavy concentration of ACA enrollment at low incomes, various factors will preserve some availability of zero-premium coverage for most enrollees with income up to 150% FPL, and often well beyond, or foster enrollment in other ways. These include:

Silver loading, Part I. In October 2017, Trump cut off federal reimbursement of insurers for providing the Cost Sharing Reduction (CSR) subsidies mandated by the ACA to low-income enrollees in silver plan. (His basis: while the ACA mandated this reimbursement, it left funding those reimbursements up to Congress, and Republican Congresses declined to do so. The Obama administration had found reimbursement funds in couch cushions.) CSR increases the value of a silver plan to a roughly platinum level for enrollees with income up to 200% FPL. The move had been anticipated, and most state regulators responded by allowing or encouraging insurers to price CSR directly into silver plans, since CSR is available only with silver plans. Since ACA income-adjusted premium subsidies are set to a silver benchmark, this “silver loading” led to sharp reductions in net-of-subsidy premiums for bronze and gold plans (and less often, for the one silver plan priced below the benchmark in each market). As a result, the Kaiser Family Foundation (KFF) calculated that beginning in 2018, most enrollees with income up to 150% FPL, and a good number with higher incomes, had access to $0 premium bronze plans.

Silver loading, Part II. As direct CSR reimbursement had been contested from the start, prior to Trump’s cutoff, analysts at CMS, the Urban Institute, and CBO had anticipated the effects of pricing CSR into silver premiums. All three analyses anticipated that gold plans would be priced below silver plans, because most silver plan enrollees have incomes qualifying for strong CSR, and silver plan enrollees on average therefore obtain higher actuarial value than gold plan enrollees. It mostly didn’t shake out that way, as insurers have strong incentives to underprice silver plans. In about fifteen states, however, varying by year, gold plans are available at premiums below benchmark silver, sometimes by regulatory or statutory requirement, and sometimes by insurer choice (usually of a dominant insurer). Most notably — and astonishingly — Texas enacted a law requiring marketplace insurers to price gold plans far below silver. (That makes sense, since 75% of enrollees in Texas have income below 200% FPL, and the average actuarial value of a silver plan in the state is over 90%, compared to 80% for gold plans.) Accordingly, Texas’s 2.2 million enrollees with income under 150% FPL will have access to $0 premium gold plans if the ARPA subsidy boosts expire, barring other changes.

Zero-deductible bronze plans. While median bronze plan deductibles were $7,200 for a single enrollee in 2023, bronze plans with $0 medical deductibles have become an increasingly common marketplace option. At low incomes, the premium is often $0 as well. These plans have their coverage traps, like a $3,000 drug deductible (usually not applicable to generics) or a $1,500 first-day hospital inpatient copay, but they can be attractive to healthy enrollees, or those willing to trade risk and out-of-pocket cost for a more expansive provider network and doctor visits not subject to the deductible.

Agent/broker commitment. The first Trump administration cut ACA marketplace advertising and nonprofit enrollment assistance, but they did cater to and encourage participation from agents and brokers — implementing a “help on demand” feature on HealthCare.gov, and encouraging development of commercial “enhanced direct enrollment” (EDE) platforms that streamline brokers’ work and organize their client files. Thanks in part to that support, along with the ARPA subsidy boosts and insurers’ re-commitment to the marketplace, broker registration with HealthCare.gov increased from 49,000 in 2018 to 83,000 in 2024. In OEP 2024, 78% of active enrollments in HealthCare.gov states (which includes all the nonexpansion states) were broker-assisted. This growing broker engagement has been something of a mixed bag, as the widespread availability of free coverage, coupled with inadequately controlled broker access to client data in EDE platforms, has led to a major outbreak of broker fraud and encouraged entry into the market of high-volume, high-speed call centers that likely provide poor service short of outright fraud. Broker participation will probably also be pared back as offerings become less affordable (assuming ARPA subsidy expiration) — e.g., if insurers also pull back and/or reduce broker commissions. But pre-ARPA, one of the marketplace’s chief limitations was the general public’s ignorance of what was on offer. Several years of broker outreach have probably increased awareness and the likelihood of being approached among those who may need coverage.

Year-round enrollment at low incomes. In early 2022, CMS implemented a rule enabling people with income under 150% FPL to enroll year-round, via permanent availability of a monthly “Special Enrollment Period.” The model here was Medicaid; the presumption is that circumstances change often for low-income people. As a result, the gap between marketplace enrollment totals as of the end of OEP and annual “average monthly enrollment” (AME) has narrowed considerably, and AME has grown even more dramatically than end-of-OEP totals (AME was 99% of OEP enrollment in 2023 and may have actually exceeded it in 2024, when the Medicaid unwinding sent a stream of newly uninsured people into the marketplace). This too has been something of a mixed bag, as the monthly SEP has facilitated broker fraud in the form of unauthorized plan-switching of enrollees. Year-round first-time enrollment could be maintained without providing a monthly SEP. More likely, the Trump administration will rescind the rule. (Originally, the rule was contingent on $0 premium silver coverage remaining available at incomes up to 150% FPL, but that condition was later withdrawn.)

A preview of low-income options in a post-ARPA marketplace

Pre-ARPA, enrollees with income up to 138% FPL paid 2% of income for a benchmark silver plan. We can preview what options may look like in this lowest income bracket by looking at what’s available to someone paying 2% of income for benchmark silver in the 2025 marketplace — that is, a single enrollee with an income of $30,000. That’s a shade under 200% FPL, and at that income, benchmark silver is currently $49/month.

Let’s look at what’s available for a single 40 year-old at this income in Houston and Miami. (You can replicate these results or look at other markets most easily via HealthSherpa’s plan shopper.) In Florida and Texas combined, as of the end of OEP 2024, 4.1 million enrollees had income in the 100-138% FPL range — about 60% of the 6.9 million enrollees in that income range nationwide.

In Houston in 2025 (zip code 77005), a single 40 year-old earning $30,000 would pay (as noted above) $49/month (1.94% of income) for the benchmark silver plan, which has a $500 deductible and a $3,000 out-of-pocket (OOP) maximum. That’s what enrollees with income up to 138% FPL will pay for benchmark silver if the ARPA subsidy boosts expire and there are no further changes. That premium could be prohibitive at a lower income (100% FPL in this year’s marketplace is slightly over $15,000/year). But in 2025, this same Houstonian also has access to two gold plans for $0 premium, as well as to a bronze plan with a $0 deductible for $15/month (and to the one silver plan below benchmark for $38/month). Both of the $0 premium gold plans are from Blue Cross, a desired brand in Texas, whereas the lowest-cost silver BCBS plan available to this person is $61/month.

Switch the scene to Miami (zip code 33134), and the 40-year old with the $30k income, paying 1.94% of income for the benchmark silver plan, does not have cheap gold available, nor any premium difference between the benchmark and lowest-cost silver (both $49/month). But this Miamian does have access to two $0 deductible bronze plans for $5 or $6/month — not to mention a ridiculous eleven zero-premium bronze plans with high deductibles.

The zero-deductible bronze plans both have OOP maxes of $9,200, whereas the two cheapest silver plans for enrollees with income under 150% FPL in Miami this year have OOP maxes of $1,800 and $2,000 respectively. Forgoing CSR vastly expands financial risk. Nonetheless, in two of the largest ACA markets for people with income that would put them in Medicaid in expansion states, no-premium or ultra low-premium options would be available if they had to pay 2% of income.

Zero premium, but more risk

Expiration of the ARPA subsidy enhancements will exacerbate a negative long-term trend in marketplace coverage: reduced silver plan selection by enrollees with income low enough to qualify for strong CSR, which raises the actuarial value of a silver plan to 94% (at income up to 150% FPL) or 87% (at income from 150-200% FPL). That compares to 60-65% AV for enhanced bronze (the commonest bronze type now) or 80% for gold. Perhaps most damagingly, those who forgo strong CSR give up an OOP max that’s capped at $3,000, and averaged just $1,388 in 2024 at incomes up to 150% FPL), in favor of bronze and gold OOP maxes that usually top $7,000 and are often set at the maximum allowable $9,200.

At incomes up to 150% FPL, silver plan selection has dropped from 89% in 2017 to 76% in 2024 (see Table 2 below). At 150-200% FPL, silver selection has plummeted from 83% in 2017 to 57% in 2024. Competition among marketplace insurers to offer lowest-cost coverage has narrowed provider networks over time, and some enrollees may be forgoing CSR to access a lower level of coverage from an insurer with a more robust network. I look at such tradeoffs in specific markets here.

ARPA subsidy expiration will be a major blow to coverage availability for those who lack access to other affordable options. It will probably be combined with other blows: it may be coupled with a major Trump 2.0 effort to stand up and promote an alternative market of medically underwritten, lightly regulated plans, and/or, most damagingly, with major assaults on Medicaid eligibility and funding. “This won’t be the worst harm you suffer” is pretty cold comfort when staring down the barrel of an ignorant, cruel, corrupt government in formation. But the ACA marketplace proved resilient during the Trump 1.0 regime, and it may prove resilient once again.

Table 1: ACA Marketplace Enrollment Growth by Income, 2021-2024

Table 2: Silver Plan Selection at Incomes up to 200% FPL

CMS’s marketplace enrollment Public Use Files are available here.

Photo by cottonbro studio

If the ACA is not repealed

Trump said during the campaign that he would not work to repeal the Affordable Care Act. Of course, Trump’s promises are not worth the breath polluted by his utterance of them, and Republicans in Congress, flush with the Republican trifecta, are talking about block-granting Medicaid. that would probably go hand in glove with phasing out the ACA Medicaid expansion — the ACA’s most important and effective program, responsible for most of the drop in the uninsured rate achieved by the ACA.

But let’s undertake the dubious exercise of taking Trump at his word. Repealing and replacing (or, in the case of the Medicaid expansion, declining to replace) the ACA’s core programs may fail, as it did in 2017. Congress may take some hacks at the law, as they did in 2017 (remember the individual mandate? or the tax on insurers?) without changing the core subsidy structure of the ACA marketplace (minus the subsidy enhancements enacted in the American Rescue Plan, which expire in 2026) or entirely eliminating the Medicaid expansion. And the Trump administration may do much to reshape the marketplace administratively — as Trump 1.0 did, but much more. As I noted earlier this fall, JD Vance sketched out how Trump might, um, “build on” (that is, partially dismantle) the ACA marketplace.

Let’s look at the various means by which the Trump administration and Republican Congress might reshape the ACA’s core programs (Medicaid expansion and Marketplace) short of wholesale repeal.

Medicaid work requirements. Billed as a way to help “able-bodied” (read: unworthy, largely nonwhite) Medicaid enrollees discover the joys and dignity of wage labor, this Republican favorite is actually a means pushing eligible people off the rolls, as brief experiments in Trump 1.0 demonstrated. As the Kaiser Family Foundation never tires of documenting, a large majority of Medicaid expansion enrollees do work, and almost all the rest are otherwise constructively engaged, e.g., in school or caring for loved ones, or disabled. During Trump 1.0, CMS administrator Seema Verma enthusiastically promoted work requirements, inviting states to file waivers enacting them. While thirteen states had waivers approved, only Arkansas implemented them, with reporting requirements so onerous and so incompetently administered that the D.C. Circuit Court shut the program down after some 18,000 had been gratuitously disenrolled, holding that the waivers were unlawful because CMS failed to consider the impact on coverage, as required for Medicaid waivers. The pandemic then interrupted other states’ waiver plans, and the Biden administration removed approvals. Various court challenges were paused or dismissed as moot. Get ready for Round 2.

Medicaid de-expansion. The path on this front is laid out by the Trumpist Paragon Institute, headed by Brian Blase, who was on Trump’s NEC. The Paragon proposal, like the 2017 ACA repeal bills, would end the federal government’s 90% match rate (FMAP) for Medicaid enrollees rendered eligible by ACA criteria (eligibility for legally present adults (except recent immigrants) with income up to 138% of the Federal Poverty Level, reducing the FMAP to each state’s FMAP for other Medicaid programs, which ranges from 50% in the wealthiest (a.k.a. blue) states to 77% in Mississippi. The proposal would allow states to drop the eligibility threshold to 100% FPL, which would then be the starting point for marketplace subsidy eligibility, as it is now in the ten remaining states that have refused to implement the ACA expansion (Paragon notes gleefully that those states, which include Texas and Florida, will never enact the expansion under this proposal). In an extra swipe at wealthy (blue) states, Paragon would drop the minimum FMAP from 50% to 40%. This proposal nominally does not end the expansion, and allows states to avoid a “coverage gap” on the current Wisconsin model (Medicaid to 100% FPL, marketplace from 100% FPL). As the federal government pays 100% of marketplace premium subsidies, the shift of enrollees in the 100-138% FPL range offset a bit of the cost to states of the reduced match rate. But most states would probably find maintaining expansion at the reduced FMAP unsustainable — especially if the Republican Congress also imposes block grant funding or per capita caps on federal Medicaid spending, which would slow-strangle Medicaid over the course of a decade or two.

Undercutting the ACA marketplace. With Republicans in control of Congress, the enhanced premium subsidies enacted through 2022 by the American Rescue Plan Act and extended through 2025 by the Inflation Reduction Act are almost certainly dead. (Had Democrats won the House, they might have leveraged the expiration of Trump’s income tax cuts to preserve the ARPA subsidy boosts.) Those subsidy increases were doubtless the main cause of a 78% enrollment increase nationally from 2021-2024 and a 123% increase in the nonexpansion states, where the enhanced subsidies rendered benchmark silver coverage free for enrollees with income in the 100-150% FPL range. That means that the pre-ARPA income cap on subsidy eligibility, 400% FPL ($60,240 annually for a single person in 2024), will snap back into place, rendering ACA-compliant coverage unaffordable for several million people.

From there, a supercharged version of Trump 1.0’s parallel, ACA-noncompliant market will go to work to reduce the ACA marketplace to a sort of high risk pool. Vance sketched out how this might work, as I noted a few weeks ago. If more wholesale repeal/redesign does not happen, Trump 2.0 might

rebuild Trump 1.0’s alternative market of medically underwritten, ACA-noncompliant plans (so-called Short-Term Limited Duration, or STLD, plans*), and 2) prompting states to implement measures like the waiver concepts put forward by Trump’s former CMS administrator, Seema Verma. These “concepts” included 1) replacing ACA premium subsidies with a lump-sum health savings account that could be used to pay premiums for any plan; 2) inviting states to restructure the federal premium subsidy as they wished; 3) allowing states to grant premium subsidies for ACA-noncompliant plans; and 4) creating state high risk pools. Options 1 and 3 could effectively convert the ACA-compliant marketplace as we know it into a high risk pool of sorts, and in combination with option 4, could create the multiple stratified risk pools that Vance described in followup comments.

Verma’s waiver concepts plainly violated the ACA statute, as alternative state schemes outlined in an ACA Section 1332 “innovation waiver” proposal has to provide coverage at least as comprehensive as that stipulated in the ACA; provide coverage and cost sharing protections that are at least as affordable; and cover a comparable number of residents. Amending the waiver provision to allow Verma-esque concepts would be low-hanging fruit for a Republican Congress, and in fact was under negotiation after Republican repeal attempts failed in 2017.

Georgia made a brief attempt to take Verma up on her waiver concepts, filing a waiver proposal in late 2020 that would have eliminated a state-sponsored exchange, relying on commercial Enhanced Direct Enrollment (EDE) platforms commissioned by the federal government, and, in one early iteration, allow plans that did not include all Essential Health Benefits to be paid for with federal subsidies. That provision was cut from the submitted waiver, as it violates the ACA statute too plainly even for Trump 1.0’s administrators. But this may be the ACA’s future, if it has any future: some semblance of the ACA subsidy that can be used for ACA noncompliant plans — including, perhaps, medically underwritten plans. As those plans would be cheaper for healthy people, ACA-compliant marketplace enrollment might be reduced to those who would a) pay very little for comprehensive coverage because of low income and a favorable shakeout of price spreads, and/or b) those who know they need comprehensive coverage or who could not get a viable offer when subjected to medical underwriting.

The big enchilada, now as ever, is the ACA Medicaid expansion — and behind that, sustained federal funding for Medicaid programs generally. Perhaps Republicans will divide the baby, and cut the ACA expansion’s enhanced FMAP, say to 80% instead of 90% — and/or cut it disproportionately for states with higher per capita income. Under the best scenarios, the uninsured rate will rise substantially, and the individual market for health insurance will degrade considerably. Just one set of the many pillars of our society likely to come crashing down in coming months and years.

Update: It’s worth noting that when the Trump administration created an ACA-noncompliant market by extending the allowable limit of so-called Short-term, limited duration (STLD) plans to a full year, renewable twice, many states imposed or preserved strict limits of their own. (The STDL plans are medically underwritten, don’t have to cover the ACA’s ten Essential Health Benefits, and are not subject to medical loss ratio limits, e.g., the requirement to spend at least 80% of premiums on patients’ medical expenses.) As of this year, 14 states have either banned STDL plans outright or effectively regulated them out of existence, and others had either imposed other duration limits (often six months) or new coverage rules. If the ACA is not substantially repealed, state markets will doubtless diverge further in how they regulate and reshape coverage.

Update 2, 11/16/24: Charles Gaba, by means of Georgetown’s Edwin Park, runs down the more extreme scenario, Republicans enacting their proposals to destroy Medicaid

* Borrowing my last quick rundown of the Trump admin’s STLD program:

The Trump administration’s major initiative to “build on” the ACA marketplace after repeal failed was to stand up (by administrative rule in 2018) a parallel market of medically underwritten, lightly regulated plans by extending the allowable duration of already-existing so-called “short-term, limited duration plans” (STLD) to up to one year, renewable twice. The Obama administration had limited STLD duration to three months, though not until 2016. In combination with the Republican Congress’s zeroing out of the tax penalty for failing to obtain ACA-compliant insurance, the STLD market was an alternative for people who were priced out of the regulated ACA marketplace — as several million people were before the Biden administration removed the income cap on subsidy eligibility via the American Rescue Plan Act (ARPA) in March 2021. (The ARPA subsidy enhancements were temporary, and extended by the Inflation Reduction Act only through 2025.)
STLD plans can refuse access to people with pre-existing conditions or exclude coverage for the condition. They do not have to cover the ACA’s Essential Health Benefits and generally offer very limited prescription drug coverage, if any, and no substance abuse coverage. They are not subject to the ACA requirement to spend at least 80% of premiums on members’ medical bills (and on a few allowed other expenses) and have been reported to spend as little as 45% of premiums on claims. They do not have to offer a provider network and can pay providers what they deem appropriate, exposing enrollees to balance billing. They do not have to provide an annual out-of-pocket cost cap on covered benefits, though some do. They are much like the plans offered in the pre-ACA individual market.