ACA marketplace renewals: reflections on retention and year-round enrollment
Writing about the ACA marketplace at this point has an elegiac feel (as does writing about almost any not-depraved or degraded aspect of American life). As of 2026, coverage is likely to be far less affordable, out of pocket costs will rise as more enrollees fall back on bronze plans, and enrollment is likely to drop precipitously. And that’s a near-best-case scenario, barring repeal or some kind of not fully imaginable regulatory assault (or judicial assault, e.g., on free preventive care).
That said, I’ve been mulling this year’s new peak in enrollment, driven primarily by a large number of renewals. That’s fueled not only by the 31% year-over-year enrollment surge in OEP 2024 but by continuous off-season enrollment for applicants with income under 150% FPL (in place since early 2022) and by off-season enrollment from those disenrolled from Medicaid during the unwinding, which was essentially done by last summer. (New enrollment dropped from 5.2 million in OEP 2024 to 3.9 million in OEP 2025.)
All that said, I wanted to look at year-round enrollment patterns through the Trump and Biden eras. I no longer refer to the January-to-December enrollment fluctuations as a question of “retention,” as there has been so much off-season enrollment since 2022; a slight drop or even increase from January to December does not give a clear picture of how long people are remaining enrolled. On the other hand, one perhaps-obvious fact that jumps out from the table below is that in all years, almost everyone who is enrolled as of December renews, actively or passively. And in the Biden years, thanks mainly to the ARPA subsidy boosts, fall-off from the end of OEP to “early effectuated enrollment in February is minimal - -that is, the auto-re-enrolled do not drop coverage in droves when the first premium payment (often $0, thanks to ARPA) is due (retention also improved through the Trump years, as cuts to outreach and marketing and a shortened OEP apparently culled more marginal enrollees).
Below, I’ve charted enrollment renewals in each year (2017-2025) against various points in the prior year: end of OEP, early effectuated enrollment (February in the Trump years, March in the Biden and Obama years), September (to capture a sense of post-OEP enrollment in Biden years especially) and December.
The sources are CMS’s annual early effectuated enrollment snapshots (links below — as far as I know they’re not indexed on one page) and enrollment public use files for end-of-OEP totals.
ACA marketplace renewals as a percentage of prior-year enrollment points
I’m not sure how illuminating this is, TBH. A few observations below. Maybe others will find other figures in the carpet.
The jump in percentages in the Biden reflects not only improved retention (as high percentages of enrollees were paying either $0 premiums or less than $10/month) but also full-year enrollment for many prospective enrollees (e.g., all enrollees during the emergency SEP of 2021, low-income enrollees from early 2022 forward, and Medicaid “unwindees” in 2023-25).
Since 2021, enrollment has either increased or barely dropped from March to September (there was also high off-season enrollment in 2020, as the state marketplaces opened emergency SEPs, HealthCare.gov streamlined individual SEPs, and Covid led to mass if mostly short-lived layoffs).
In all years, almost everyone who retains coverage through December renews. And as I’ve noted previously, the drop in enrollment from the end of OEP to the first effectuated enrollment snapshot is also quite high and has also fallen steadily since 2016 — indicating that those who are passively auto-renewed mostly retain coverage for at least some months in the new year. Thus retention as well as year-round new enrollment are part of the year-long stability of enrollment totals in recent years.
What else?
- - -
Here are the effectuated enrollment reports:
2016 https://downloads.cms.gov/files/effectuated-enrollment-snapshot-report-06-12-17.pdf
2017 https://www.cms.gov/cciio/programs-and-initiatives/health-insurance-marketplaces/downloads/2018-07-02-trends-report-1.pdf
2018 https://www.cms.gov/sites/default/files/2019-08/08-12-2019%20TABLE%20Early-2019-2018-Average-Effectuated-Enrollment.pdf
2019 https://www.cms.gov/cciio/resources/forms-reports-and-other-resources/downloads/early-2020-2019-effectuated-enrollment-report.pdf
2020 https://www.cms.gov/document/early-2021-2020-effectuated-enrollment-report.pdf
2021 https://www.cms.gov/files/document/early-2022-and-full-year-2021-effectuated-enrollment-report.pdf
2022 https://www.cms.gov/files/document/early-2023-and-full-year-2022-effectuated-enrollment-report.pdf
2023 https://www.cms.gov/files/document/early-2024-and-full-year-2023-effectuated-enrollment-report.pdf
2024 https://www.cms.gov/files/document/early-2024-and-full-year-2023-effectuated-enrollment-report.pdf
Mar 4, 2026:
As I think I commented on one of your other posts, I've been trying to get a better fix on those who dropped ACA on-exchange coverage. (Better than the current 1.2 to 1.4 million less enrolling during open enrollment for 2026 vs 2025.)
I had a problem with that 1.2 million to 1.4 million number, because there was a big annual uptrend 2023 to 2024 (5 million) and 2024 to 2025 (3 million) , due, apparently, to people either finding out about reasonably-priced coverage available or reconsidering from a prior stance, and deciding that health insurance, at the available price, might be good to have after all. (And this factor causing the uptrend should still be going on, even with somewhat higher prices.)
To de-confound the two factors, I thus tried to look at non-re-enrollments, and got:
"ACA Exchanges: I get 3.5 million More (4.7 million) non-returned in 2026 (vs. 1.2 million in 2025)"
https://normspier828307.substack.com/p/aca-exchanges-i-get-35-million-more
Now, since the CMS OE reports do not give you the number with end-of-prior-year enrolled status (that is, the count of possible re-enrolls), I used prior year open enrollment enrolled. Which could be a little off, though, hopefully the error each year is similar, due to the difference being always within expanded subsidy years.
But, from a little poking around, I see that, if I wait a few weeks or month for the next-year version of this
https://www.cms.gov/files/document/effectuated-enrollment-early-snapshot-2025-and-full-year-2024-average.pdf ,
then I will have much closer to the 2025 and 2026 "potential returner" s. (Dec. effectuated.)
(I guess I'll have to do a revision of my 3.5 million when it comes out.)
On search, I also caught that you routinely publish what will be my revised number, which you, in fact, have on a table in the post I am commenting on.
Though, I think that this year, as it catches those chased away by the higher prices from the expired expanded subsidies, and separates ("unconfounds") those people from those other people deciding to enroll for the first time despite the higher prices, it has a special usefulness that I don't know of any other way to capture.