What do Congressional Budget Office (CBO) or Congressional Research Service (CRS) estimates say about ACA-related coverage losses in 2025?
Executive summary
CBO estimated that letting enhanced ACA marketplace premium tax credits expire would increase the number of uninsured by an average of about 3.8 million people per year from 2026–2034 and that, in analyses of the 2025 reconciliation legislation, CBO projected a net increase in the uninsured of roughly 10.9 million (on top of 5.1 million in its baseline for a total of about 16 million in some summaries) with ACA-marketplace changes accounting for roughly 2.1–4.2 million of that rise depending on the calculation [1] [2] [3] [4]. Coverage estimates vary across CBO, CRS summaries, and secondary analyses and hinge largely on whether one isolates the expiration of enhanced tax credits, the reconciliation bill’s marketplace provisions, Medicaid changes, or interactions among provisions [1] [4] [3].
1. What the CBO explicitly estimated about the 2025 subsidy expiration
CBO told congressional requesters that if the enhanced premium tax credits enacted in ARPA and extended through 2025 are not continued, the number of uninsured would rise substantially: CBO reported an average annual increase in the uninsured of about 3.8 million from 2026–2034 and highlighted steep premium increases for marketplace enrollees if the credits expire [1] [5]. Other CBO summaries presented the expiration’s effect as 4.2 million more uninsured in 2034 specifically, reflecting year‑by‑year and endpoint framing differences [4].
2. How the reconciliation bill changed the arithmetic: CBO’s broader estimate
When CBO scored the 2025 reconciliation (so‑called One Big Beautiful Bill) it estimated the law would increase the uninsured by 10.9 million in 2034 relative to the baseline — a figure presented as the incremental effect of the bill’s provisions — and commentators often add CBO’s baseline‑projected coverage losses (another 5.1 million) to describe a combined 16 million increase in the uninsured under some framings [2]. CBO’s breakdown attributes most of the increase to Medicaid cuts but also to Marketplace changes; KFF and other summaries allocate portions of the total across Medicaid and ACA marketplace effects [3] [4].
3. Where CRS reporting fits into this picture
CRS products summarize and tabulate CBO estimates and committee impacts; a CRS table cited CBO’s estimate that a particular amendment would reduce federal Medicaid outlays and lead to 4.8 million people losing insurance in 2034 — illustrating CRS’s role in organizing CBO figures for lawmakers and the public rather than producing independent microsimulation projections [6]. CRS and Congress.gov materials therefore relay CBO’s core numbers while linking them to specific bill provisions.
4. Why different numbers (2.1M, 3.8M, 4.2M, 10.9M, 16M) appear
The divergence reflects which phenomenon is being counted and the time frame: (a) the marketplace premium tax‑credit expiration alone is often framed as ~3.8 million average annual uninsured increases (2026–2034) or 4.2 million in 2034 [1] [4]; (b) CBO’s estimate of the reconciliation bill’s net effect is 10.9 million uninsured in 2034 from the bill’s combined Medicaid and Marketplace changes [2]; (c) some summaries allocate CBO’s combined 10‑plus‑million effect across Medicaid (~7.5–7.8 million) and ACA marketplace (~2.1–2.4 million) components, producing the smaller ACA‑specific figure of ~2.1–2.4 million in some state‑level breakdowns [3] [7]. Adding CBO’s baseline trend (5.1 million) to the incremental 10.9 million yields higher “total” tallies that appear in some analyses [2].
5. Alternative estimates and research groups that are frequently cited
Independent modelers such as the Urban Institute and state‑level analysts have produced higher marketplace‑coverage loss estimates in some reports (e.g., millions losing ACA coverage in 2026), and organizations like the Commonwealth Fund, KFF, and Georgetown’s Center for Children and Families have used CBO numbers combined with other models to produce state or sector estimates — but those specific alternative numbers are documented in secondary analyses that often rely on or reconcile with CBO’s core scoring [8] [4] [7]. Available sources do not provide detailed independent microsimulation methods here; instead they reference and repackage CBO findings [8] [4].
6. Caveats, political context, and why estimates can be weaponized
CBO produces standardized budget‑window projections; CRS organizes those findings for lawmakers. Both agencies do not take policy positions, but their numbers are used by advocates and lawmakers with opposing agendas: proponents of extending subsidies cite CBO’s millions‑uninsured warnings to urge action, while supporters of the reconciliation bill emphasize projected deficit savings and premium effects cited by CBO and JCT [9] [10]. Analysts also warn that interactions among provisions (Medicaid cuts, eligibility rules, marketplace integrity changes) complicate attribution and that year‑specific vs. multi‑year averages yield different headline figures [3] [7].
7. Bottom line for readers
For 2025 coverage: CBO’s central messages in available reporting are [11] letting the enhanced premium tax credits expire would cause several million people to lose marketplace coverage (commonly summarized as ~3.8 million annually or ~4.2 million in 2034), and [12] the broader 2025 reconciliation law was scored by CBO as increasing the uninsured by about 10.9 million in 2034 with Medicaid provisions accounting for most of that rise and marketplace changes contributing a smaller but material share [1] [2] [3]. Readers should note that different framings (annual average vs. endpoint year vs. incremental vs. total vs. allocated across programs) drive the varying headline numbers in the public debate [2] [4].