How many people could become uninsured if enhanced ACA subsidies expire in 2026 according to different analysts?

Executive summary

Analysts place the number of people who could become uninsured in 2026 if the enhanced ACA premium tax credits lapse between roughly 2.2 million and nearly 5 million, with most high-profile estimates clustering around 4–4.8 million; differences reflect divergent modeling assumptions about enrollee behavior, the base year of enrollment, and whether people shift to other coverage or drop insurance entirely ^{[1] [2] [3] [4]}. The range matters: a low-end CBO-like outcome implies a modest reversal of recent gains in coverage, while Urban Institute and Commonwealth estimates imply a dramatic backslide of several million losing insurance and large spillovers for hospitals and labor markets ^{[4] [2] [3]}.

1. CBO and government-linked estimates: 2.2 million to ~3.8 million

The Congressional Budget Office has been cited with multiple estimates that differ by release and context; one widely reported figure—2.2 million people projected to lose insurance in 2026—appears in reporting that cites a December brief to Congress (AHA News summarizing CBO) and has circulated in policy coverage ^[1]. Other CBO references in secondary reporting and trackers put the agency’s projection higher—around 3.8 million additional uninsured on average each year over 2026–2034—showing that even within CBO outputs the estimate can vary depending on the specific scenario and averaging period used ^[4]. In short, CBO-based public estimates land from roughly 2.2 million to 3.8 million, with the lower number tied to particular short-run assumptions and the higher tied to multi‑year averaging and different behavioral responses ^{[1] [4]}.

2. Urban Institute and academic work: 4.8 million

The Urban Institute’s September analysis projects that 4.8 million more people would become uninsured in 2026 if enhanced premium tax credits are allowed to expire, and that 7.3 million would lose subsidized marketplace coverage—figures driven by higher-than-expected 2025 enrollment and detailed microsimulation of take-up and affordability ^{[2] [5]}. Urban’s modeling emphasizes that many people who gained zero- or low-premium plans under the enhancements would find nongroup coverage unaffordable under standard PTCs and either forego insurance or be priced out entirely, which is why its estimate sits well above the lower CBO figure ^[2].

3. Think tanks, health foundations and near‑topline press: ~4 to nearly 5 million

A cluster of nonpartisan and advocacy-linked analyses echo Urban or push the estimate slightly higher: Commonwealth Fund and several policy briefs report “nearly 5 million” could become uninsured in 2026 absent an extension, while outlets summarizing multiple analysts commonly use “roughly 4 million” as a shorthand ^{[3] [6] [7]}. Health policy shops such as KFF emphasize the scope of people affected by premium increases (millions seeing costs rise) but focus their numeric claims on enrollment and premium impacts rather than a single uninsured tally ^{[8] [9]}.

4. Why estimates diverge — enrollment baselines, behavior, and timing

Differences in headline numbers reflect three predictable drivers: which enrollment baseline is used (2024 vs. 2025 plan selections), assumptions about how many people will switch to other coverage (employer-sponsored plans, Medicaid) versus become uninsured, and whether analysts model one-year shocks or average effects over a multi‑year window; Urban used a large 2025 enrollment base and assumed limited substitution, producing 4.8 million uninsured, while some CBO products used alternate assumptions and shorter windows that produced 2.2–3.8 million ranges ^{[5] [4] [1]}.

5. State- and subgroup-level nuance changes the picture

State analyses and exchange operators show concentrated impacts: California warned up to 400,000 could become uninsured without state action, and analysts note disproportionate effects on middle‑income households between 250–400% FPL and on communities of color—details that matter for policy responses even if national tallies differ ^{[10] [11] [2]}. These subnational figures underscore that national ranges conceal sharp local variation driven by state policy choices and population composition ^{[10] [2]}.

6. Bottom line and limits of available reporting

The best read of the reporting is a bounded but consequential range: roughly 2.2 million on the low end (per some CBO readouts) to about 4.8–5.0 million on the high end (Urban Institute and Commonwealth/others), with many independent sources and media accounts using “around 4 million” as a central shorthand for public discussion ^{[1] [2] [3] [7]}. This summary relies on published analyses; it cannot adjudicate which projection is “correct,” only map why and how estimates differ across credible analysts and which assumptions drive the spread ^{[4] [5]}.