The Subsidy Cliff and the Risk Pool: An Actuarial Analysis of Adverse Selection and Premium Stability in the ACA Individual Market after the Expiration of the Enhanced Premium Tax Credits

Abstract

Background. The enhanced Affordable Care Act premium tax credits expired on 31 December 2025, restoring the original
contribution schedule and the cliff at 400% of the federal poverty level.
Methods. Net premiums were computed for every HealthCare.gov rating area from CMS public use files. A Tweedie cost model
was estimated on 68,119 MEPS person-years and a selection-and-repricing model was calibrated to 2026 plan selections.
Income-band responses were tested against 2024 to 2025 as a placebo, in state-by-band difference-in-differences regressions
with the unsubsidized premium burden as a dose, and in a county panel. Loss ratios for 2021 to 2024 and the cost per personyear
of four restoration schedules were computed.
Results. The median benchmark premium rose 24.3%; crossing 400% of poverty costs a 40-year-old $1,487 and a 60-year-old
$10,163. In 30 HealthCare.gov states, selections above 400% fell 34.6% against 3.5% below, after rising 9.0% in the placebo
year. A single price response calibrated to the total predicts a 5.1% loss above the threshold; fitting the segments separately
requires responses differing by a factor of 11. Selections at 300 to 400% of poverty exceed their counterfactual by 122,752,
71% of the loss immediately above. The exposed segment’s loss rose 3.25 points per point of unsubsidized burden; the placebo
year shows no gradient. Full restoration costs $30,786 per person-year of coverage; extending the 9.96% cap above 400%
costs $9,636.
Conclusions. The subsidized market absorbed the reversion; the unsubsidized segment contracted sharply, partly through
income reporting at the notch. No single elasticity describes both.