From reviewing actuarial memoranda across multiple ACA filing cycles, the 2026 rate season stands apart. The combination of a discrete policy discontinuity (enhanced premium tax credit expiration), compounding medical trend acceleration (GLP-1 drugs, specialty pharmacy), and a new layer of trade-policy uncertainty (tariffs on medical devices) created a pricing environment where standard trend-projection methods were insufficient. Pricing actuaries had to build explicit adverse selection adjustments from limited analogue data, and the results varied widely: premium changes ranged from –10% to +59% across the 312 insurers that filed 2026 rates (Peterson-KFF Health System Tracker).
This article breaks down the four-step methodology that health pricing actuaries used to construct adverse selection morbidity adjustments, based on a review of 105 actuarial memoranda from 20 jurisdictions published by the Peterson-KFF Health System Tracker, supplemented by Georgetown CHIR’s analysis of 178 filings across 28 states and the Urban Institute’s decomposition of the 21.7% average premium growth figure.
Step 1: Segmenting the Enrolled Population by Subsidy Sensitivity
The first step in pricing adverse selection from a subsidy cliff is identifying which enrollees are most likely to leave. Enhanced premium tax credits, enacted under the American Rescue Plan Act in 2021 and extended through 2025, capped marketplace premiums at 8.5% of household income for all enrollees and eliminated premiums entirely for those below 150% of the federal poverty level. Expiration means net-of-subsidy premiums increase sharply for subsidized enrollees, with the CBO projecting marketplace enrollment falling from 22.8 million to approximately 18.9 million.
Pricing actuaries segmented their enrolled populations along several dimensions to estimate subsidy sensitivity:
Income distribution and subsidy magnitude. Enrollees receiving the largest dollar-value subsidies face the steepest premium increases when enhanced credits revert to pre-2021 levels. Members below 200% FPL, who often paid zero or near-zero net premiums under enhanced credits, face the sharpest affordability shock. Actuarial memoranda from Connecticut, for example, cited 3.7% to 6.8% additional premium increases across carriers specifically attributable to the subsidy cliff (Peterson-KFF).
Age and metal tier. Younger enrollees choosing bronze and silver plans tend to exhibit higher price elasticity of demand. They are more likely to be healthy enough to forgo coverage when premiums rise. The ACA’s 3:1 age-curve constraint means younger member departures do not proportionally reduce premium revenue, because their premiums are already compressed relative to expected costs. Losing young, healthy enrollees disproportionately worsens the risk pool’s average morbidity.
Claims history as a proxy for price sensitivity. Several insurers stratified their books by prior-year risk scores, applying differential lapse assumptions by decile. MVP Health Care in Vermont assumed that healthy subsidized individuals would drop coverage “at twice the rate of other subsidized individuals” (Georgetown CHIR). This two-to-one differential lapse ratio was common across filings, though the exact multiplier varied by carrier and state.
Step 2: Modeling the Morbidity Differential Between Leavers and Stayers
Once the likely-to-leave population is identified, the next step is quantifying the morbidity difference between those who exit and those who remain. This is the core of the adverse selection adjustment: the remaining risk pool is sicker, on average, than the pre-expiration pool.
Actuarial memoranda consistently described this dynamic using the same directional language. Optimum Choice in Maryland noted that “healthier members to leave at a disproportionately higher rate than those with significant healthcare needs, increasing market morbidity” (Peterson-KFF). Community Health Option in Maine anticipated that the “remaining risk pool in 2026 will have higher healthcare needs, on average, as healthier consumers more likely to lapse” (Georgetown CHIR).
The quantification method typically involved claims stratification by risk score decile:
- Bottom three deciles (lowest-morbidity members): assumed 40–60% lapse rates, representing healthiest, most subsidy-sensitive enrollees
- Middle four deciles: assumed 15–25% lapse rates, reflecting moderate health needs and mixed price sensitivity
- Top three deciles (highest-morbidity members): assumed 5–10% lapse rates, reflecting members whose healthcare needs make coverage essential regardless of premium
The resulting shift in average plan-level risk scores depends on the specific enrollment distribution and lapse assumptions, but the industry consensus centered on a 3–5 percentage point morbidity deterioration factor. UnitedHealthcare applied a 1.044 adjustment factor (4.4 percentage points) specifically for the subsidy expiration effect (Peterson-KFF). Blue Cross Blue Shield of Vermont included “an additional 6.6 percent” from enhanced PTC loss within its 23.3% total rate increase (Georgetown CHIR).
Step 3: Translating Morbidity Shift into Rate-Level Impact
The ACA’s single risk pool requirement means that all individual market enrollees within a state are rated together, with premiums varying only by age (3:1 band), tobacco use, and geography. This constraint means the morbidity shift cannot be isolated to specific sub-pools; it flows through the entire rating structure.
The translation from morbidity factor to rate impact follows a multiplicative chain:
| Component | Typical Range | Mechanism |
|---|---|---|
| Baseline medical trend | 8–10% | Unit cost inflation, utilization growth, mix shift |
| Adverse selection morbidity | 3–5 pp | Risk pool composition change from subsidy cliff |
| Marketplace Integrity Rule | 0.25–1 pp | Tighter eligibility verification, additional healthy-member attrition |
| GLP-1 / specialty drug layer | 1–3 pp | Pharmacy trend in excess of baseline (detail below) |
| Tariff adjustment | 0–3 pp | Medical device and pharmaceutical supply cost increases |
The interquartile range of total proposed rate changes fell between 12% and 27%, consistent with a base trend of 8–10% compounded by 4–8 points of policy-driven adjustments and 1–3 points of pharmacy acceleration (Peterson-KFF). The median of 18% suggests that most actuaries applied the adverse selection load conservatively rather than at the upper bound.
One important technical detail: because the 3:1 age curve is fixed, a morbidity shift concentrated among younger enrollees (who have lower rating factors) has a smaller dollar impact per departing member than a shift among older enrollees. Actuaries modeling the subsidy cliff had to account for the age distribution of the leaving population and its interaction with the compressed age curve. The net effect is that a 10% enrollment decline concentrated among 25-to-34-year-old bronze plan members produces a smaller per-member-per-month cost increase than a 10% decline distributed evenly across all ages.
How Risk Adjustment Interacts with Shrinking Enrollment
The ACA risk adjustment transfer formula is designed to neutralize selection differences between plans within a market, not to offset market-wide adverse selection. This distinction is critical for understanding why risk adjustment does not eliminate the morbidity-driven premium increase.
The transfer formula calculates each plan’s payment or charge based on the difference between its average risk score and the statewide average risk score, multiplied by the statewide average premium. When enrollment shrinks and the remaining population is sicker:
- The statewide average risk score increases (sicker remaining pool)
- The statewide average premium increases (higher costs flow through rates)
- Plans that previously received transfer payments may see those payments shrink if the market-wide risk level rises to meet their own risk profile
- Plans that previously made transfer payments may see their charges increase if their risk score falls further below the new, higher statewide average
The net result is that risk adjustment redistributes costs across plans but does not inject new money into the market. When the entire pool deteriorates, every plan faces higher expected per-member costs, and risk adjustment transfers recalibrate around a higher baseline. A plan whose membership was already relatively sick may actually see reduced transfer receipts, because the statewide average moved toward its own risk profile. Conversely, a plan that retained a healthier-than-average book faces larger transfer charges.
This dynamic creates a secondary pricing uncertainty: actuaries must project not only their own plan’s enrollment composition but also the statewide average risk score and statewide average premium, both of which depend on every other insurer’s assumptions about the same adverse selection. The circularity is inherent to the ACA risk adjustment design, and it widens the confidence interval around plan-level rate adequacy.
The GLP-1 and Specialty Drug Trend Layer
Separate from the adverse selection adjustment, many insurers added explicit loads for GLP-1 receptor agonist cost acceleration. The Peterson-KFF analysis documented that MVP Health Care in Vermont reported GLP-1 allowed costs increasing 25–30% per quarter throughout 2024, with Q4 2024 costs “nearly double” the 2023 total. Kaiser in Washington projected script utilization increases of 18% in 2025 and 7% in 2026, driven by oncology and GLP-1 drugs. Blue Cross Blue Shield of Massachusetts reported over $300 million in 2024 GLP-1 spending and estimated that discontinuing weight-loss GLP-1 coverage reduced its rate increase by approximately 3 percentage points.
For pricing actuaries, separating the GLP-1 trend from the adverse selection morbidity adjustment is essential. Both factors increase expected per-member-per-month costs, but through different mechanisms. GLP-1 costs reflect utilization growth and unit cost increases for an expanding drug class. The morbidity adjustment reflects a compositional shift in the risk pool. Double-counting is a real risk: if an actuary builds GLP-1 trend into the base medical/pharmacy trend assumption and also includes GLP-1 utilization growth within the morbidity adjustment (because sicker stayers may have higher GLP-1 utilization), the rate filing overstates expected costs.
Specialty drugs more broadly accounted for 2% of utilization but “more than 50 percent” of drug spend for at least one filing reviewed (Excellus, per Peterson-KFF). This concentration effect amplifies the impact of any compositional shift in the risk pool: if stayers have disproportionately high specialty drug utilization, the per-member pharmacy cost increase from adverse selection exceeds the medical cost increase.
Tariff and Regulatory Uncertainty Adjustments
A subset of insurers added a separate tariff adjustment, averaging approximately 3 percentage points where applied, to account for anticipated cost increases on imported medical devices, pharmaceuticals, and supplies (Peterson-KFF). In Rhode Island, one insurer added a 3% pharmacy trend bump and a 0.5% CPI adjustment specifically for tariff effects. The majority of insurers, however, chose not to include a tariff adjustment, citing uncertainty about implementation timing, scope, and pass-through dynamics.
The CMS Marketplace Integrity and Affordability Rule introduced additional pricing uncertainty. Celtic Insurance Company (Ambetter) in Texas noted that stricter scrutiny on subsidy eligibility “may lead to further enrollment erosion among the healthier populations,” compounding the subsidy-cliff morbidity adjustment (Georgetown CHIR). Most actuarial memoranda treated the Marketplace Integrity Rule as adding 0.25% to 1 percentage point of additional morbidity deterioration on top of the subsidy expiration effect.
Credibility Challenges: Pricing Without Precedent
The fundamental actuarial challenge in the 2026 filing cycle is that no prior experience data exists for this specific policy discontinuity. Enhanced premium tax credits were enacted in 2021 and remained in effect continuously through 2025. There is no observation period where they were removed and then restored; the closest analogues are:
- Pre-2021 enrollment patterns: Before enhanced credits, ACA marketplace enrollment was lower and the risk pool was different, but the policy and competitive environment also differed substantially
- CSR payment cessation (2017): The abrupt termination of cost-sharing reduction payments created a partial analogue for subsidy disruption, but the mechanism (silver loading) and population affected were different
- Wakely Consulting projections: Georgetown CHIR cited Wakely projections of a 47–57% average enrollment decline, with non-expansion states facing up to 64% losses. If realized, these declines would reduce marketplace enrollment to levels “not seen since early years of Marketplaces, if not lower”
Under ASOP No. 25 (Credibility Procedures), actuaries must disclose when limited data requires reliance on professional judgment rather than statistically credible experience. The 2026 filings represent a case where every insurer’s morbidity adjustment reflects substantial professional judgment, constrained by directional reasonableness checks against the analogues listed above but not calibrated to directly observed experience.
Several states required dual-rate submissions: one assuming enhanced PTCs continue, one assuming expiration. Connecticut data shows the expiration scenario added 3.5% to 6.8% to carrier-level rate increases. Washington estimated the difference at up to 6.4 percentage points. These dual filings provide a useful, if imperfect, measure of the pure adverse selection load, isolated from the baseline trend component.
Why This Matters for Pricing Actuaries
The 2026 ACA filing cycle is a case study in pricing a discrete, policy-driven adverse selection event without credible experience data. Three dimensions of this problem deserve ongoing attention from health pricing actuaries.
The feedback loop between adverse selection and premium adequacy. If morbidity adjustments are too conservative and actual adverse selection exceeds projections, carriers face reserve deficiency and potential market exit. Aetna’s withdrawal from 17 states affecting over 1 million consumers illustrates the stakes (Georgetown CHIR). If morbidity adjustments are too aggressive, premiums accelerate the disenrollment spiral they were designed to price, as higher rates push additional marginal enrollees out of the market.
The risk adjustment recalibration problem. As the market shrinks and average acuity rises, the statewide risk adjustment parameters shift. Actuaries projecting 2027 rates will have the first full year of post-subsidy-cliff experience data, but that data reflects a market still adjusting to the new enrollment equilibrium. The credibility challenge persists into subsequent filing cycles until enrollment and morbidity stabilize.
The separation of morbidity, trend, and policy loadings. Regulators reviewing 2026 filings face the challenge of evaluating whether carriers double-counted the GLP-1 trend within both the base pharmacy assumption and the adverse selection morbidity adjustment. State rate review programs, which scrutinize any proposed increase above 15%, lack a standard benchmark for adverse selection loads in a subsidy-cliff scenario. The tension between actuarial conservatism (protecting solvency) and consumer affordability (preventing a death spiral) is inherent to this pricing problem, and it will define the regulatory posture for the 2027 cycle as well.
Further Reading
- ACA Benchmark Premiums Jump 21.7% in Largest Surge Since 2018 – The carrier-level decomposition of 312 rate filings that provides the market context for the adverse selection adjustments analyzed here, including GLP-1 cost drivers and tariff loads.
- ACA Marketplace 2026: Subsidy Cliff, Enrollment Shock, and Actuarial Implications – The foundational analysis of enhanced PTC expiration, enrollment trajectories, and state-level responses that set up the adverse selection problem priced in this filing cycle.
- Healthcare Cost Trends 2026: Forces Reshaping Medical Spending – The 8.5–9.5% baseline medical trend projections and GLP-1 pharmacy trend acceleration that form the non-selection component of 2026 ACA rate increases.
- ACA 2027 Rate Filings: Pricing Actuaries Face a GLP-1 Credibility Problem – How ASOP No. 25 credibility procedures apply to the GLP-1 pharmacy trend loading, with methodology parallels to the adverse selection credibility challenge in 2026 filings.
- ACA 2027 Proposed Rule: Actuaries Model a 2M Enrollment Drop – How the 2027 NBPP layers additional enrollment contraction and wider de minimis ranges on top of the subsidy cliff documented here.
Sources
- Peterson-KFF Health System Tracker: How Much and Why ACA Marketplace Premiums Are Going Up in 2026 – Analysis of 105 actuarial memoranda from 20 jurisdictions and 312 rate filings across all 50 states and DC.
- Georgetown CHIR: Early 2026 Rate Filings Show Marketplace Policy Changes Contribute to Eye-Popping Rate Increases – Review of 178 actuarial filings across 28 states with carrier-specific morbidity assumptions.
- Urban Institute: Understanding the Extraordinary Increase in ACA Premiums in 2026 – Decomposition of the 21.7% average premium growth into medical trend and adverse selection components.
- Congressional Research Service: Enhanced Premium Tax Credit and 2026 Exchange Premiums (R48290) – FAQ on enhanced PTC mechanics, 2026 applicable percentages, and benchmark plan premium calculations.
- CMS Rate Review Database – Federal repository of rate filings subject to the 15% threshold review requirement.
- MoneyGeek: ACA Premiums 2026, 50-State Analysis – State-level premium change data and affordability impact projections.
- Risk Transfer Formula for Individual and Small Group Markets Under the Affordable Care Act (PMC) – Technical specification of the ACA risk adjustment transfer formula mechanics.