Washington 2027 ACA Filings Reveal Post-Subsidy Adverse Selection Load

From tracking ACA rate filings across multiple cycles, the morbidity adjustment component has always been the most carrier-specific input to the index rate development. Medical trend varies within a manageable band. Pharmacy trend, outside of GLP-1 disruption, follows observable pipelines. But the morbidity adjustment, the factor that captures who is in the risk pool and how sick they are relative to the pricing base period, diverges dramatically when enrollment is shifting. The May 26 Washington filings are the clearest illustration yet of that divergence. Thirteen carriers filed requests ranging from 8.6% to 27.8%, and the gap is not primarily about trend disagreements. It is about fundamentally different assumptions regarding which enrollees leave the market when subsidies shrink.

Decomposing the 22.4% Average: Three Layers of Rate Development

A 22.4% average rate increase does not mean every dollar of the increase comes from one source. Pricing actuaries building the 2027 index rate for Washington's individual market decompose the total adjustment into at least three components, each requiring independent assumption development and documentation in the actuarial memorandum filed with SERFF.

The first layer is medical and pharmacy trend, covering unit cost inflation and utilization changes. National benchmarks from Milliman, PwC, and Segal place commercial medical trend in the 7% to 9% range for 2027, with pharmacy trend running higher at 10% to 14% depending on GLP-1 and specialty drug assumptions. A carrier projecting 8% blended trend contributes roughly 8 percentage points of the 22.4% average.

The second layer is the morbidity adjustment factor. This captures the expected change in average risk score or allowed PMPM attributable to the compositional shift in the enrolled population. When 13% of Washington exchange enrollees drop coverage after EPTC expiration, the remaining pool is not a random subsample of the original. Healthier, younger, more price-sensitive individuals disproportionately exit when their net premium rises. The remaining enrollees carry higher average morbidity, and the pricing actuary must estimate the magnitude of that shift. This adjustment typically adds 3 to 8 percentage points to the rate, depending on the carrier's enrollment profile.

The third layer is an explicit adverse selection loading that accounts for the feedback loop between premium increases and further enrollment attrition. Higher rates cause additional healthy members to leave, which further worsens the pool's morbidity, which requires still-higher rates. Carriers with larger on-exchange, subsidy-dependent populations must load more heavily for this spiral risk than carriers with commercially stable enrollment.

Why the Carrier-Level Spread Is 19 Points Wide

The 8.6%-to-27.8% range across the thirteen Washington filers is not noise. It reflects structural differences in each carrier's book of business that produce legitimately different morbidity assumptions.

Coordinated Care filed at 27.8% with 97,979 enrollees, the largest individual market carrier in Washington. Its enrollment is heavily concentrated on-exchange and disproportionately subsidized. When enhanced APTCs expired, Coordinated Care's population faced the largest net premium shock. The carrier's pricing actuaries must assume that the healthiest slice of their 97,979-member base, the younger adults who were paying $0 to $50 per month after subsidies and now face $200 to $400, will leave. The remaining pool shifts older and sicker, producing a morbidity adjustment factor that may exceed 1.08 (an 8% increase in average allowed PMPM from composition alone, before any medical trend is applied).

Kaiser Northwest filed at 9.5% with 8,180 enrollees. Kaiser's integrated delivery model anchors a membership base that is less subsidy-dependent and more commercially stable. Its enrollees chose Kaiser for the delivery system, not solely for the lowest-premium option. When subsidies shrink, Kaiser's lapse rate is lower, and the composition of its remaining pool shifts less dramatically. Its morbidity adjustment factor may be closer to 1.01 or 1.02.

Regence BlueShield filed the lowest request at 8.6%. Regence's individual market block includes a meaningful off-exchange component where enrollees have been paying full premium without subsidies throughout the EPTC period. These members are already self-selected for willingness to pay, and their retention through subsidy changes is expected to be high. Regence's morbidity adjustment may be negligible.

The pattern is consistent: carriers whose enrollment is most concentrated among subsidy-sensitive, on-exchange members project the steepest adverse selection load. Carriers with commercially anchored or integrated-delivery enrollment project modest adjustments.

Demographic Morbidity Relativities: Quantifying the Composition Shift

The morbidity adjustment factor is not a guess. Pricing actuaries construct it from demographic morbidity relativities, which assign expected cost weights to enrollees based on age, sex, and (where available) historical risk score. The calculation follows a straightforward framework.

Start with the base period enrollment distribution: the age-sex composition of the 2025 or early 2026 enrolled population, weighted by the number of member-months in the experience period. Assign each age-sex cell its morbidity relativity factor, typically derived from the ACA age curve (the 3:1 age band with granular age-sex factors published by CMS) or from the carrier's own experience-rated relativities. Compute the weighted average morbidity relativity for the base period.

Next, project the 2027 enrollment distribution. This is where carrier-specific assumptions diverge. A carrier expecting 20% lapse concentrated in the 18-34 age band will project a 2027 distribution that is older and more heavily weighted toward the 45-64 age cells, which carry morbidity relativities 2x to 3x higher than the youngest cells under the ACA age curve. A carrier expecting uniform lapse across age bands projects a smaller shift.

The morbidity adjustment factor is the ratio of the projected 2027 weighted average morbidity relativity to the base period weighted average. If the base period average is 1.00 and the projected 2027 average is 1.06, the morbidity adjustment factor is 1.06, adding 6 percentage points to the rate increase independent of medical trend.

For Coordinated Care, with nearly 98,000 members skewing toward the subsidy-eligible population, even a modest shift in the age distribution, say the 18-34 share dropping from 28% to 22%, can produce a morbidity adjustment factor of 1.05 to 1.08. For Kaiser Northwest, where the age distribution is already older and more stable, the same subsidy shock might shift the 18-34 share from 20% to 18%, producing a factor of 1.01 to 1.02.

Risk Adjustment Transfers Under 45 CFR §153: Partial Offset, Not a Solution

The ACA's risk adjustment program, codified at 45 CFR §153, transfers funds from carriers with lower-than-average risk scores to carriers with higher-than-average risk scores within each state market. The mechanism is designed to attenuate adverse selection at the market level by removing the penalty for enrolling sicker members.

In a stable market, risk adjustment works reasonably well. Carriers that attract sicker populations receive transfer payments that partially compensate for higher claims costs, reducing the incentive to avoid high-risk enrollees through plan design or marketing.

In a contracting market, the mechanism's effectiveness degrades. When total enrollment drops 13%, the aggregate risk adjustment transfer pool shrinks proportionally. Carriers that were net receivers (those with sicker-than-average pools) receive smaller absolute transfers even if their relative risk position is unchanged. More critically, the enrollment shift itself changes each carrier's risk score relative to the market average. If healthier enrollees disproportionately leave the market entirely rather than switching carriers, the market-average risk score rises, compressing the spread between high-risk and average-risk carriers and reducing the per-member transfer amount.

Each carrier's pricing actuary must therefore make assumptions about the carrier's net transfer position for 2027, which depends on the carrier's projected risk score relative to the market average, both of which are moving simultaneously. This creates a circular reference in the pricing model: the carrier's premium depends on its risk adjustment transfer assumption, but the transfer amount depends on every other carrier's enrollment and risk score, which in turn depend on their premiums. The standard resolution is iterative: set a preliminary premium, estimate market-level transfers, recalculate the premium, and repeat until convergence.

Credibility Challenges When Enrollment Drops 13% in a Single Year

The 13% enrollment decline on the Washington Health Benefit Exchange creates a credibility problem for 2027 trend projection that goes beyond the usual ASOP No. 25 considerations. The issue is not just that the experience base is smaller. It is that the 2026 experience base represents a fundamentally different population than the one being priced for 2027.

Standard actuarial credibility procedures under ASOP No. 25 weight a carrier's own experience against an external benchmark (typically a Milliman or Wakely market composite) based on the volume of exposure. A carrier with 100,000 member-months might receive 85% credibility on its own experience; a carrier with 10,000 member-months might receive 40%. When enrollment drops 13%, the credibility weight on own experience mechanically decreases.

But the more fundamental problem is that the 2026 experience data reflects claims from a population that included the members who have since left. Those members, the healthier, lower-utilizing enrollees who dropped coverage when their net premium rose, contributed to the 2026 experience period's average allowed PMPM. Projecting that experience forward to a 2027 population that excludes those members requires an explicit adjustment: either removing the departed members' claims from the experience base (if individual-level data permits) or applying a prospective morbidity trend factor that accounts for the compositional shift.

Neither approach is straightforward. Removing individual claims data requires linking enrollment terminations to claims history, which is feasible for carriers with integrated data systems but operationally complex for those relying on EDGE server data. Applying a prospective morbidity factor relies on the same demographic relativity assumptions discussed above, introducing model risk into the trend selection.

The result is that carriers filing 2027 rates in Washington are making trend selections with wider-than-usual confidence intervals. A carrier that normally estimates medical trend within a +/-1% band around 8% may face a +/-3% band when the morbidity adjustment uncertainty is compounded with the credibility reduction from a shrinking enrollment base.

Single Risk Pool Constraints and the Providence Exit

ACA Section 1312(c) requires that each carrier price its individual market products as a single risk pool. A carrier cannot segment its on-exchange and off-exchange enrollees into separate pricing pools or apply different base rates to subsidized versus unsubsidized members. The single risk pool requirement means that a carrier's entire individual market book absorbs the morbidity shift collectively.

This constraint amplifies the adverse selection problem. A carrier with 70% on-exchange enrollment and 30% off-exchange enrollment cannot protect the off-exchange block from the morbidity deterioration concentrated in the on-exchange segment. The 30% of off-exchange members, who may be commercially stable and price-insensitive, receive the same percentage rate increase as the 70% of on-exchange members whose composition is driving the morbidity adjustment. This produces a secondary adverse selection effect: off-exchange members facing a 22% increase driven by on-exchange composition shifts may themselves exit, further worsening the pool.

Providence Health Plan's decision to exit the 2027 Washington individual market entirely, withdrawing its 254 enrollees, illustrates the extreme case. When a carrier determines that its projected 2027 risk pool is not viable at any attainable premium level, exit is the rational actuarial response. But the exit concentrates Providence's residual enrollees (those who must find new coverage) among the remaining twelve carriers. If those enrollees carry above-average morbidity, as is typical for members of an exiting carrier's block, their redistribution worsens the risk pool for every remaining issuer.

The effect is small in Providence's case (254 enrollees in a market of roughly 300,000), but the mechanism matters. In states where larger carriers exit, the redistribution effect can add 1 to 3 percentage points to remaining carriers' required rate increases, a dynamic the American Academy of Actuaries has documented in its carrier exit repricing briefs.

How the WA OIC Reviews These Filings

The Washington Office of the Insurance Commissioner will complete its actuarial review of all thirteen filings by September 2026. The review process involves independent validation of each carrier's actuarial memorandum against OIC benchmarks for medical trend, pharmacy trend, morbidity adjustment factors, and risk adjustment transfer assumptions.

The OIC's actuaries maintain their own morbidity models calibrated to Washington-specific enrollment data from the Health Benefit Exchange. When a carrier's morbidity adjustment factor deviates significantly from the OIC's independent estimate, the carrier must provide documentation justifying the deviation: specific enrollment composition data, age-sex distribution projections, and risk score analyses supporting its assumption.

Historically, the OIC has approved rates close to the requested levels for carriers whose actuarial memoranda are well-documented, while requiring modifications for carriers whose trend or morbidity assumptions appear inconsistent with market-level data. For the 2027 cycle, the OIC faces the additional challenge that market-level benchmarks are themselves shifting rapidly. The 13% enrollment decline means that the OIC's own historical trend benchmarks, built from a larger and differently-composed population, require the same compositional adjustments that carriers are making in their individual filings.

Why This Matters for Pricing Actuaries

Washington's May 26 filings are the first comprehensive state-level release of 2027 individual market rate requests with carrier-level detail. They matter beyond Washington for three reasons.

First, the 19-point carrier-level spread demonstrates that post-EPTC adverse selection loading is not a single-number industry assumption. It varies by a factor of three or more depending on enrollment composition. Pricing actuaries in other states filing 2027 rates will face the same decomposition challenge: medical trend is largely consensus, but the morbidity adjustment factor is carrier-specific and will drive the variance in requested increases across every state exchange.

Second, the filings surface the credibility problem inherent in pricing a contracting market. Standard loss ratio projection from a stable base period does not work when the base period population is compositionally different from the target population. Actuaries must choose between data-intensive individual-level adjustment and model-dependent morbidity trending, and the Washington filings show that different carriers are making different choices with materially different results.

Third, the Wakely morbidity data and OBBBA Medicaid churn dynamics compound the Washington-specific enrollment decline with national-level structural shifts. Carriers pricing 2027 rates in every exchange state must layer Washington-type adverse selection modeling onto the broader enrollment shocks from Medicaid redetermination churn and potential federal subsidy policy changes. The result is a rate development cycle with wider assumption bands and more carrier-level variance than any filing season since the ACA's early years.