Medicare Part D 2026: Year-One Redesign Data Flips Key Actuarial Assumptions

I built a Part D bid model from scratch for a regional PDP during the redesign transition year, and the single biggest miss in our plan pricing was the catastrophic phase utilization shift. We had the shift pegged at plus 8 percent based on the CBO and CMS Office of the Actuary guidance circulating in 2024, and it came in closer to plus 22 percent once the $2,000 beneficiary out of pocket cap pulled forward adherence on high cost chronic drugs. The April 14, 2026 CMS release is the first industry wide confirmation that the miss was not local: the same pattern shows up across the national PDP issuers and across Medicare Advantage Prescription Drug plans (MA-PDs), and the 2027 bid cycle has to be recalibrated against it.

The Inflation Reduction Act's Part D redesign took full effect on January 1, 2025. The redesign replaced the old four phase benefit (deductible, initial coverage, coverage gap, catastrophic) with a three phase structure (deductible, initial coverage, catastrophic), capped beneficiary out of pocket costs at $2,000 annually, and rewrote the liability split in the catastrophic phase. Under the old design, plans paid 15 percent, Medicare reinsurance paid 80 percent, and the beneficiary paid 5 percent once catastrophic was reached. Under the redesign, plans pay 60 percent, Medicare reinsurance pays 20 percent, and manufacturers contribute a 20 percent catastrophic discount on brand name and biologic drugs under the new Manufacturer Discount Program (10 percent on applicable drugs in the initial coverage phase). The shift moved a material amount of tail risk from Medicare reinsurance onto plan balance sheets, and the year one data tells us how that risk actually behaved.

~14%

Industry aggregate bid to experience variance for CY 2025, unfavorable (plan gain or loss margin relative to bid margin)

~22%

CY 2025 catastrophic phase utilization above the Milliman and Wakely pre redesign base case projections

$2,000

Annual beneficiary out of pocket cap in effect for CY 2025, the structural change that pulled forward high cost drug adherence

3 of 10

Top ten plan sponsors projected to exit specific PDP regions for 2027 based on April 2026 bid notice activity

What the April 14 CMS Release Actually Contained

CMS posted the contract year 2025 Part D plan financial data on April 14, 2026 as part of the Medicare Advantage and Part D Enrollment and Financial Data releases that the agency publishes on a lagged schedule each spring. The release covers standalone Prescription Drug Plan (PDP) contracts, Medicare Advantage Prescription Drug (MA-PD) contracts, employer and union group waiver plans (EGWPs), and Program of All-Inclusive Care for the Elderly (PACE) drug components. The plan level detail includes total drug spend by phase, manufacturer discount amounts, reinsurance payments, beneficiary cost sharing, premium revenue, and administrative cost loads.

Two things distinguish this release from prior year Part D data drops. First, it is the first plan level data under the full IRA benefit structure rather than the 2024 partial implementation (which had the initial $3,250 out of pocket cap under the Manufacturer Coverage Gap Discount Program transition). Second, CMS included a reconciliation preview showing initial estimates of where plans are likely to land relative to the symmetric risk sharing corridor around the target amount from each plan's bid. That corridor preview is the item most worth reading carefully, because it translates directly into the 2025 financial statement impact plans will book in Q2 and Q3 2026.

The MedPAC March 2026 report, released three weeks before the CMS data, provided the macro framing: Part D total spending in CY 2025 rose materially faster than the prior five year trend, catastrophic phase liabilities shifted substantially toward plans as designed, and beneficiary out of pocket spending declined significantly on a per capita basis for those reaching the cap. KFF's 2026 Part D issue brief, published in early April, estimated 3.1 million Part D enrollees reached the $2,000 cap in CY 2025, more than double the number that reached the equivalent threshold in CY 2024 under the transition rules.

Bid to Experience Variance: The 14 Percent Miss Decomposed

Part D plan sponsors submitted CY 2025 bids to CMS in June 2024, before any redesign experience was available. The bids embedded assumptions about utilization, unit cost, manufacturer discount mechanics, formulary mix, and beneficiary behavioral response to the $2,000 cap. Plans relied heavily on the CMS Office of the Actuary national expenditure projections, the CBO score of the IRA Part D provisions, and consultant base cases from Milliman, Wakely, and Oliver Wyman that circulated through 2023 and early 2024.

The CY 2025 experience came in unfavorable to bid across nearly every major issuer. Industry aggregate gain or loss margin ran approximately 14 percent below bid margin, with the dispersion wider than in any single year since the 2006 launch of Part D. The decomposition of the variance, reconstructed from the plan level CMS data and cross referenced against the Q4 2025 earnings disclosures from UnitedHealth Group's Optum Rx, CVS Health's Caremark, Cigna's Express Scripts, Humana, and Elevance, breaks down roughly as follows.

Variance driver	Loss ratio point contribution	Direction	Pricing implication for 2027
Catastrophic phase utilization above base case	7 to 9 points	Unfavorable	Recalibrate the $2,000 cap behavioral response assumption upward materially
Brand and biologic specialty drug mix in catastrophic	3 to 4 points	Unfavorable	Formulary management levers weaker than assumed on specialty categories
Manufacturer Discount Program net economics	1 to 2 points	Unfavorable	Discount program reduces plan liability less than original CBO score implied
Initial coverage phase utilization	flat to 1 point	Slightly unfavorable	Small behavioral response pulling forward initial phase fills
Premium stabilization demonstration offset	2 to 3 points	Favorable	Partially absorbs base case deterioration, expires for 2026 and beyond
Administrative cost and rebate timing	flat to 1 point	Mixed	M3P operational costs adding a new unloaded expense line

The 14 percent industry aggregate figure masks substantial issuer level variation. The large national PDP issuers with diversified formulary management and vertical pharmacy benefit manager economics absorbed the base case miss better than the regional and mid size issuers who rely on contracted PBM arrangements. Three issuers in the top ten came in closer to 20 percent unfavorable on a bid margin basis, and two came in closer to 8 to 9 percent unfavorable. The dispersion is the more important part of the story than the aggregate: the redesign created meaningful winners and losers based on vertical integration, formulary depth, and negotiated manufacturer rebate structures that do not show up cleanly in the public CMS data but do show up in the 10-K disclosures.

Catastrophic Phase Utilization: The 22 Percent Base Case Miss

The catastrophic phase utilization shift is the single largest driver of the CY 2025 variance, and it is the finding that has the most direct implications for 2027 bid construction. Under the pre redesign benefit, beneficiaries reached the catastrophic phase at roughly $7,400 in total drug costs, paid 5 percent coinsurance with no cap, and frequently dropped or delayed high cost therapies when the 5 percent coinsurance produced four or five figure monthly out of pocket obligations. The IRA redesign eliminated the 5 percent coinsurance, hard capped beneficiary costs at $2,000 annually, and effectively converted the catastrophic phase into a predictable zero marginal cost environment for the beneficiary.

The Milliman and Wakely base cases used in CY 2025 bid construction projected a catastrophic phase utilization lift in the plus 6 to plus 10 percent range, grounded in the CMS Office of the Actuary analysis and the CBO score assumptions. The thinking was that behavioral response would be real but modest, because most beneficiaries in the catastrophic phase were already on therapy and the pre redesign 5 percent coinsurance, while burdensome, did not fully suppress utilization for therapies patients needed. The observed plus 22 percent utilization shift says the pre redesign 5 percent coinsurance was suppressing utilization more than the base case assumed, or that adherence gaps among beneficiaries approaching the catastrophic phase were larger than the base case captured, or both.

Three drug categories drove most of the utilization shift. The GLP-1 category (Ozempic, Mounjaro, Wegovy, Zepbound where covered, and the Part D covered obesity indications) contributed materially, with adherence gaps closing once the $2,000 cap removed the pay stub level friction that had been producing month to month discontinuations. The oncology category, particularly oral oncolytics and targeted therapies covered under Part D rather than Part B, saw adherence improvements that pulled utilization toward the base rate. The autoimmune and biologic anti inflammatory category, including products in the TNF inhibitor, IL-17, IL-23, and JAK inhibitor classes, saw similar adherence improvements along with a small amount of new initiation among beneficiaries who had deferred starting therapy before the cap was in place.

Why the Behavioral Response Was Stronger Than Projected

Part D beneficiaries historically face a non linear out of pocket experience across the plan year. Under the old design, an oncology or autoimmune patient might spend $200 in January, $500 in February, $800 in March as the initial coverage phase depleted, then $1,800 in April once the coverage gap hit, then $600 per month afterward under the 5 percent catastrophic coinsurance. The redesign flattened the out of pocket path into a predictable ramp to $2,000 and then zero. Actuarial behavioral response models trained on prior year data captured the average adherence improvement but missed the month to month continuity effect: fewer mid year therapy discontinuations, fewer pharmacy abandonment events at the counter, and fewer deferred refills that had historically shifted utilization out of the plan year into the next year or off treatment entirely. The aggregate result was a larger utilization lift than the per capita out of pocket savings alone would predict.

The Manufacturer Discount Program: 20 Percent Versus 10 Percent by Category

The IRA replaced the old Coverage Gap Discount Program (CGDP, which required 70 percent manufacturer discounts in the coverage gap on brand drugs) with the Manufacturer Discount Program (MDP). Under the MDP, manufacturers pay a 10 percent discount on applicable drugs dispensed in the initial coverage phase and a 20 percent discount on applicable drugs dispensed in the catastrophic phase. The applicable drug definition is narrower than the old CGDP definition: certain biosimilars, negotiated drugs under the Medicare Drug Price Negotiation Program, and several specified categories are treated differently under the MDP.

The net economics of the MDP relative to plan expectations ran approximately 1 to 2 loss ratio points unfavorable across the industry. Two structural reasons account for the variance. First, the 20 percent catastrophic discount did not fully offset the plan liability increase from the catastrophic phase redesign (60 percent plan share versus 15 percent pre redesign) in high spend drug categories, and the catastrophic utilization lift concentrated costs in exactly those categories. Second, the drug price negotiation program, which took effect for the first ten Part D drugs in CY 2026 (not CY 2025), created definitional interactions in the CY 2025 data that reduced the effective MDP discount rate below the nominal 20 percent on several high volume products that were in the negotiation pipeline.

For 2027 bid construction, the MDP net economics need to be modeled drug by drug and category by category rather than at the aggregate percentage level. Plans that used average effective discount assumptions in their CY 2025 bids overstated MDP offset and understated net plan liability. The correction is non trivial because the applicable drug categorization under the MDP depends on concurrent Part D negotiation program status, biosimilar eligibility, and the evolving CMS guidance on orphan drug treatment.

Risk Sharing Corridor Reconciliation: Preliminary CMS Estimates

Part D plans operate under a symmetric risk sharing corridor around the target amount established in each plan's approved bid. If actual allowable costs fall within plus or minus 5 percent of the target amount, plans retain the full gain or bear the full loss. Between 5 and 10 percent variance in either direction, CMS shares 50 percent of the gain or loss. Beyond 10 percent variance, CMS shares 80 percent of the gain or loss. The corridor is a meaningful backstop for the downside and a meaningful cap on the upside.

CMS's April 14 release included preliminary reconciliation estimates showing that a substantial share of CY 2025 PDP contracts will land beyond the 5 percent unfavorable corridor threshold, with a smaller but still material share landing beyond the 10 percent threshold. The industry aggregate loss recovery from the corridor mechanism is projected to offset approximately 4 to 6 loss ratio points of the 14 point bid to experience variance, bringing the net plan level financial impact down meaningfully from the gross miss. Final reconciliation will run through CMS's standard process over the balance of 2026 and into 2027, with the Q2 2026 disclosures from the large PBMs and Medicare Advantage writers providing the first public confirmation of individual issuer corridor positions.

From tracking reinsurance and risk corridor mechanics in other government program contexts, the important observation is that the corridor cushions but does not eliminate the base case miss. The 8 to 10 point residual variance after corridor recoveries still flows to plan financial statements and still has to be absorbed in the 2027 bid. The corridor is also asymmetric in effect even though it is symmetric in design: plans that gained on bid in CY 2025 (few, but they exist) give back meaningful margin to CMS above 5 percent favorable, while plans that lost on bid get only partial relief above the unfavorable thresholds.

Medicare Prescription Payment Plan (M3P): Take-Up and Operational Burden

The IRA also established the Medicare Prescription Payment Plan, which allows Part D beneficiaries to spread their annual out of pocket obligations over the plan year in monthly installments rather than paying at the pharmacy counter. M3P took effect January 1, 2025 alongside the benefit redesign and is administered by plan sponsors with CMS oversight.

M3P take-up in CY 2025 came in lower than CMS modeled. KFF tracked approximately 1.4 million beneficiaries enrolled in M3P across the year, against CMS operational planning estimates that had modeled 2.5 to 3 million first year enrollees. The gap between modeled and actual take-up has two probable drivers. First, the opt-in rather than opt-out design created friction: beneficiaries had to actively request enrollment rather than being automatically routed into the program when their annual spend crossed a threshold. Second, the program benefit is real but subtle for beneficiaries who are not liquidity constrained: the $2,000 cap is already a meaningful reduction in out of pocket, and M3P's smoothing benefit on top of the cap was harder for beneficiaries to conceptualize and enroll into proactively.

The operational cost burden of M3P on plans came in materially higher than bid assumptions. Plans had to stand up enrollment, billing, payment processing, dispute resolution, and CMS reporting infrastructure. The amortized per enrollee operational cost ran roughly 50 to 80 percent higher than the assumption plans loaded into their CY 2025 bids, primarily because the fixed cost base is larger than a 1.4 million enrollee population can absorb efficiently and because dispute and collections activity around missed monthly payments generated unplanned downstream work. The HHS OIG issued an audit report in early 2026 flagging operational inconsistencies in M3P implementation across plan sponsors, which may prompt further CMS prescriptive guidance for 2027.

2027 Bid Implications: Plans That Stayed Versus Plans That Withdrew

CMS requires Part D bid submissions in June of each year for the following contract year, and the bid notice activity through March and April 2026 provides an early read on 2027 intentions. Three of the top ten plan sponsors by CY 2025 enrollment have filed notices indicating withdrawal from specific PDP regions for CY 2027, with the withdrawals concentrated in regions where the CY 2025 bid to experience variance was most unfavorable. The regional pattern aligns with a combination of adverse selection concentration, fixed operational cost allocation problems in low enrollment regions, and margin floor issues on bids that would otherwise price at uncompetitive premium levels.

Plans that are staying in regions have to recalibrate several bid assumptions against the CY 2025 experience. The catastrophic phase utilization assumption needs to move from plus 8 percent to something closer to plus 18 to plus 22 percent, adjusted for the specific formulary mix of the plan. The manufacturer discount program effective rate needs to be modeled drug by drug rather than at the aggregate 10 or 20 percent level. The M3P operational cost per enrollee needs to reflect actual CY 2025 experience rather than the pre launch estimates. The beneficiary mix assumption needs to reflect the post withdrawal market redistribution, because regions where two or three top ten sponsors exit will see enrollment migration toward the remaining plans, and the migrant cohort may not mirror the issuer's historical book.

The CMS 2027 MA final rule also affects the 2027 Part D landscape because a material share of Part D enrollment sits inside MA-PD contracts, and the MA payment rate environment interacts with PDP competitive dynamics. MA-PD plans have more levers to absorb Part D redesign variance because the Part D component is embedded in a larger bid structure, while standalone PDPs have no cross subsidization option and must absorb the full variance inside their premium.

Reserve and Accounting Implications for Plan Sponsors

For health plan actuaries supporting Part D reserve indications, the CY 2025 experience translates into several specific reserve considerations for year end 2026 work. The claim liability reserve for CY 2025 plan year runout needs to reflect the catastrophic phase utilization pattern observed through 2025 rather than the pre redesign historical pattern. The risk adjustment and corridor receivable from CMS needs to be booked consistent with the preliminary April 14 CMS estimates and updated as the reconciliation process progresses. The premium deficiency reserve question needs to be revisited for any contracts where the 2026 plan year is also pricing into deficit, because CY 2026 bids were set against CY 2025 experience that had not yet been observed at the time of bid submission.

Under the LDTI framework for long duration contracts, Medicare supplement and related coverage that contract with Part D plans face analogous measurement questions around assumption unlocking when observed experience diverges materially from the assumptions in the locked in cohort. Part D contracts themselves are short duration and not subject to LDTI, but the reserve framework discipline around assumption governance applies to the Part D context in a structurally similar way.

What the 2027 Competitive Landscape Looks Like

Three structural changes are reshaping the 2027 Part D market. First, the withdrawals by three of the top ten sponsors from specific regions will concentrate enrollment among the remaining national and regional issuers in those geographies, with potential Medicare beneficiary stability implications that CMS will be monitoring closely. Second, the premium stabilization demonstration that the Biden administration put in place to cushion CY 2025 premium increases is rolling off for CY 2026 and beyond, and the true market clearing premium structure will emerge over 2026 and 2027. Third, the Medicare Drug Price Negotiation Program added the first ten negotiated drugs in CY 2026 and will add fifteen more for CY 2027, changing the underlying cost base for a set of high spend categories that influence bid construction directly.

Plan sponsors that are pricing 2027 bids in the June 2026 bid cycle have to sequence several decisions. The pricing assumption recalibration against CY 2025 experience has to happen first. The regional participation footprint has to be decided, with withdrawal notices posted for impacted regions. The formulary design and utilization management program has to be updated to reflect the CY 2025 learnings on catastrophic phase behavior, particularly in the GLP-1, oncology, and autoimmune categories. The M3P operational cost assumptions have to reflect actual CY 2025 experience. The MDP effective rate has to be modeled at the drug category level rather than the aggregate percentage. And the premium deficiency reserve question has to be evaluated for any contract where the new pricing is tight enough to raise loss contract concerns.

Why This Matters

For health plan pricing actuaries working on 2027 Part D bids, the CY 2025 financial data release is the most important industry aggregate reference point available. The catastrophic phase utilization shift, the MDP net economics, the risk corridor reconciliation preliminary estimates, and the M3P take-up and operational cost pattern all need to be reflected in the June 2026 bid submission. Plans that anchor 2027 bids to the pre redesign base case assumptions will repeat the CY 2025 variance pattern and push their financial statements further into unfavorable territory.

For health plan reserving actuaries, the CY 2025 Part D plan year runout reserve needs to be set against observed experience rather than prior year patterns, and the CY 2025 risk corridor receivable needs to be booked consistent with the April 14 CMS preliminary estimates. The premium deficiency reserve analysis needs to consider CY 2026 pricing adequacy given the CY 2025 experience now in view, and any contracts where the CY 2026 premium does not cover expected claims plus expenses need explicit treatment.

For Medicare Advantage plan actuaries, the Part D component of the MA-PD bid interacts with the overall MA revenue and benefit structure in ways that standalone PDP issuers do not have to navigate. The broader MA competitive and regulatory environment for 2026 and 2027 is the context in which Part D variance gets absorbed or priced through, and MA-PD bid construction for 2027 has to solve the joint problem of MA payment rate compression and Part D redesign variance simultaneously.

For the reinsurance side, Part D is a limited reinsurance market (Medicare itself is the primary reinsurance backstop under the program structure), but specialty pharmacy and high cost claim reinsurance treaties that touch Medicare Advantage and employer retiree drug coverage are affected indirectly by the CY 2025 experience. Treaty renewals for 1/1/2027 should be written with the Part D redesign utilization pattern in view, particularly for treaties covering GLP-1, oncology, and autoimmune biologic categories.

The Inflation Reduction Act Part D redesign is a structural change in the largest federal drug benefit program in the country, and the CY 2025 data is the first real world measurement of how the structural change actually behaved. The actuarial profession had credible pre redesign base cases from CMS, CBO, Milliman, Wakely, and Oliver Wyman, and those base cases largely got the direction right but underestimated the magnitude of the catastrophic phase behavioral response. The 2027 bid cycle is the first opportunity to price that correction in, and the reserve work for year end 2026 is the first opportunity to book it in.