MedPAC's July 2026 Data Book: MA at 55 Percent, Benchmarks at 124 Percent of a Shrinking FFS Base

Medicare Advantage now prices against a population it has spent fifteen years absorbing. MedPAC's July 2026 data book counts 35.5 million MA enrollees in February 2026, 55 percent of eligible beneficiaries, and puts 2026 benchmarks at 124 percent of what fee-for-service spending would have been for those enrollees (MedPAC, July 2026). Every county bid target anchors to per capita spending in a program covering a minority of beneficiaries.

The Medicare Payment Advisory Commission's annual data book, "Health Care Spending and the Medicare Program," is the reference volume plan actuaries mine for bid-season exhibits. This year's edition documents a payment system whose two central quantities are moving in opposite directions. Enrollment in the risk-paid program keeps compounding: 35.5 million in February 2026, up from 34.4 million a year earlier and 11.7 million in 2011, when penetration stood at 26 percent (MedPAC, July 2026). Meanwhile the FFS population whose spending defines every county benchmark keeps contracting, and the machinery that converts that spending into benchmarks, bids, and rebates deserves a closer read than the enrollment headline received.

The Enrollment Detail Behind the 55 Percent

The composition of the growth is more informative than its level. Total MA enrollment grew about 3 percent over the year, but regional PPO enrollment collapsed 31 percent to 163,000 and private fee-for-service plans held just 36,000 members (MedPAC, July 2026). Local coordinated-care plans now hold 35.3 million enrollees, over 99 percent of the program. More than three-quarters of the past year's local-plan growth came from special-needs plans. KFF's parallel analysis of March 2026 enrollment puts SNPs at 8.2 million members, 23 percent of MA enrollment, and attributes 85 percent of the year's net program growth to them (KFF, June 2026). The growth engine is now the dual-eligible and chronic-condition products, which carry the highest rebates and the highest coding intensity in the program.

Geographic dispersion remains wide. Fourteen states sit at 60 percent MA penetration or higher, with Michigan at 67 percent and Puerto Rico at 96 percent, while Alaska sits at 2 percent and Vermont at 13 percent (MedPAC, July 2026). The dollars scale accordingly: Medicare paid MA plans an estimated $537 billion in 2025, more than double the 2019 total, and MedPAC's March 2026 report projects $615 billion for 2026 (MedPAC, March 2026). The Congressional Budget Office expects penetration to reach 63 percent by 2034 (KFF, June 2026). The program's minority partner is FFS Medicare, and it is the minority partner that still sets the prices.

The Benchmark Formula and Its Shrinking Denominator

The statutory mechanics have not changed. Counties are ranked into quartiles by per capita FFS spending, and benchmarks run from 95 percent of projected local FFS spending in the highest-spending quartile through 100 and 107.5 percent in the middle quartiles to 115 percent in the lowest. Plans rated 4 stars or better receive a 5 percentage point benchmark increase, 10 points in double-bonus counties (MedPAC, March 2026). What has changed is the population underneath the projection. As of June 2024, MA-eligible beneficiaries split 32.8 million in MA against 27.6 million remaining in FFS (MedPAC, July 2026). The remainder skews toward beneficiaries with Medigap coverage, 20.5 percent of all beneficiaries in 2023, whose first-dollar wraps are associated with higher utilization, and away from the dual-eligible groups enrolling in MA at rates of 59 to 78 percent.

MedPAC's own front matter concedes the measurement problem. The data book's notes warn that aggregate FFS sector spending changes now partly reflect enrollment migration to MA rather than cost trend, and that "Fee-for-service spending per capita may present a more complete picture of spending changes" (MedPAC, July 2026). The Commission has also criticized the projection base itself: CMS's county FFS estimates include beneficiaries with Part A only, who cannot enroll in MA, and in 2023 national per capita FFS spending ran about 6 percent higher when that group was excluded (MedPAC, March 2026). MedPAC has recommended since 2017 that benchmarks be computed only from beneficiaries holding both Part A and Part B.

The CY2027 rate cycle showed how sensitive the anchor has become. CMS's April 2026 rate announcement finalized an effective growth rate of 5.33 percent, up from 4.97 percent in the January advance notice (CMS, April 2026). CMS's own fact sheet describes that growth rate as driven largely by growth in Original Medicare per capita costs, as estimated by the Office of the Actuary. Georgetown's Medicare Policy Initiative traced 0.36 points of that revision to a single additional quarter of FFS claims experience (Georgetown CHIR, April 2026). One quarter of data from the shrinking half of the program moved the growth engine for $615 billion of annual payments. The overall update went from 0.09 percent in the advance notice to 2.48 percent at final, roughly $13 billion. The larger share of that swing came from CMS retreating to a milder risk-model revision rather than from the growth rate itself.

Layered on the projection base is the selection problem, which MedPAC now quantifies annually. Favorable selection, the tendency of the FFS-calibrated risk model to overpredict what MA enrollees would have cost in FFS, added between 9.4 and 11.2 percent to MA payments in every year from 2015 through 2023 (MedPAC, July 2026). The decomposition below is the data book's own arithmetic for 2026, and it is the single most bid-relevant exhibit in the volume.

2026 estimate, share of FFS spending	Benchmarks	Bids	Payments
Before coding and selection effects	107%	83%	99%
Estimated coding effect	+4	+3	+4
Estimated selection effect	+12	+9	+11
Overall estimate	124%	95%	114%

Read the first column twice. Statute alone puts benchmarks at 107 percent of projected FFS spending; coding intensity and favorable selection inflate the effective anchor to 124 percent of what the same enrollees would actually have cost (MedPAC, July 2026). Payments land at 114 percent. That is MedPAC's headline finding: Medicare will spend 14 percent more for MA enrollees in 2026 than FFS coverage would have cost, a projected $76 billion, equal to 12 percent of all MA payments (MedPAC, March 2026). That figure is down from the 20 percent estimated for 2025, and the Commission credits the completed phase-in of the V28 risk model for most of the improvement. Selection, not coding, is now the dominant wedge: $57 billion of the $76 billion.

Rebate Dollars and the Bid Margin Underneath Them

Rebates are where benchmark generosity becomes product design. A plan bidding below its benchmark keeps 65 or 70 percent of the difference, depending on star rating (MedPAC, July 2026). It must convert that rebate into supplemental benefits, reduced cost sharing, or premium buy-downs after loading administrative cost and margin. For 2026 the data book puts the average rebate at $218 per member per month across all plans, with HMOs at $235, local PPOs at $187, and SNPs at $274, up from $170 as recently as 2021 (MedPAC, July 2026). Annualized, the average plan is financing roughly $2,600 per enrollee per year of extras. Working the statutory shares backward, a $218 monthly rebate implies plans are bidding roughly $310 to $335 per member per month below their benchmarks. The data book confirms the spread directly: enrollment-weighted bids average 77 percent of benchmarks, and virtually every plan bid below its benchmark for 2026.

The bids themselves embed a specific actuarial claim. Before adjusting for coding and selection, plan bids average 83 percent of projected FFS spending (MedPAC, July 2026). Plans are telling CMS they can price the Part A and Part B package about 17 percent below what FFS would spend on the same risk profile. Some of that is genuine managed-care efficiency, and some is the same selection and coding arithmetic that inflates the benchmark. The V28 phase-in supplied a live experiment in how plans defend that spread. The model's final year completed in 2026, and MedPAC's March report found that plans absorbed the risk-score compression in their bids while holding aggregate rebates at a record level (MedPAC, March 2026). Payment pressure landed in bids and in benefit design at the edges, a trade-off the site examined in the 2026 benefit-design paradox: rebate competition is the enrollment engine, and plans defend it before they defend margin.

Quality bonuses sit on top of the rebate stack, financed with added program dollars rather than budget neutrality. The star-rating benchmark increases will add about $16 billion to MA payments in 2026, roughly 3 percent of the program, with 64 percent of enrollees projected to sit in bonus-rated plans (MedPAC, March 2026). A half-star movement is a compound payment event, shifting both the 5-point benchmark add and the 65-versus-70 rebate share, which is why star-rating methodology has become annual litigation. The site's coverage of Elevance's suit against CMS over $115 million in bonus revenue walks through how cut-point mechanics turn a quality label into a rate input. MedPAC, for its part, has recommended replacing the quality-bonus program outright since 2020.

Three Organizations, 58 Percent of Enrollment

Bid competition is the system's intended discipline, and the data book's companion status report shows how little of it operates at the margin. UnitedHealth Group, Humana, and CVS Health together enrolled 58 percent of all MA members as of July 2025, up from 32 percent in 2008, on shares of 29, 17, and 12 percent respectively (MedPAC, March 2026). At the county level, where bids actually compete, the largest organization holds a weighted-average 43 percent of enrollment and the top three hold 82 percent (MedPAC, March 2026). Ninety-five percent of MA enrollees live in counties that qualify as highly concentrated on the Herfindahl-Hirschman index. The program lists 164 parent organizations offering 5,492 plans, and the average beneficiary can choose among 39 of them; the enrollment, even so, pools with three bidders.

The 2026 plan year rearranged share inside the oligopoly rather than beyond it. KFF's March 2026 data show UnitedHealth's enrollment down nearly 647,000 members to a 26 percent share while Humana added 1.3 million to reach 20 percent (KFF, June 2026). The two firms still hold 46 percent of the program between them. Membership that churned out of one national carrier's exiting counties largely re-enrolled with another. MedPAC cites the research consequence: less-concentrated MA markets are associated with more generous benefits and lower premiums, and dominance by a small set of firms weakens the pressure that would otherwise force bids down toward cost (MedPAC, March 2026).

Concentration also changes what plans compete on. MedPAC found that 8 of the 10 largest MA organizations posted 2024 coding intensity at least 5 percentage points above the 5.9 percent adjustment CMS applies (MedPAC, March 2026). Eight organizations exceeded FFS coding levels by more than 20 percent. A large organization that out-codes its rivals converts documentation into risk score, risk score into rebate, and rebate into supplemental benefits that win enrollment. The Commission says that loop distorts both plan competition and the incentive to manage medical cost. In an 82 percent top-three county, the rival's benefit sheet constrains a bid more tightly than the benchmark does.

Implications for 2027 Bid Development

Three items belong in every MA pricing team's 2027 assumption memo. First, benchmark projections deserve explicit uncertainty treatment. The CY2027 cycle demonstrated a 0.36-point growth-rate revision from one quarter of FFS experience, and the FFS cells generating county projections keep thinning fastest exactly where penetration is highest, in the fourteen states already at 60 percent or more. Counties with small residual FFS populations will produce noisier projections. The sensitivity runs through every downstream quantity: a benchmark revision moves the rebate pool by 65 to 70 cents on the dollar before a single medical trend assumption changes.

Second, the V28 reprieve is a cycle, not a policy. CMS declined to recalibrate the model with 2023-2024 data for CY2027, keeping the 2018-2019 calibration, and softened the proposed risk-model impact from a 3.32 percent reduction to 1.12 percent (CMS, April 2026). The same announcement embeds projected MA risk-score growth of 2.50 percent from coding trend. Recalibration exposure returns in the CY2028 cycle, and bids built on the assumption that the April 2026 generosity repeats are carrying unpriced regulatory risk. MedPAC's estimate that net coding effects still add $22 billion to 2026 payments is the standing argument for a future adjustment above the statutory minimum, which CMS has never exercised.

Third, the fiscal arithmetic is now large enough to be a pricing assumption in its own right. The Committee for a Responsible Federal Budget extended MedPAC's $76 billion estimate across CBO's February 2026 baseline and projects roughly $1.3 trillion of above-FFS payments over the coming decade (CRFB, March 2026). CRFB argues the Hospital Insurance trust fund would remain solvent well beyond its current depletion date without them. MedPAC adds that higher MA payments raise Part B premiums about $11 billion in 2026, roughly $14.61 per beneficiary per month, including for the FFS enrollees who receive no supplemental benefits from the arrangement (MedPAC, March 2026). Benchmark rebasing, Part A-only exclusion, and a discretionary coding adjustment all sit on MedPAC's standing recommendation list; each is a repricing event for a book that now runs $615 billion a year. The 55 percent milestone is not the finding. The finding is that the reference price under all of it is measured on the other 45 percent, and both MedPAC and CMS spent this spring demonstrating how far that measurement can move.