CHAPTER 01
The benchmark: what Medicare offers to pay
Before any plan bids anything, CMS sets a ceiling for every county in the country: the benchmark. It starts from what Original Medicare spends per person per month in that county (the county's fee-for-service base), then applies a percentage set by how expensive the county is. The cheapest quarter of counties gets 115% of local costs, the next quarter 107.5%, then 100%, and the most expensive quarter gets 95%. Cheap counties get a markup to attract plans; expensive counties get a haircut. A second rule sits behind all of it: no county's benchmark may exceed what it would have been under the pre-2010 formula. That ceiling turns out to matter enormously, and Chapter 2 is where you can see it bite.
Star ratings then bend the percentage upward. The bonus adds percentage points: a 115% county with a 4-star plan becomes a 120% county, not 115% times 1.05. Here is this county's actual published row, all three tiers, to the penny:
CHAPTER 02
Stars: three gates, not one
A plan's star rating moves its payment through two different levers, and they trip at different thresholds. At 3.5 stars the plan keeps a bigger share of its bid savings (the rebate share rises from 50% to 65%). At 4.0 stars the benchmark itself grows: the county's applicable percentage gains 5 points, or 10 in the 279 counties that qualify for a doubled bonus (Chapter 1's county line tells you whether the one you picked is among them). At 4.5 stars the rebate share rises again, to 70%. Drag the rating across each line and watch which lever moves.
One timing note the slider cannot show: the rating that sets a plan's 2026 bonus is not measured during 2026. It is the 2025 Star Rating, published in October 2024, previewed to plans that November, and therefore already fixed before the plan writes its 2026 bid. Quality is paid on a two-year delay.
WHAT THE FILE SHOWS
CHAPTER 03
The bid and the rebate: where the extra benefits come from
Each plan tells CMS what it would charge to deliver Original Medicare's benefits in that county: its bid. Bid under the benchmark and the plan does not pocket the difference. It must return at least half the savings to members as a rebate (half at under 3.5 stars, rising to 70% at 4.5), and that rebate is what funds everything MA plans advertise. By regulation it can go to exactly three places: supplemental benefits such as dental, vision, hearing, and lower cost sharing; a reduction in the Part D drug premium; or a reduction in the member's Part B premium. Bid over the benchmark and members pay the difference as a monthly premium instead.
The bid is a cost, not a share of the benchmark, so it does not move when the star rating does. That is what makes the star thresholds in Chapter 2 matter twice over: scroll back and raise the rating, and the bid above stays exactly where you left it while the benchmark climbs and the share of the savings rises with it. The same bid in the same county funds visibly different benefit packages at 3.0, 3.5, and 4.5 stars, and the same plan funds different packages in a capped county and an uncapped one.
CHAPTER 04
Risk adjustment: the multiplier on everything
Risk scores come from the CMS-HCC model, a separate machine that deserves its own page. For this one, the point is only that everything above is quoted per member per month at a risk score of exactly 1.0, and real payments scale member by member.
SOURCES & FINE PRINT
- County rates, FFS bases, caps, and bonus-county flags: CMS, 2026 Ratebook and supporting data (the ratebook and rate-calculation files). Every rate on this page was recomputed from those components and matches the published figure to the penny.
- Applicable percentages, quality bonus points, and qualifying counties: 42 CFR 422.258(d).
- Rebate shares and permitted uses: 42 CFR 422.266.
- Benchmark cap: Social Security Act §1853(n)(4), as discussed in the CY 2026 Rate Announcement, Section C.
Simplifications, disclosed. Rates shown are the published Parts A&B non-ESRD county rates; the fee-for-service base is CMS's adjusted figure, which excludes the phased-out indirect medical education and kidney acquisition amounts, so it is not raw county Medicare spending. A plan serving several counties is paid an enrollment-weighted blend of their rates, not a single county's. ESRD, PACE, employer-group plans, and regional PPO benchmarks are separate books. New plans get a 3.5-point bonus and count as 3.5-star for rebate purposes. Counties that changed cost quartiles get a blended percentage during transition. Los Angeles appears twice in CMS's file as two payment areas with identical rates, and is shown once here. This page names no companies on purpose: it explains the machine, not any player in it.