On October 15, the Medicare annual enrollment period opens and the mailers start arriving: $0 premiums, dental cleanings, a gym membership, a card for over-the-counter purchases. How does a private insurer afford all that while charging less than the government plan it replaces? The answer lies in how Medicare Advantage plans get paid, which is stranger than most people assume. Those extras aren’t a gift from the insurer. They’re a slice of a gap between two numbers, and the size of that slice was set by a quality score the plan earned more than a year before you saw the ad.
Medicare Advantage plans get paid a flat monthly amount per person, not per claim
Original Medicare pays doctors and hospitals claim by claim. Medicare Advantage works the other way around. The government hands a private plan a fixed monthly payment for each person enrolled, called a capitated payment, and the plan takes on the job of paying the bills. If its members cost less than the payment, the plan keeps the difference. If they cost more, the plan eats the loss.
The scale is large. The Medicare Payment Advisory Commission, the independent agency that advises Congress, reported in March 2026 that Medicare paid MA plans an estimated $537 billion in 2025, not counting drug coverage, and that the average payment in 2026 will be $16,242 per enrollee per year. According to KFF, just over 35 million people were enrolled as of February 2026, which is 55% of everyone eligible. More than half of Medicare now runs on this flat-fee model, and for 2027 the Centers for Medicare & Medicaid Services finalized an average payment increase of 2.48%, or more than $13 billion, in its April 2026 rate announcement. That raise is part of what’s funding the mailers now landing in your parents’ mailbox.
The benchmark and the bid: a reverse auction that funds the extras
Every year, each plan submits a bid, which is its own estimate of what it will cost to provide standard Medicare Part A and Part B benefits to an average-health enrollee in a given county. Medicare, separately, sets a benchmark for that county. The benchmark is the most the government will pay, and it’s pegged to what Original Medicare spends locally, ranging from 95% of that spending in high-cost counties to 115% in low-cost ones, according to KFF.
Most plans bid below the benchmark. When they do, the gap between the two numbers is split. Part of it goes back to the Treasury, and the rest goes to the plan as a rebate, which the plan must spend on its members. That rebate is where the extras come from. MedPAC found the average rebate reached $2,660 per enrollee per year in 2026, more than double what it was in 2018. Conventional plans project spending about 35% of it on lower copays and deductibles, 22% on benefits Original Medicare doesn’t cover such as dental and vision, 26% on richer drug coverage, 9% on lowering members’ Part B premiums, and about 8% on administration and profit.
So when a plan advertises a $0 premium and a dental allowance, the plain translation is that it bid low enough, relative to its county benchmark, to generate a rebate big enough to pay for those things.
Medicare Advantage star ratings change both the ceiling and the split
Most enrollment guides skip this next mechanism. Star ratings, the one-to-five score Medicare publishes for every plan contract, aren’t just a consumer label. They move money in two separate ways.
First, a contract rated four stars or higher gets its benchmark raised, usually by five percentage points, which lifts the ceiling the bid is measured against. Second, the rating sets how much of the gap the plan keeps. KFF’s analysis of the quality bonus program lays out the schedule: a plan rated 4.5 stars or higher keeps 70% of the difference between benchmark and bid, a plan rated 3.5 or 4 stars keeps 65%, and anything lower keeps 50%.
Work it through with round numbers. Picture a county where the benchmark is $1,300 a month and a plan bids $1,000. If the plan’s contract is rated four stars, its benchmark rises 5% to $1,365. The gap is $365, and at a 65% share the rebate is $237.25 a month, or $2,847 a year per member. Now suppose that plan slips to 3.5 stars. It loses the benchmark bump, so the gap shrinks to $300. It still keeps 65%, so the rebate falls to $195 a month, or $2,340 a year. That’s $507 per member per year that simply disappeared from the extras budget, from a half-star drop. At three stars, the share falls to 50% and the rebate drops to $150 a month, or $1,800 a year. Same plan, same bid, same members, and more than $1,000 less per person to spend on dental and lower copays.
This is why plans fight so hard over their ratings. In 2026, KFF counted 209 contracts at four stars or above, down from 261 the year before, and the share of enrollees in bonus-eligible plans fell from 75% to 68%. Humana’s largest contract dropped from 4.5 to 3.5 stars for the 2025 rating year, and the company sued Medicare over it. So far, the courts have sided with the government. Medicare will still spend at least $13.4 billion on these quality bonuses in 2026, KFF estimates.
The star rating you see while shopping pays for benefits two years later
For your own decision, the timing matters most. Medicare publishes star ratings each October, but they don’t touch payments until the year after next: the ratings released in the fall of 2025 set the bonus status and rebate share behind 2027 benefits. The ratings published this October will shape plan money for 2028.
That lag has two practical consequences. A plan that just fell below four stars may still look generous this year, because this year’s extras were financed by last year’s rating, and the cuts arrive when you renew. And a plan whose rating jumped this fall can’t yet spend the money that comes with it. When you compare plans during open enrollment, look at the rating history as well as the current number. CMS publishes prior years’ ratings, and a plan that has slid from 4.5 to 4 to 3.5 is telling you something about its future budget for extras.
Risk scores and coding intensity, the other lever
There’s one more adjustment. Payments are scaled up or down by each member’s risk score, which is built from the diagnoses recorded in their medical records. A sicker member brings a bigger payment. That creates an incentive to record every possible diagnosis, which MedPAC calls coding intensity, and it estimates that MA risk scores run about 4% higher than those of comparable people in Original Medicare. It also found that people who choose MA tend to be healthier than their risk scores predict, which it calls favorable selection. Together, MedPAC estimates, these make Medicare spend about 14% more on MA enrollees than it would if the same people stayed in Original Medicare, roughly $76 billion in 2026. That’s down from a projected 20% in 2025, largely because Medicare finished phasing in a new risk model called V28 that trims the payment effect of coding.
Everyone pays for part of that gap, including people who never join an MA plan. MedPAC estimates Part B premiums are about $11 billion higher in 2026 because of the extra MA payments.
Using the mechanism when you choose a plan
I don’t think any of this makes Medicare Advantage a bad deal, or a good one. It explains where the extras come from and why they move. If a plan’s appeal is mainly its dental allowance or its $0 premium, you’re betting that its rebate holds up, and the rebate depends on a star rating and a benchmark that can both change. The network, the prior authorization rules, and the out-of-pocket maximum, which can reach $9,250 in 2026 per MedPAC, are what shape your costs in a bad year.
So when you’re helping a parent compare plans before December 7, read the star history, then the out-of-pocket maximum, and only then the extras. Once you know how Medicare Advantage plans get paid, the glossy mailer becomes something you can check instead of something you have to trust. For the drug side of the same decision, see our explainer on Medicare Part D’s 2026 changes and our guide to how pharmacy benefit managers work.