What Medicare Advantage plans file for agent pay, 2018 to 2026

CMS sets a maximum agent payment per member per year, higher in four rate regions and one national figure for everywhere else. For CY2026 the national number is $694. The obvious assumption is that carriers compete somewhere underneath it, the way employers pay somewhere under a salary band.

They don't. Of the 5,056 Medicare Advantage plans filed in the national rate region for CY2026, 99.6% put the initial payment at $694 on the nose. Not $690, not $650. It has held above 98% every year since 2020, and the one dip, 87.1% in 2019, traces mostly to plans still carrying 2018's $455 in that year's file rather than to carriers underpaying. CMS isn't capping a market here so much as setting the price for it.

That single fact is useful before anything else. If you sell an MA policy in a national-region state this year, the compensation your carrier filed is a number you can look up, and it's almost certainly $694.

The filing has two halves. Alongside that maximum, a plan states its minimum, and 95.6% of CY2026 plans state $0. The attested range for a typical Medicare Advantage plan is therefore "somewhere between nothing and $694".

Weighted by enrollment it's starker still. 28,465,681 of the 29,849,833 enrollees we could match sit in a plan filing a $0 floor. Humana, CVS Health, Elevance and Kaiser file it on 100% of their plans. UnitedHealth's filings cover 7.6 million members, and 99.4% of them sit in a $0-floor plan.

The share has been climbing for the whole period, from 76.0% of plans in CY2018 to 95.6% in CY2026, and nine of those points arrived in the final year, right when CMS changed its file format.

CMS doesn't define what a $0 minimum means, and none of the 5,056 national-rate filings leaves the field blank. The likeliest innocent reading is operational: initial pay is pro-rated by month and clawed back entirely on a rapid disenrollment, so a $0 floor can be a carrier stating its worst case net of chargebacks. A few plans file floors of $57.83, one month of $694, and 181 file a floor equal to their ceiling. What the field can't do, either way, is tell an agent what a given enrollment will net them.

The one part of the picture with almost no spread is renewal. In every year from 2018 to 2026, at least 98% of plans set renewal compensation at half of the initial payment, rounded to the nearest dollar. For CY2026, where half of $694 is an even $347, 99.4% of plans file exactly that.

So the national renewal figure to reconcile against is $347 per member per year. What lands on a statement still moves with months enrolled, chargebacks and plan-type rules, so the filed rate is the expected value a line reconciles against, not a number every line must equal.

The nominal maximum has risen every year, which makes the headlines look generous. Deflated by CPI-U into constant 2026 dollars, the picture is duller. The maximum reached $657.86 in 2021 and then fell to $643.07 by 2025, a real-terms decline of 2.2% over four years. It lost ground to inflation outright in 2022, 2024 and 2025.

CY2024 is the clearest case. The maximum went from $601 to $611, a rise of 1.7% in a year when CPI ran ahead of it, so every MA agent in the country took a real pay cut while the trade press reported an increase.

CY2026 is the correction: $626 to $694 is 10.9% nominal and 7.9% real, the largest single-year real rise in the nine years we measured, and it leaves the cap at its highest real value in the series, above even 2021.

Put the findings together and the public record gives you two things you can verify against any statement. The filed ceiling for a national-region MA policy is $694 for CY2026, on 99.6% of plans, and renewal is half of it. Both are per member per year, both are published, and neither requires anyone's permission to look up.

What the public record can't give you is your own contracted rate. For 95.6% of plans the filed floor is $0, so the range CMS holds on your carrier spans the entire distance between nothing and the maximum. Your actual rate exists in your contract and in the statements you're paid against, and nowhere else.

That's the practical reason line-by-line reconciliation matters for Medicare agents specifically. The expected side of most of an MA book starts from a published figure, adjusted by what your own records hold: months enrolled, chargebacks, plan type. It's the kind of check that stays boring right up until the month a renewal comes through at the wrong number.

CMS publishes the compensation filings as annual bundles going back to CY2013, though the split between initial and renewal payment only starts in CY2018, so that's where our window opens. We deduped to one row per plan per state, since compensation is filed per county and a plan sold in 600 counties would otherwise drown out one sold in three. Standalone drug plans are excluded because they run on a schedule closer to $100 than $700.

Each year's maximum is recovered from the filings themselves rather than copied from an industry blog, then checked against the published CMS memos for 2021, 2025 and 2026 across all four rate regions. That's twelve comparisons and twelve exact matches. The pipeline refuses to write its output if any of them drift.

The full derived dataset and the methodology are published alongside the study, including the limits: a filing is what a carrier attests, not what it paid, and CMS suppresses enrollment counts of ten or fewer, which biases the weighted totals slightly against small rural plans.