[HUM, UNH, CVS, ELV, ALHC] CY26 Medicare Advantage Rates: Likely Upswing in the Final

By Beth Steindecker Published on January 11, 2025 PDF

CMS’s Friday release of the CY26 Advance Notice (AN) for Medicare Advantage (MA) insurers [HUM, UNH, CVS, ELV, ALHC, CLOV, CNC] came in better than our net YoY benchmark expectations with a proposed net increase of 2.23%, which excludes the estimated 2.10% risk score trend. (We note, though, that the Medicare agency is telegraphing a preliminary MA rate increase of 4.33%, inclusive of this trend, which plans tend to ignore.) However, there is much room for the CY26 MA benchmark rates to rise come spring, when the final MA rate announcement is released (statutory deadline is the first Monday in April (Apr. 7)).

Such an increase, potentially in the mid-single-digit range, would afford MA insurers much more leeway on their plan design and benefit generosity as well as value-based care and healthcare providers [AGL, EVH, PRVA, DVA, FMS, HCA, THC, CYH]potentially more breathing room on payments/reimbursements.

As we previously wrote, the Trump administration has various levers it can – and we expect it to – pull for improvement, given the overall bias in favor of MA vis-à-vis traditional Medicare, past pushback from insurer stakeholders, and political calendar. We do expect that in the spring CMS is likely to finalize a YoY MA benchmark change solidly in the mid-single digit range. The key levers that CMS is likely to consider for the final announcement include:

  • Use more recent fee-for-service (FFS) expenditure data from CY24 for calculating CY26 FFS growth rates; for the AN, CMS incorporated Part A claims experience only through the end of CY23 (which CMS representatives seemed to acknowledge as a bit dated) and Part B claims through mid-CY24.
  • Change the FFS growth rate methodology to reflect the calculation used to forecast FFS growth in CMS’s ACO REACH demonstration project; the ACO REACH model relies upon a different dataset of years, expenditures, and adjustments.
  • Pause or slow down the implementation of the technical adjustment applied to the FFS growth rate to phase out medical education costs; CMS estimates in the AN that implementing the remainder of its three-year phase-out of these costs lowers the preliminary CY26 FFS growth rate by 1.42%.
  • Pause or slow down the implementation of the version 28 (V28) risk adjustment model that is being phased in over three years with CY26 being the final year.  CMS estimates that the impact from the raw risk adjustment model revisions phase-in for CY26 is a 4.31% reduction, and when combined with a +1.30% impact from the normalization factor, results in a net 3.01% cut to the benchmarks in the AN.

Source: CMS, CPP

After the close Friday, CMS released its CY26 AN, which the agency estimates would increase MA spending by $21B YoY. This proposal came out a few weeks before the statutory deadline in early February. Aside from when President Trump’s CMS during his final year in office in 2022 issued the CY23 AN more than four months early, this is the only other time we can recall in the past two decades that CMS has posted its AN before the statutory deadline. We suspect that the Biden administration rushed to get its proposal out before Trump’s CMS would put its imprint on the preliminary notice, potentially undercutting some of the proposed changes in the current AN.

Even so, our assessment is that the AN is better than we expected. We had thought the current CMS would be stingy in its forecasting of cost growth while maintaining some of its already finalized policies like the phase out of the MA-related medical costs from FFS cost growth and implementation of the v28 risk adjustment model.

To some extent CMS was, since there was a belief that the FFS growth rates would be even higher than what was in the AN and therefore pumping up the preliminary benchmarks. Some investors thought that CMS would use the estimates projected in the agency’s ACO REACH demonstration model that revealed an 8%+ growth trend. We were skeptical of such adoption, given CMS did not do that in the past and there are differences in the datasets used for calculating the FFS growth for MA vs. ACO REACH. Yet, the ACO REACH data, along with higher utilization and expectation of prior-year adjustments, was suggestive of higher growth compared to last year.

Moreover, CMS somewhat admitted that the AN’s growth rates do not reflect much of the claims’ experience for CY24, which is the base year for this proposal. Officials said the experience incorporated into the calculations is incomplete, since it only reflected Part A claims through December 2023 and Part B through June 2024. They expect the final announcement will be more comprehensive but would not specify the anticipated period for the claims experience.

Key components of the CY26 AN for MA  

Effective growth rate: CMS projected a preliminary CY26 effective growth rate (i.e. the starting point for the calculation of MA benchmark changes YoY) at 5.93%. This rate is derived from the: 1) FFS US per-capita cost growth, 5.67%, which drives most of the calculation; and 2) total US per-capita cost growth, 7.70%. These estimates account for CMS implementing the rest of the technical adjustment for eliminating the MA-related medical education costs.

Medical education cost phase-out: As expected, CMS proposed to remove the remainder of the MA-related medical education costs from its cost growth calculations for CY26. In the CY24 MA rate announcement, CMS finalized a three-year schedule to phase out these costs via a technical adjustment with CY26 as the final year. For CY26, it estimated that implementing the rest of this adjustment that it had been phasing in during CY24 and CY25 would cut the growth rate by 1.42%. So, by pausing this policy, that would reverse this hit and add another $7B in MA spending for CY26, according to CMS.

Version 28 (v28) Risk Adjustment and FFS Normalization Factor: In sync with its finalized three-year timeline, CMS proposed to maintain its policy to adopt fully its overhauled risk adjustment model, which began in CY24. The agency projected that full implementation of the v28 risk adjustment model for computing risk scores, along with maintaining the updated FFS normalization factor methodology to keep risk scores relative to 1.0, would lower the MA benchmarks by 3.01%, which reflects a 4.31% reduction for the risk adjustment and a 1.30% addition for normalization. CMS estimated that if the implementation to the v28 model is paused, MA spending would rise by $3.4B, likely reversing the above benchmark impacts. Separately, it indicated plans to further recalibrate the v28 model with MA encounter data for improved payment predictability, potentially starting in CY27. We think this idea may have legs with Trump’s CMS.

Coding Intensity Factor: CMS proposed to maintain the minimum statutory coding intensity cut of 5.90%, which is applied to risk scores for preventing risk score inflation. We note some investors were worried that, in a parting shot, the Biden administration would propose to lower MA risk adjustment spending via a: 1) hike the coding intensity cut much higher, ala 8%+; 2) requirement for two-years’ worth of diagnostic data to compute scores; and 3) disallowance of diagnoses from in-home risk assessments for use in risk adjustment. We were skeptical that they would be proposed, but we don’t rule out variations of them down the road.

Rebasing / repricing: As with prior years, CMS proposed to rebase and reprice rates to reflect the most recent data and price changes under legislation. We won’t know until the final rate announcement the impact of this change on the benchmark rates, but it tends to be between -0.50% and +0.50%.

Star rating change impacts: CMS proposed to update weights, retire quality measures, and solicit input for future ones, while affirming the goal to refocus more on clinical care and outcomes as opposed to process. It also sought input about adding geography as a component to the health equity index reward factor that is slated to begin in CY28. However, we are dubious that this policy will come to fruition under Trump, who seems uninterested in the health equity emphasis.

Risk score aka coding trend: CMS estimated a risk score trend of 2.10%, far less than prior years that used three-years’ worth of data pre-pandemic. It said it was changing its calculation to reflect more recent data that is representative of changes in population and coding practices but for CY26, it would only use two years’ worth of MA risks scores that were available after 2021 when COVID-19 was most prevalent. CMS intends to revert to the three-year timeframe of risk scores to project trend, starting in CY27.