Medicare Advantage Overpayment: MedPAC Stands by $76 Billion

Medicare Advantage is the private-plan option within Medicare, the US health programme for people aged 65 and over. The government pays insurers a set amount for each person they enrol. Medicare's independent advisers, the Medicare Payment Advisory Commission (MedPAC), estimate that those payments run 14% higher than Medicare would spend on the same people in traditional Medicare.

That is about $76 billion in 2026.¹ On 22 September 2026, MedPAC's chair defended the figure at a forum run by the insurers' own lobby group.² KFF Health News carried the exchange in its national briefing.³ Insurers say it is too high, and a House committee has advanced a bill that would make MedPAC also publish a version that leaves part of the estimate out. This article explains what the number counts and why the two sides disagree.

What MedPAC estimates

MedPAC's March 2026 report says Medicare spends "approximately 14 percent more" for Medicare Advantage members than it would in traditional fee-for-service Medicare. That is "a projected $76 billion in 2026".¹

It names two causes. The first is favourable selection: people who cost less than their risk score predicts are more likely to join private plans.¹ MedPAC estimates this raises payments by roughly 11% in 2026.¹

The second is coding intensity. Private plans record more diagnoses for their members, which raises risk scores and payments.¹ MedPAC estimates risk scores end up about 4% higher than for similar people in traditional Medicare, even after the government's own adjustment.¹

LeadingAge, a group representing aged-care providers, reports that favourable selection accounts for $57 billion of the $76 billion.⁴

The estimate has moved

The figure is smaller than last year's. In March 2025, MedPAC put the gap at "approximately 20 percent", or a projected $84 billion in 2025.⁵ The 2025 release also put favourable selection at roughly 11%.⁵ That suggests the fall to 14% came mainly from the other parts of the estimate.

Why insurers say it is too high

The dispute centres on favourable selection. Humana Healthcare Research is the research arm of a large private-plan insurer. Its study, published in Health Affairs Scholar in July 2026, estimated favourable selection at 4.9% to 5.8% for 2020 to 2022.⁶ It says that is 42% to 65% lower than the 10% to 15% range often cited, including by MedPAC.⁶

The study attributes the gap to method. It used encounter data for the whole private-plan population, rather than extrapolating from a subset, it says.⁶ It is worth noting who produced it: an insurer with a direct financial interest in the answer. That does not make it wrong, but it is a reason to read both methods.

The insurers' lobby, the Better Medicare Alliance, argues that MedPAC's method does not account for structural differences between private plans and traditional Medicare.² Insurers also point to a study by the Centers for Medicare and Medicaid Services, released in January, with much lower estimates, Healthcare Dive reports.² That study was not available to review for this article.

The bill that would split the number

On 16 September 2026, the House Ways and Means Committee advanced HR 4093, the Apples to Apples Comparison Act.⁴ It would require MedPAC to publish two analyses: "one that accounts for favorable selection and another that does not."⁴

LeadingAge, which supports transparency in principle, warns that competing estimates "could make it harder for lawmakers and their staff to evaluate the costs of health policy reforms".⁴

MedPAC's chair, Dr Amol Navathe, responded at the insurers' forum on 22 September. "We hope that as we put evidence out there and raise issues, stakeholders will engage seriously and provide meaningful feedback," he said.²

What the headline number leaves out

The $76 billion is a projection, not a measured overspend. It depends on how favourable selection is estimated, which is exactly the point in dispute.¹⁶

It is also not a judgement on value. MedPAC found private plans spent roughly $24 billion in 2023 on supplemental benefits that traditional Medicare does not cover.¹ Whether that spending is worth the extra payment is a separate question.

How to read the next figure you see

Check which estimate is being quoted, and from which year: MedPAC's 2025 figure of $84 billion, or its 2026 figure of $76 billion.¹⁵

Ask whether favourable selection is included. If HR 4093 becomes law, MedPAC will publish versions with and without it, and the two will differ by tens of billions.⁴

And check who produced any rival estimate. A study funded by an insurer, or by the commission that advises Congress, has a point of view. The method is what lets you judge it.

Footnotes

1. https://www.medpac.gov/wp-content/uploads/2026/03/March_2026_MedPAC_Report_Press_Release_SEC.pdf — Medicare Payment Advisory Commission, press release on the "March 2026 Report to the Congress: Medicare Payment Policy" (published Mar 2026). Accessed Oct 3, 2026. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹

2. https://www.healthcaredive.com/news/medpac-medicare-advantage-overpayment-analysis-defense-bma/831126/ — Healthcare Dive, "MedPAC chair defends Medicare Advantage overpayment analysis at insurer lobby event" (published Sep 23, 2026). Accessed Oct 3, 2026. ↩ ↩² ↩³ ↩⁴

3. https://kffhealthnews.org/morning-briefing/thursday-september-24-2026/ — KFF Health News, Morning Briefing (published Sep 24, 2026). Accessed Oct 3, 2026. ↩

4. https://leadingage.org/?p=1173239 — LeadingAge, on HR 4093, the Apples to Apples Comparison Act (published Sep 2026). Accessed Oct 3, 2026. ↩ ↩² ↩³ ↩⁴ ↩⁵

5. https://medpac.gov/wp-content/uploads/2025/03/March_2025_MedPAC_Report_Press_Release_SEC.pdf — Medicare Payment Advisory Commission, press release on the "March 2025 Report to the Congress: Medicare Payment Policy" (published Mar 2025). Accessed Oct 3, 2026. ↩ ↩² ↩³

6. https://news.humana.com/press-room/press-releases/2026/new-humana-healthcare-research-study-finds-favorable-selection-i — Humana, "New Humana Healthcare Research study finds favorable selection in Medicare Advantage is substantially lower than MedPAC estimates" (published Jul 28, 2026). Accessed Oct 3, 2026. ↩ ↩² ↩³ ↩⁴

Show all 6 sources