MedPAC Findings Versus the Insurance Industry: Is MA Paid More Than Traditional Medicare?

Editor’s Note: This perspective was originally published in Health Affairs Forefront on April 7, 2026.

After years of explosive growth in enrollment, private Medicare Advantage (MA) plans now enroll 51.6 percent of all Medicare beneficiaries (35.5 million). As of October 2025, MA enrolled 56.0 percent of beneficiaries with both Part A and Part B, a requirement to enroll in MA. MA attracts enrollees by offering enhanced benefits, financed by plans receiving 15 percent more revenue ($2,660 per enrollee in 2026) than plans’ average annual cost of providing Traditional Medicare (TM) benefits. The attractiveness of MA has led to marked declines in both the number and share of individuals in TM, in spite of the total number of Medicare beneficiaries increasing significantly.

Recent decreases in MA gross margins have prompted insurers to take actions aimed at improving MA profitability, such as withdrawing from low-performing markets, reducing benefits, and more vigorously managing spending on health services. Insurers also argue for higher MA rates and against repeated findings that MA is overpaid. Insurance companies like UnitedHealthcare and advocates for plans contend that MA delivers important value to beneficiaries not available in TM. As part of a multi-faceted advocacy campaign, AHIP (formerly America’s Health Insurance Plans), the Blue Cross Blue Shield Association (BCBSA), and the Healthcare Leadership Council (HLC) have supported studies intended to undermine analyses showing that CMS pays more for MA enrollees than if they participated in TM; these groups have also sought Congressional support for legislation that would undermine the work of the Medicare Payment Advisory Commission (MedPAC), a non-partisan, highly-respected public agency that advises Congress.

The advocacy campaign attempts to discredit MedPAC’s findings assessing how much more Medicare pays plans than the program would pay if MA enrollees instead participated in TM, with the most recent estimate showing Medicare will pay MA plans 14 percent ($76 billion) more in 2026. Overpayment from favorable selection is one cause—this involves CMS paying average capitation to plans for beneficiaries whose expected costs are below average on a risk-adjusted basis without any management or other intervention from plans. Overpayment from upcoding is the second cause and reflects MA plans receiving inflated monthly capitation when they increase enrollee risk scores by coding additional diagnoses compared to norms in TM.

This contribution starts by describing the question that MedPAC annually addresses, which differs from the question that many MA advocates want answered. Next, it analyzes MedPAC’s methodology and findings, including how MedPAC has updated its analyses to reflect more recent data and introduce methodological refinements; we also discuss the results emerging from peer-reviewed literature. The third section analyzes the critiques of MedPAC’s findings made in two reports advanced by the industry. The final section contrasts the question MedPAC answers with the points raised in the reports sponsored by MA advocates.

MedPAC Compares MA Spending To What MA Enrollees Would Have Cost In TM

MedPAC answers a very specific budgetary question: How much more does Medicare, estimated by CBO to cost $1.1 trillion total in 2026, spend on MA enrollees than it would spend if those beneficiaries were enrolled in TM? Critics fault MedPAC for addressing the wrong question, wanting instead a focus on the value of MA, not its cost. However, as the MedPAC Chair Michael Chernew and Executive Director Paul Masi explain in Health Affairs Forefront, the purpose of the analysis is:

“to inform policymakers about the spending implications for the Medicare program if a beneficiary enrolls in MA versus TM. The Commission acknowledges that a portion of these increased payments to MA plans are used to provide more generous supplemental benefits and better financial protection for MA enrollees, such as lower cost sharing and reduced premiums.”

MedPAC acknowledges the analytic challenges and commits to continuing to refine its work:

“Estimating the effects of coding and selection on Medicare’s payments for beneficiaries in MA and TM is analytically challenging. Given the inherent complexity of this work, it is important for methods to be refined and estimates compared with each other. MedPAC devotes a chapter in its March Report to the Congress detailing the specific methods it used in its estimates and will continue to refine its methods over time.”

Chernew and Masi respond from a policy perspective to concerns that “the [MedPAC] estimate is flawed because it fails to account for differences in the benefit structures of MA and TM, or the practices of MA plans,” or how participants in MA and TM differ. They reiterate MedPAC’s focus on budgetary cost: “Considerations about MA value may be important for policy discussions. Yet, they do not affect the estimate of the spending gap.”

MedPAC Estimates Of MA Overpayments From Favorable Selection And Upcoding

MedPAC’s estimate of the added cost of MA beneficiaries (compared to if they were instead participating in TM) includes both overpayments from favorable selection and upcoding and the effects on MA costs and bids of several statutory requirements and other factors. These requirements chiefly involve bonuses associated with MA plan “star quality ratings” and the quartile system of setting MA benchmarks, which provides MA plans in lower-spending counties with higher percentages of Medicare fee-for-service spending than plans in higher-spending counties. As noted above, the key drivers of MedPAC’s estimate are (1) favorable selection, which arises when enrollees’ risk-adjusted expected costs are systematically less than the risk-adjusted revenue paid to plans and (2) upcoding, which refers to actions to more intensively code diagnoses in MA than in TM that inflate beneficiary risk scores and the monthly capitation paid to plans by Medicare.

Extensive claims data reliably detail spending and utilization for TM beneficiaries, but comparable data are lacking for MA enrollees. Even though the completeness and accuracy of MA “encounter data”—diagnoses, procedures, visits, and other care provided to MA enrollees—are improving, the lack of data comparable to TM claims data presents a fundamental challenge when estimating MA overpayments. As a result, the starting point for MedPAC and other analyses estimating overpayments from favorable selection focuses on data associated with beneficiaries who switch from TM to MA.

Risk Adjustment And Favorable Selection

Medicare has a highly skewed spending distribution that causes the CMS hierarchical condition categories (HCC) system of risk adjustment to overpay plans when prospectively trying to align capitation with expected costs. Multiple studies analyzing many years of data show CMS systematically overpays MA plans due to persistent favorable selection, in part because mean risk-adjusted expenditures are about three-and-a-half times the median and 70 percent of beneficiaries spend less than the mean. Risk adjustment does not adequately address the highly skewed nature of FFS spending: A 2023 study we coauthored shows virtually identical distributions of spending for all beneficiaries without risk adjustment and for beneficiaries with identical risk scores. (See exhibit 1 of 2023 study.)

Beneficiaries who switch from TM to MA have spending that is substantially below their risk-adjusted mean. Overpayments from favorable selection occur because CMS pays MA plans based on the mean risk-adjusted expenditure of a beneficiary, which on average substantially exceeds the expected cost. In two Technical Appendices, MedPAC details how it compares the base cost of the MA population to TM (12-A) and estimates favorable selection (12-B).

Upcoding Increases Risk Scores And Capitation Paid To Plans

In TM, the services recorded in fee-for-service claims determine reimbursement, but diagnoses have a quite limited role, and risk scores have no role at all. In MA, by contrast, the CMS risk adjustment model computes risk scores that, when multiplied by MA rates, increase or decrease plan capitation. The risk scores that adjust MA capitation, in turn, incorporate reimbursement and diagnoses reported on TM claims. Thus, additional diagnoses harvested by MA plans inflate enrollee risk scores—and plan revenue—compared to the HCC risk scores calibrated based on coding in TM. This provides MA plans with powerful economic incentives to invest in behaviors that generate additional diagnoses compared to the diagnoses reported in TM claims, as MedPAC explains in its most recent Report and in Technical Appendix 12-C.

Since first reporting MA overpayments, MedPAC has regularly improved its estimates by refining its methodologies, using additional data as they become available, critically assessing new published research, and having outside experts peer-review its work. MedPAC’s estimates align closely with academic articles measuring upcoding. A recent Forefront article noted that both CMS and MedPAC estimated that overpayments from upcoding in 2022 were 10 percent under the Version 24 (V24) risk adjustment model that was used then.

For 2026, MedPAC has reduced its estimate of overpayments from upcoding to 4 percent ($22 billion). Two main factors account for the decrease from the 2025 upcoding estimate, which, like 2022’s estimate, was 10 percent. First, MedPAC has incorporated more recent data that capture the effects on upcoding of the new CMS risk adjustment model (V28), which was designed to limit upcoding’s impact. Second, the 2026 upcoding estimate includes the full effects of the V28 model, which is now completely implemented after a three-year phase-in.

Using MedPAC’s “DECI” methodology to project upcoding overpayments under a scenario that assumed a fully implemented V28 model in 2022, CMS officials estimated that upcoding overpayments that year would have been in the range of 1.5 percent to 2.0 percent. CMS did not estimate upcoding beyond 2022, the last year for which actual data are available. Because MedPAC analyzed the effect of upcoding on spending in 2026, its estimate of 4 percent includes increases from an upcoding trend factor applied to the years after 2022, a routine practice also employed by the CMS Office of the Actuary when setting MA rates.

Although a number of factors complicate comparing CMS’s 2022 scenario and MedPAC’s 2026 estimate, the two estimates are generally consistent and reflect the significant reduction in upcoding associated with the V28 model. MedPAC and CMS both rely on MedPAC’s “DECI” methodology for computing overpayments from upcoding. (As explained in an earlier Forefront article by Chernew and coauthors, “the DECI method assumes that, after adjusting for demographic characteristics—age, sex, dual eligibility for both Medicare and Medicaid, and institutional status—MA beneficiaries have a similar prevalence of health conditions as FFS Medicare beneficiaries and that higher, diagnosis-based risk scores in MA therefore result primarily from more intensive coding.”)

Critiques Of MedPAC’s MA Work

MA advocates typically have a quite different focus and conflate the budgetary question with other attributes of MA, such as the value of extra benefits to enrollees and MA’s diverse enrollment, including lower-income beneficiaries. For example, in 2024, two current and two former MedPAC Commissioners rejected MedPAC’s estimate of MA overpayments from upcoding and favorable selection, stressed the value of MA, and mischaracterized MedPAC’s estimates of MA overpayment. Similarly, the September 2025 Healthcare Leadership Council report stressed the programmatic differences between MA and TM while contending that MedPAC’s methodologies for estimating overpayments from upcoding and favorable selection were flawed, but the report did not identify an alternative approach that MedPAC could adopt to answer its budgetary question.

Inovalon Study

Intended to help stakeholders “understand how the [MA and TM] programs are different,” the Inovalon study “Medicare Advantage Versus Traditional Fee-For-Service Medicare: Different Populations, Different Outcomes,” supported by AHIP, claims a “major advancement” in research that “provide[s] strong evidence of the value and efficiency of MA.” Inovalon combines “actual raw MA encounter-level claims data” with its proprietary data on “detailed patient characteristics, diagnoses, and healthcare utilization pre-65” and “information on beneficiaries’ social drivers of health (SDOH) characteristics.” The authors critique the findings reached by MedPAC and other researchers regarding the cost of MA compared to TM, upcoding, and favorable selection.

Inovalon’s findings are based on 10,287 individuals who were continuously insured for three years prior to age 65, joined MA plans within three months of turning 65, and remained continuously enrolled in MA for two years after that. This group was compared to 35,243 TM beneficiaries deemed similar to the MA enrollees based on statistical matching using multiple characteristics. While the approach is innovative, the generalizability and robustness of the findings are limited by the selected sample of 10,287 recent retirees and their “matched pairs” of TM beneficiaries, a universe that omits millions of individuals who enrolled in plans more than three months after turning 65 or are disabled or have ESRD, as well as people uninsured prior to age 65. Two other Inovalon studies similarly rely on matched pairs of recent retirees and TM beneficiaries; interestingly, one found greater favorable selection among beneficiaries who enrolled in MA upon their initial Medicare eligibility than MedPAC. While previous studies have reported the greater propensity of dually eligible, lower-income, and minority beneficiaries to enroll in MA, it is unclear that Inovalon adequately controlled for its “actual, raw MA encounter-level data” to address the well-documented limitations that have led MedPAC and other researchers to eschew relying on MA encounter data.

Inovalon concludes that MA plans experience unfavorable selection and decrease Medicare costs. The organization says its findings are based on “real world data,” contrasting its methods to MedPAC’s use of extensive TM data to simulate the costs of MA beneficiaries. The Inovalon findings contradict a substantial body of peer-reviewed literature (discussed by MedPAC on p. 389–390) and are reached by projecting MA costs, utilization, health status, and social and demographic characteristics based on a sample of recent retirees that is not representative of the 35.5 million individuals enrolled in MA plans.

FTI 2025 Report

The FTI Consulting report “Reviewing Methodological Issues in MedPAC’s Analysis of Favorable Selection in Medicare Advantage,” supported by AHIP and BCBSA, finds that “MedPAC’s conclusions are based on data which limit the scope of their findings and methodologies that affect the accuracy of their estimates.” FTI casts doubt on using TM but not MA data to estimate favorable selection, as well as ignores plan bid data and differences between MA and TM. MedPAC and academic researchers typically avoid relying on MA encounter data because these data are not sufficiently complete and reliable to be comparable to TM claims data. And neither bid data nor programmatic differences affect the dollars-and-cents question that MedPAC answers. In addition, MedPAC has significantly improved its methodology for assessing the scope of favorable selection as well as incorporating additional data.

Moreover, the combination of refinements in MedPAC’s methodology and sound programmatic reasons discount the significance of six additional factors that FTI says MedPAC does not consider but that could, in FTI’s view, impact MedPAC’s conclusions. For example, FTI criticizes MedPAC’s exclusion of beneficiaries with only Part A. However, MedPAC’s approach is consistent with the fact that only beneficiaries entitled to Part A and enrolled in Part B are eligible to enroll in MA; moreover, MedPAC now incorporates an adjustment for Part A-only beneficiaries. FTI also criticizes MedPAC for previously excluding MA beneficiaries with end-stage renal disease (ESRD) from its analysis, disregarding MedPAC’s commitment to include beneficiaries with ESRD once plans have sufficient data on this newly eligible category of beneficiaries—and in fact, now that sufficient data are available, MedPAC includes the impact of ESRD beneficiaries in its analysis of MA. MedPAC also now takes into account dual-eligibility status and Medicare as secondary payer in its methodology, obviating the FTI criticism that it insufficiently addressed these factors. Finally, MedPAC’s limiting of its analysis to beneficiaries who live in the same county over a given year likely has little, if any, impact.

FTI criticizes findings of favorable selection, concluding:

“Both the MedPAC and Schaeffer Center reports base their conclusions on insufficient data, overlook significant countervailing facts, suffer from methodological deficiencies, and misconstrue findings. A key limitation in their analysis is the lack of MA claims or encounter data.”

The Schaeffer Center findings of persistent favorable selection and the skewed distribution of risk-adjusted spending continue to be cited in scholarly articles; while only speculation, we note the FTI critique is consistent with AHIP’s unhappiness with those findings and the conclusion that Medicare overpaid MA plans by billion. (The authors of this piece were coauthors of the USC analysis; Ginsburg and Lieberman also discussed the themes in their analysis in a Forefront article they coauthored.) FTI states:

“A key limitation in their analyses is the lack of MA claims or encounter data. Instead, the reports create a proxy for the costs of MA beneficiaries using only fee-for-service data.”

FTI discusses potential issues that could complicate using TM data to create a proxy for the expected cost of MA beneficiaries. Especially in light of MedPAC’s methodological refinements, we question the validity and importance of the issues raised by FTI. More fundamentally, the objective of the FTI critique is to preclude answering the question MedPAC poses: How much more do beneficiaries cost Medicare by being in MA than TM?

MedPAC Credibly Answers A Budget Question That MA Advocates Avoid Addressing

We find the MedPAC analyses highly credible in assessing how much more Medicare pays MA plans due to favorable selection and upcoding than it would spend if MA beneficiaries were in TM. MedPAC answers its dollars-and-cents question employing a transparent, fully described methodology that is annually peer-reviewed. MedPAC staff carefully assess the relevant literature, painstakingly improve their methodology, incorporate new data and findings into their analysis, and evaluate how their results compare with those in the peer-reviewed literature.

MA advocates bridle at the finding that MA costs more than what Medicare would spend if the enrollees were in TM. In part, the industry studies are attempts to refocus from dollars and cents to less clearly defined notions of value, but they also raise issues with analyses by MedPAC and independent researchers. Beneficiary advocates want to preserve the extra benefits, decrease beneficiary costs, and expand the ability of plans to address vulnerable populations as well as social determinants of health.

Clearly, there are important public policy debates to be had regarding the $1.1 trillion annual cost of Medicare. Policymakers and stakeholders can discuss the fairness of the roughly half of beneficiaries enrolled in MA receiving a total of $76 billion—or an average of $2,660 per enrollee—in extra benefits that are not available to TM beneficiaries; these benefits are financed by the difference between what CMS pays MA plans and the plans’ cost of providing traditional Medicare benefits. The cost of the extra benefits for MA increases Part B premiums for all beneficiaries by an estimated $15 billion.

Policymakers should consider proposals that would reduce MA overpayments, with some of the savings used to expand benefits in TM and/or decrease Medicare spending. But all of those issues, important as they are, do not invalidate the factual question of whether payment to MA plans exceeds what would have been spent had the enrollees been in TM.

Authors’ Note

Paul Ginsburg served as a MedPAC Commissioner from 2016 to 2022, the last three years as Vice Chair.

Lieberman SM, Ginsburg PB, MedPAC findings versus the insurance industry: Is MA paid more than traditional Medicare? Health Affairs Forefront, April 7, 2026, https://www.healthaffairs.org/content/forefront/medpac-findings-versus-insurance-industry-ma-paid-more-than-traditional-medicare

MedPAC Compares MA Spending To What MA Enrollees Would Have Cost In TM