Ch12 – MedPAC Report to Congress 2025

What you are reading here is a piece on the costs of Medicare Advantage Plans versus Traditional Medicare. Why is one so different than the other? The differences in MA and FFS are in coding intensity and selection. The later does not have such.

MedPAC March 2025 report to the Congress Pages 338 to Page 343

The Commission estimates MA payments (including rebates that finance supplemental benefits) substantially exceed what would have been spent on those beneficiaries had they been in FFS. This continues the trend of higher levels of payment throughout the history of Medicare’s payment policy for managed care. Before applying adjustments for favorable selection or coding intensity, we estimate that Medicare payments to MA plans have generally been similar to historical spending for FFS beneficiaries with both Part A and Part B coverage. However, the effects of favorable selection (prior to any coding differences) consistently cause the risk scores of MA enrollees to overpredict what the spending would have been in FFS. This increases MA payments by an estimated 11 percent ($44 billion) above FFS spending in 2025.

Also, diagnostic coding by MA plans overstates the health differences between MA and FFS enrollees are assumed in risk scores which further increases MA payments by an estimated 10 percent ($40 billion) above FFS spending in 2025. Twenty-twenty-five Medicare’s payments to MA plans will total a projected $538 billion (about $507 billion after excluding projected payments for enrollees with end stage renal disease). The estimate is based upon the differences between MA and FFS in coding intensity and selection.

Medicare spends 20% more for MA enrollees than it would spend if those beneficiaries were enrolled in FFS Medicare. The difference translates into a projected $84 billion or 17% of total payments to MA plans. This excludes end-stage renal disease paymentsin 2025.

The payments above FFS spending correspond with our projections of plan rebates in 2025, which we also estimate to be 20 percent of FFS spending and 17 percent of total payments to plans. These higher payments relative to FFS Medicare vary across MA organizations (as shown by the variation in the effects of coding intensity across MA organizations, among other factors). Additionally, these payments are above what would have occurred in FFS are not an estimate of plan profits and administrative expenses. The additional payments are the primary source of funding used to provide more generous supplemental benefits and better financial protection for MA enrollees. Such will attract enrollees and increase plans’ total revenues.

The estimate is reached first by projecting actual payments to MA plans in 2025. We project those payments (on average) will be equal to about 100 percent of CMS’s projections of FFS spending in 2025 (Table 11-4) and that benchmarks and bids will be 108 percent and 83 percent of FFS spending.

To accurately compare MA and FFS spending, ensure the two populations are comparable by accounting for differences in coding intensity and favorable selection.

• “Coding intensity” refers to the tendency for more diagnosis codes to be recorded for MA enrollees. Such causes risk scores for the same beneficiaries to be higher when they are enrolled in MA than they would be in FFS.

• “Favorable selection” in MA occurs when beneficiaries with lower actual spending relative to their risk scores tend to enroll in MA. It is the extent to which risk standardized spending of MA enrollees would be lower than the FFS average without any intervention from MA plans (including coding intensity).

As in prior years, we have made technical refinements to our methods for estimating coding intensity and favorable selection, and these updates are reflected in both projected and historical comparisons of MA payments and FFS spending.

Accounting for greater coding intensity and favorable selection in MA increases our projections of benchmarks, bids, and plan payments as a share of FFS spending. Indeed, we project that benchmarks in 2025 are 130% of FFS spending. In the aggregate, Medicare could spend up to 30 percent more for beneficiaries enrolled in MA than it would spend if the same beneficiaries were in FFS Medicare. Actual MA payments are 120 percent of FFS spending because most plans bid below those benchmarks. After accounting for coding intensity and favorable selection, plan bids in 2025 are 100 percent of FFS spending. What this means? On average, under the existing payment system, a plan’s estimate their costs to provide the standard Medicare benefit are about the same as FFS Medicare’s costs.

On average, 13 percent of plans’ bids are projected to be nonmedical expenses for administration and profit. Our review of private-plan payments suggests that over a 40-year history, the many iterations of full-risk contracting with private plans have never yielded aggregate savings for the Medicare program.

Evaluations of private-plan payment rates under Medicare demonstrations occurring before 1985 found payment rates were 15 percent to 33 percent higher than FFS Medicare spending (Langwell and Hadley 1990). Between 1985 and 2004, risk adjustment was inadequate, and researchers estimated that private-plan payments were 5% to 7% higher than FFS Medicare spending in the late 1980s and through the mid-1990s (Brown et al. 1993, Medicare Payment Advisory Commission 1998, Newhouse 2002, Riley et al. 1996).

Figure 11-3 shows that since 2007, payments to MA plans have been substantially above the amount FFS Medicare would have spent for the same beneficiaries, primarily due to the effects of favorable selection and coding intensity. Throughout the 19-year period from 2007 through 2025, we estimate that MA payments were at least 8 percent more than FFS spending for comparable beneficiaries in each year. Between 2011 and 2017, relative MA payments decreased from 23% above FFS spending to 16% above FFS spending. This change is largely explained by declining benchmarks resulting from ACA policies. The changes to benchmarks were fully implemented in 2017. The MA payments again increased relative to FFS spending through 2025 as driven by the combined effects of coding intensity and selection.

Figure 11-4 shows the higher payments to MA relative to what spending would have been in dollar terms if enrollees were in FFS. (In estimating the payment amount above FFS spending, we removed MA payments for beneficiaries with end-stage renal disease (ESRD), which we exclude from all of our analyses and estimate will account for about 6 percent of MA payments in 2025.30) We estimate that Medicare will pay MA plans a total of $84 billion more in 2025 than the program would have spent if enrollees had been in FFS Medicare. will account for about 6 percent of MA payments in 2025.30) We estimate that Medicare will pay MA plans a total of $84 billion more in 2025 than the program would have spent if enrollees had been in FFS Medicare.

Before accounting for the effects of coding intensity and favorable selection, MA payments are generally similar to what FFS spending would have been

The first component of the Commission’s payment comparison involves constructing a base comparison of MA payments with FFS spending for beneficiaries with similar risk scores and counties of residence. We begin with a base comparison because it aligns with how CMS constructs its estimates of risk-standardized FFS spending used to set county benchmarks. However, the base comparison does not account for how coding intensity and favorable selection cause CMS’s estimates of risk-standardized FFS spending to overpredict costs for the average MA enrollee. Those effects are incorporated as separate components in the Commission’s payment comparisons. The Commission’s estimate that Medicare payments to MA plans in 2025 average 100% of projected FFS spending (without any adjustments for coding intensity or favorable selection) reflects several
aspects of MA payment policy (Table 11-4, p. 337).

• First, benchmarks are set above projected FFS spending in some counties and below FFS spending in other counties.

• Second, benchmarks are increased based on plan quality scores which were resulting in quality-bonus payments that are financed with additional program dollars. We estimate that quality bonuses will account for about 3% (an estimated $15 billion) of MA payments in 2025. Combining both of those factors, we project benchmarks to be 108 percent of FFS spending in 2025 before accounting for coding and selection.

• Third, payments to MA plans are below benchmarks because bids are generally lower than benchmarks, and plans receive a percentage (based on quality score) of the difference between bids and benchmarks in additional payment.

The Commission uses historical data to estimate the effect of each of the three components on MA payments for the most recent year for which data are available (2022 for the base-comparison and favorable-selection components, 2023 for coding intensity) and earlier years. We then project what we expect each component to be through 2025 in years for which data are not yet available, to provide the Congress with our best estimate for the current year. Those projections are subject to uncertainty. In future reports, we will provide updated payment comparisons for those years using historical data.

The Commission uses historical data on the actual FFS spending of beneficiaries with both Part A and Part B coverage (that is, people who would be eligible to enroll in MA). This is to estimate the base payment comparison for the most recent year for which data are available (2022).31 Our estimates of the base payment comparison for years in which we do not have historical data (2023 to 2025) rely instead on CMS’s projections of FFS spending.

Those projections differ from the actual FFS spending of MA-eligible beneficiaries for several reasons, including, the projections include Part A. Only enrollees and that CMS’s projection of the FFS spending trend is subject to uncertainty. The method used to produce its FFS projections has been criticized because it includes beneficiaries who have Part A but not Part B coverage, while MA enrollees are required to have both Part A and Part B coverage. The Commission recognizes this shortcoming in the CMS methodology and previously recommended that CMS calculate MA benchmarks using FFS-spending data only for beneficiaries with both Part A and Part B coverage (Medicare Payment Advisory Commission 2021a, Medicare Payment Advisory Commission 2017).

Despite the shortcomings in CMS’s projections, our analysis of historical data shows the projections have not systematically impacted our base comparison of MA spending with FFS payments. We estimated the base comparison of MA payments with FFS spending using both CMS’s projections and the observed FFS spending of MA-eligible beneficiaries in each year from 2016 to 2022 (Figure 11-5). Both methods produced results within 1 percentage point of each other for every year except 2020 and 2021, which were affected by the coronavirus pandemic. The similarity of the estimates for the two methods indicates that relying on CMS’s projections of FFS spending (including adjustments for Medicare as a secondary payer) for estimating differences in MA and FFS spending has not introduced meaningful error in our analysis, and we expect that relying on CMS’s projections has a minimal effect on our payment comparisons for 2023 to 2025.

Prior to coding differences, favorable selection causes risk-based payments to overpredict spending for MA enrollees

The second component of the Commission’s payment comparison adjusts the base comparison of MA payments relative to FFS spending for favorable selection. When setting MA benchmarks and paying plans for each enrollee, CMS implicitly assumes that if MA enrollees were in FFS Medicare, their average Medicare spending would be equal to that of current FFS enrollees in the same county after adjusting for differences in risk scores. However, applying MA risk scores to FFS spending averages may overpredict the actual spending that MA enrollees would have had in FFS for two reasons: coding intensity and favorable selection. The Commission’s estimates of coding intensity are discussed in the next section of this chapter.

“Favorable selection” refers to the tendency for Medicare’s risk-adjustment model to (on average) overpredict the spending the MA-enrolled population would have had if they were enrolled in the FFS program, even for beneficiaries with similar coding intensity. Favorable selection can occur due to unmeasured differences in health status but can also result from factors such as differences in beneficiaries’ propensities to seek care for reasons that are unrelated to their health (Medicare Payment Advisory Commission 2024).

Risk models are imperfect, and beneficiaries with the same risk score typically have a wide distribution of actual spending relative to the spending predicted by their risk score (Lieberman et al. 2023). Each year, a mix of beneficiaries enroll in MA who either have lower spending than predicted by their risk score or higher spending than predicted by their risk score.

The MA program as a whole will experience favorable selection if the average MA enrollee has less spending than predicted by their risk score (prior to any effects from plans’ utilization management and coding efforts).

A substantial body of research has found evidence of favorable selection in MA and provides support for the Commission’s estimates and methodology (Brown et al. 2014, Curto et al. 2021, Curto et al. 2019, Fuglesten Biniek et al. 2024, Goldberg et al. 2017, Government Accountability Office 2021, Jacobs and Kronick 2018, Jacobson et al. 2019, Lieberman et al. 2023, Medicare Payment Advisory Commission 2023a, Medicare Payment Advisory Commission 2012a, Meyers et al. 2019, Newhouse et al. 2015, Rahman et al. 2015, Riley 2012, Ryan et al. 2023, Teigland et al. 2023). The Commission’s method uses data from the portion of the MA population who previously switched (i.e., “switchers”) to MA from the FFS program to estimate selection for the entire population of MA enrollees, including those who enrolled upon their initial eligibility for Medicare. Analyses of switchers have been used in other studies (Jacobson et al. 2019, Lieberman et al. 2023, Newhouse et al. 2015, Teigland et al. 2023). A recent white paper also estimated favorable selection among MA beneficiaries who enrolled upon their initial eligibility for Medicare and found estimates that were even larger than our estimates for switchers in similar years (Teigland et al. 2023).36 That study provides support for our approach that generalizes estimates based on switchers to the newly eligible population. We made several technical improvements to our methods this year, including updating certain components to more comprehensively account for mortality differences between the MA and FFS populations. More detail on our methodology for estimating favorable selection can be found in Technical Appendix 11-A (p. 377)

Evidence of favorable selection throughout the period from 2007 to 2022

We estimate that favorable selection of the overall MA population resulted in MA payments that were 10 percent above FFS spending in 2022, a slight decrease from 11 percent above FFS spending in 2021 (Figure 11-6, p. 344). The selection percentage was below 100% in every year during the 2016 to 2022 period, indicating that the spending that the FFS program would incur for the MA population would be lower than what would be predicted by their risk score. Between 2017 and 2022, the selection percentage increased from 89% to 91%. On net, favorable selection persisted throughout the study period even as a larger share of Medicare beneficiaries enrolled in MA, including enrollees who had higher risk scores. The estimates presented in Figure 11-6 use the Commission’s comprehensive method. We discuss that method in greater detail in Technical Appendix 11-A accompanying this chapter.

Using the simple method, we found evidence of favorable selection among recent switchers into MA during their last year in FFS throughout the period from 2007 to 2022 (Figure 11-7, p. 345). We estimate that the selection percentage ranged from 87 percent to 96 percent during the period. Favorable selection was highest in 2022 (MA enrollee local-area spending was 12.6 percent less than predicted by their risk score, a
selection percentage of 87.4).

MedPAC March 2025 report to the Congress, Ending Page 343

VVVVVV

MedPac Report to Congress on Medicare and Medicare Advantage – Angry Bear