Higher and Faster Growing Spending Per Medicare Advantage Enrollee Adds to Medicare's Solvency and Affordability Challenges
The Medicare Trustees’ report includes data on spending per enrollee in private plans, including the bid component, rebate component, and total amount by plan type and overall, since 2010. We use the 2019 total per enrollee in this analysis.
To calculate what Medicare Advantage enrollees would have spent if they had remained in traditional Medicare we start with total spending for Medicare beneficiaries enrolled in both Part A and Part B reported in the CMS Geographic Variation PUF, excluding hospice. We subtract spending on direct and indirect graduate medical education (estimated using data from CMS FFS Data file). Next, we calculate county-level per-person spending in traditional Medicare for this group. We adjust those estimates to account for the health status of Medicare Advantage enrollees compared to traditional Medicare beneficiaries in the same county, using the Geographic Variation PUF data, CMS Plan Payment Data and MedPAC estimates of unadjusted coding intensity. Our estimate of the national average per person spending in traditional Medicare is then calculated as the Medicare Advantage enrollment-weighted county average per person spending and administrative expenses are added. Each step is described in more detail below.
Note, we are unable to compare projected Medicare payments per Medicare Advantage enrollee to future spending per person in traditional Medicare for comparable beneficiaries. This is because detailed information about key inputs into our calculation are not available, specifically: expected health status and risk score information for Medicare Advantage enrollees and traditional Medicare beneficiaries, the geographic distribution of Medicare Advantage enrollees, and county level spending projections
Total Part A and Part B Spending: Part A spending is the sum of spending in the following service categories: Inpatient, PAC: SNF, PAC: IRF, and PAC: LTCH. Part B spending is the sum of Outpatient, Ambulatory Surgical Center, E&M, Procedures, Tests, Imaging, DME, Outpatient Dialysis Facility, FQHC/RHC, Ambulance, Part B Drugs, Other Services, and PAC: HH spending. In addition, Total Population Based Payment Reduction Costs are added to total spending. This category includes payments made to ACOs and other population-based-payment program payments. Note, we do not exclude disproportionate share hospital (DSH) program spending because the funding formula for Medicare Advantage plans includes an amount to cover these payments and may influence the prices negotiated between private plans and hospitals.
GME and IME: To estimate the amount of GME and IME spending to subtract, the GME and IME Part A spending shares in the FFS data are applied to Part A spending in the Geographic Variation file.
County Per Person Spending: The per person spending for traditional Medicare beneficiaries is calculated by dividing the sum of total spending, as described above, by the number of beneficiaries enrolled in Part A and Part B in the county.
Health Status and Unadjusted Coding Intensity: CMS provides the average risk score by county and plan type through 2017. We use these data and the Medicare Advantage enrollment data to calculate an enrollment weighted county average risk score for 2017. We then calculate the ratio between the Medicare Advantage risk scores and the average HCC score for traditional Medicare beneficiaries in each county. Since plan payment data are not available for 2019, we take the 2017 risk ratio and increase by the growth in the ratio of Medicare Advantage to traditional Medicare risk scores reported by MedPAC (2.0 percentage points between 2019 and 2017). To account for unadjusted coding intensity in 2017, we reduce the risk ratios by the Medicare Advantage coding impact on payment reported by MedPAC (1.4 percentage points in 2017, 3.2 percentage points in 2019) (Figure 12-7). There is some disagreement about the level of upcoding in Medicare Advantage. By using MedPAC’s estimates, we incorporate the more conservative estimates of coding intensity. For comparison, a recent brief by the Center for a Responsible Federal Budget suggests that coding intensity increased risk scores by 15.4 percent in 2017, more than twice MedPAC’s estimate (which was 7.1 percent in 2017, including the statutory adjustment of 5.7 percent in that year, which is reflected in the Medicare Advantage payment data in the Trustees report). The adjusted risk score ratio and the county per person spending estimate are then multiplied to get the county per-person spending for traditional Medicare beneficiaries.
National Average Per Person Spending: The contribution of each county to the national average is equal to its share of total Medicare Advantage enrollment in the analysis sample. The county per person spending amounts are multiplied by their county’s respective weight and summed. Note, we do not have complete information for counties with fewer than 11 traditional Medicare beneficiaries or fewer than 11 Medicare Advantage enrollees, and so these counties are excluded in our calculations. We also exclude counties where any category of spending is not reported due to too few beneficiaries contributing to that category of spending in the county.
Administrative Expenses: Private plans use payments received by Medicare to cover their administrative expenses, as well as the cost of Part A and B services and supplemental benefits. The data used to calculate spending among traditional Medicare beneficiaries does not include administrative expenses. We account for this by assuming that administrative expenses for the subset of beneficiaries enrolled in both Part A and Part B of traditional Medicare was the same share of Part A and B spending as it was for the full population of beneficiaries enrolled in either Part A or Part B, or 2.8 percent in 2019.
Decomposing projected growth in Medicare Advantage spending
To determine the drivers of Medicare Advantage spending growth, we first converted the Medicare actuaries’ projections of Medicare Advantage spending to 2021 dollars using the gross domestic product implicit price deflator. We then decomposed the change in annual spending into two components: growth in enrollment and growth in spending per person. Those are defined as follows.
Total spending: TSt = St * Et
- t is the year
- S is average spending per person, and
- E is total enrollment
The change in total spending in two consecutive years:
Where, the first term is the change due to the change in enrollment, the second term is the change due to the change in spending per person, and the third term is the combined effect of enrollment and spending per person. In each year, we allocated the combined effect to either enrollment or spending per person, based on the relative share of the first and second components. To calculate the cumulative impact of each component through 2029, we added the annual estimates together.
To calculate projected growth in spending per person for traditional Medicare we first calculate Part A spending per beneficiary in traditional Medicare by dividing Part A fee-for-service spending from table IV.A3 of the Medicare Trustees Report, less hospice, by the number of Part A enrollees, less part C enrollees, in table V.B3. We then calculated Part B spending per beneficiary in traditional Medicare by dividing Part B fee-for-service spending from table IV.B6 by the number of Part B enrollees, less part C enrollees, in table V.B3. Projected spending per traditional Medicare beneficiary is the sum of these two numbers. We then calculated the average growth rate between 2021 and 2029. Note, these per-person spending amounts are not directly comparable to the amounts we estimated for 2019 using the geographic variation public use file data, because they do not adjust for the distribution of Medicare Advantage enrollment or differences in health status between traditional Medicare beneficiaries and Medicare Advantage enrollees. We calculate growth in payments per Medicare Advantage enrollee using the data in table IV.C3. Data on private health plan enrollment by plan type comes from table IV.C1.
Compared to the Medicare actuaries’ most recent 2020 projections, the Congressional Budget Office (CBO) March 2020 Medicare baseline assumed that approximately 5 million more beneficiaries will be enrolled in Medicare Advantage in 2029. The CBO and OACT have similar estimates of Medicare payments per enrollee. As a result, Medicare Advantage spending is projected to grow more by 2029 under CBO’s baseline than the Medicare actuaries projected in 2020, due to higher enrollment growth. In its most recent July 2021 Medicare baseline, CBO increased its Medicare Advantage enrollment projections, and now expects 8 million more beneficiaries to be enrolled in Medicare Advantage by 2029 than was projected for that year in the 2020 Medicare Trustees Report. CBO did not, however, provide updated estimates of Medicare Advantage spending. Assuming no changes to spending per enrollee, higher enrollment growth would lead to even higher Medicare Advantage spending.