Over the past seven years, Medicare Advantage’s (MA’s) enrollment has almost doubled, adding 10 percent to its market share , now at 49 percent. Rebates, the additional dollars paid by CMS to MA plans that bid beneath their “benchmark,” have doubled in that time from $80 per beneficiary per month (PBPM) to $164. (Benchmarks are intended to represent the average per-beneficiary spending in traditional Medicare (TM) in a given service area.)
The Medicare Advantage industry’s explanation of its success is grounded in claims about MA’s ability to deliver Medicare Part A and B benefits for much less than TM. These savings are, in theory, the basis for the rebates, the incremental revenue CMS pays to plans that fund the improved benefits and lower premiums as compared to TM, which in turn help attract members to MA plans. Indeed, the Medicare Payment Advisory Commission (MedPAC) reports that MA bids average 85 percent of the FFS cost. Because these bids include approximately 15 percent for administrative costs and profits, they imply that Plan medical cost savings must be in the range of 25 to 30 percent versus the Medicare FFS cost benchmark bidding target.
However, a close examination of the bid process reveals that most of these savings are artifacts of the process and not due to better or more efficient care. They result from including “induced utilization costs” from Medicare supplemental insurance, legislated increases in the benchmarks, and risk score gaming. The inflation of benchmarks and risk score gaming, not better care, finance the rebates that drive MA market success.
To paraphrase Barry Switzer , MA was born on third base and thinks it hit a home run.
The MA Bid Process
CMS pays MA Plans a per-person revenue amount that is determined by the Plan’s bid to provide Part A (hospital) and Part B (all other medical services) to enrollees. Plans bid against a benchmark, which as noted is intended to capture the amount that Medicare would spend on TM benefits for an average TM beneficiary. If the bid is less than the benchmark, CMS keeps about one-third of the difference and pays two-thirds to the Plan as a rebate. This rebate can be used to improve benefits or reduce costs for the members. If the bid is above the benchmark, there is no rebate, and the Plan must charge the member premiums to make up the difference.
Bids include the cost of medical services as well as plan administrative costs and profits. Most plans bid sufficiently below the benchmark to offer members a “zero premium” product, often including Part D drug coverage. Conceptually, the difference between the benchmark and the bid represents “savings” that the plan generates that decrease CMS costs.
Real Vs. Apparent Savings
The difference between bids and benchmarks, i.e., the savings vs. FFS, and rebates have doubled over the past seven years, leading to improved benefits, lower premiums for members, higher profits and more rapid growth. In 2022, rebates were $164 PBPM and 66 percent of beneficiaries were in zero-premium products.
This suggests that savings for CMS have increased; however, the reality is that most of these are just “apparent savings”—not real savings—that increase costs for CMS, beneficiaries and taxpayers.
Here is how that happened.
Benchmarks Are Significantly Inflated By Including The Costs Of ‘Induced Utilization’
The total cost of care is a function of the price paid per service and the number of services patients receive. Because MA Plans are given the right to use CMS’s Medicare pricing schedule for all Medicare participating providers, the MA average price per service tends to be about the same as Medicare’s. Most savings in MA then must be due to changes in utilization of services. Is the 25-30 percent implied savings of MA really due to 25-30 percent lower utilization across the full set of health care services, or is something else leading to “apparent savings”?
The ‘Induced Utilization Effect’ Of Medicare Supplemental Insurance Leads To Higher Utilization And Costs In The TM Population
Health insurance benefits programs vary by the percentage of costs paid by the covered individual. First-dollar coverage (FDC) means that the insurer pays most of the cost of services. Non-first dollar coverage (NFDC) with deductibles, coinsurance, and copays creates financial hurdles for patients as they pay more of the cost. Actuaries have shown that populations with FDC use more services and have higher total costs. We use the term “induced utilization” to denote the additional services associated with FDC. If one assumes that the additional services are necessary and contribute to better health, this difference is better framed as “forgone services” by the population with NFDC.
TM’s population provides an ideal context to study this phenomenon. TM’s fee-for-service benefits cover about 84 percent of medical costs but the vast majority of TM beneficiaries (84 percent) have supplemental insurance coverage that covers the other 16 percent of costs, effectively giving them FDC. MedPAC commissioned two studies to examine the difference in utilization between beneficiaries with and without supplemental coverage. MedPAC cited the first study in their June 2012 report on reforming Medicare’s benefit design, concluding:
The study estimated that total Medicare spending was 33 percent higher for beneficiaries with medigap policies . . . Beneficiaries with employer sponsored coverage had 17 percent higher Medicare spending
The authors of this study updated it in 2014, using three years of additional data, through 2008. Their conclusions at that time estimated that Medicare spending was 25 percent higher for beneficiaries with Medigap policies and 14 percent higher for those with employer-sponsored coverage. Another 2019 study on induced utilization showed that Medigap increased utilization by more than 20 percent.
The way the costs from induced utilization inflate the MA benchmark calculation has major implications for the calculation of MA Plan rebates. As we shall see below, the current approach gives MA plans a massive head start on financial success, no matter how well or poorly they manage care or costs.
Effects Of Induced Utilization On TM Spending Flow Through To MA Benchmarks
At a high level, the MA benchmark is based on the average total cost for all TM beneficiaries. The 2022 TM average cost of $1,086 per beneficiary per month (PBPM) is the average of the total medical costs for two Medicare populations: those with additional coverage (TM + Coverage) and those with Medicare only (TM only). Exhibit 1, using the differences in spending cited by Hogan et al in 2014 above, demonstrates the underlying average costs for individuals with and without additional coverage. The TM-only population costs CMS $920 while the TM+ Coverage population costs $1,169. The overall weighted average cost of $1,086 is inflated above the TM-only cost by $166, or 18 percent.
The Expected Cost Plans Use In Their Bids, Based On TM-Only Benefits, Are Far Below The Inflated MA Benchmark And Result In Large “Apparent Savings” And Rebates
The MA bid process instructs plans to bid their expected cost using the standard Medicare package of services and benefits. Any improved benefits and resulting costs are part of the supplemental benefit information that explains how they will use the rebates. The intent is for the plans to demonstrate their ability to drive significant savings versus CMS’s cost. One would think that this should be a comparison of the bid with the TM-only population’s cost. But the MA benchmark used in the comparison is based on the overall average of costs for the TM +Coverage and the TM-only populations, thereby including the induced utilization costs. When the lower expected cost is subtracted from the inflated benchmark it automatically creates “apparent savings” of $166 PBPM, or 18 percent.
Exhibit 2 illustrates how these “apparent savings” roll through the MA bid to create rebates for the plans. Our model in exhibit 2 is based on MedPAC’s analysis of the industry-wide 2022 MA bids, which showed an average rebate of $164 PBPM. That analysis, combined with the Medicare 2022 average TM cost of $1,086, implies that the expected medical costs used in the bids averaged approximately $790, an actual medical cost savings of $130 PBPM (14 percent) compared to the $920 TM-only cost.
Column 1 shows what would happen if the benchmark were set at the TM-Only cost of $920 with an average risk population. With assumed administrative costs of $80 PBPM (8 percent) and profits of $50 (5 percent) the resulting bid would be $920. Because the bid and the benchmark would be both $920, plans would show no savings and receive no rebate. The total savings in the Bid are just the medical cost savings of $130, which fund the plan administrative costs and profits. With no rebate they would have to charge members for any improved or supplemental benefits. This would not be a formula for success in MA.
Column 2 uses the inflated benchmark of $1,086 that includes the $166 induced utilization effect. Medical costs do not change, the bid remains the same, and the $166 becomes the difference between the benchmark and the bid; two-thirds of the $166 becomes the rebate of $108. The total savings implied in the bid increase to $296, but 56 percent is due to “apparent savings,” which account for 100 percent of the rebate.
But this is still well below the reported 2022 $164 average rebate. Is more of this driven by plan medical cost savings?
Legislated Payments Above FFS Cost Further Inflate Benchmarks And Contribute To Apparent Savings And Rebates
The MA average national benchmark of $1,086 is based on Medicare’s national average cost for TM in 2022, as reported by CMS. This is what Medicare pays for all Part A and B services for an average beneficiary across the country. Actual MA rates are set at the county level and are adjusted from 95 percent to 115 percent of FFS Medicare expenses depending on whether a county has high or low costs relative to the national average. Approximately 80 percent of MA enrollees are also in plans that receive quality bonuses.
County bonuses and quality bonuses are added to benchmarks. According to MedPAC, in 2022 these bonuses accounted for an additional 8 percent increase in payments above the FFS cost, and 90 percent of MA members were in Plans receiving quality bonuses. MedPAC again reiterated at its January 2023 meeting that the “quality bonus program is not a good way of judging quality for the 49 percent of beneficiaries in MA.” Column 3 starts with a benchmark that is inflated $87 (8 percent) more to account for these bonuses. The bid is unchanged, increasing the difference to $253 and the rebate to the $164 reported by MedPAC. Of the savings implied in the bid, 66 percent is from benchmark inflation, as is 100 percent of the rebate.
Risk Score Gaming Acts As A Multiplier Of This Benchmark Inflation
As described in our prior article, MA Money Machine Part 1, plans systematically increase their risk scores to improve payments from CMS. Column 4 in exhibit 2 illustrates the results for a plan that increases its risk score from 1 to 1.1. In the bid process, the benchmark-bid difference is computed by comparing the actual bid to the risk-adjusted benchmark. Our prior examples had an average risk score of 1, so the risk-adjusted benchmark is the same as the benchmark. In Column 4 the benchmark is risk adjusted by multiplying the $1,173 from Column 3 by the 1.1 risk score, resulting in an increase of $117.
While the higher risk score might suggest that the population is sicker, that is an illusion created by the risk score game. The medical costs do not change. The reality is the population is the same; the plan has just collected more codes that make the population look sicker. We recently presented an example of this using data from a United Health Group (UHG)/Optum Team Study that included a comparison of HCC coding rates for FFS and MA populations.
The bid therefore remains the same. The difference has increased to $370, resulting in a rebate of $241. The savings implied in the bid increase to $500, but 74 percent of these and 100 percent of the rebate are apparent savings from benchmark and risk score driven inflation.
George Halvorson argues that the change to the Encounter Data System (EDS) from the prior Risk Adjustment Payment System (RAPS) eliminated the risk score gaming opportunity. He seems to miss the point that plans’ chart reviews, home visits and annual wellness visits drive large coding opportunities unrelated to real clinical care as the Department of Health and Human Services Office of Inspector General described in 2021. A Milliman survey report in 2020 found that EDS Scores were becoming higher than under RAPS. MedPAC just confirmed this, showing the MA coding intensity has increased at a more rapid rate since the change.
The MA Bid Process Allows Plans With Zero Cost Savings To Offer Zero Premium Products
Column 5 of exhibit 2 shows that even if the plan has no actual medical cost savings, and no increase in the risk score, the benchmark inflation from induced utilization and bonuses allows the plan to have a rebate of $100.
MedPAC has reported that rebates vary widely, suggesting that there are indeed real MA plans today that are delivering no improvement in medical costs vs. FFS but still are receiving rebates and offering zero premium products.
Even Plan Medical Cost Savings Are Uncertain
Most of our examples assume that MA beneficiaries have a risk score of 1, that is they have the same health burden as an average Medicare population. Multiple studies have shown that this is not the case. Jacobson et al demonstrated that beneficiaries who enrolled in MA in 2016 were 16 percent less costly than individuals who stayed in TM. Other researchers have used mortality rates as evidence that individuals choosing MA are healthier than those in TM. If MA beneficiaries are actually healthier and have lower medical costs, the 14 percent “real” cost savings we use in exhibit 2 would be overstated.
ACOs Do Not Benefit From Any of These Subsidies
In an earlier paper, Joe Antos and Jim Capretta asserted that “There is little question the MA plans have the capacity to deliver Medicare benefits at far less cost than unmanaged fee-for-service” and further that accountable care organizations (ACOs) deliver “far less that the savings that could be achieved by MA plans based on their bids.” While we disagree with both statements, we agree on one point: the savings implied in MA bids seem large. But they are mostly apparent not real savings.
ACOs have a very different starting point from MA plans. They don’t start with the 18 percent advantage demonstrated above, they don’t get county and quality bonuses and they are not able to benefit much from risk coding. The costs of induced utilization are in their benchmarks, but their aligned beneficiaries are representative of the TM mix of people with FDC and non-FDC. The benefits and the resulting costs match, unlike the MA bid comparison of average costs and lower plan expected costs. Any comparison of savings by the two programs needs to account for this reality. We believe that most of the differences in the “real savings” of the two programs is accounted for by claims denials and some instances of lower prices in MA vs. FFS.
MA is growing rapidly because of plans offering lower premiums and improved, supplemental benefits.
Claims by the MA industry that they are successful because they deliver more efficient care are flawed because they compare MA costs to inflated benchmarks that are much higher than the actual CMS costs to provide benefits.
Even if one assumes that MA does decrease medical costs, these savings as reflected in MA bids are not the drivers of the rebates or MA success. Indeed, 100 percent of rebates result from inflated benchmarks and risk score gaming. The resulting payments, which are in excess of TM costs, generate the additional funding for “free-to-the-member” improved benefits. While there is large variation, and some plans improve care and decrease utilization, MA industry success is a function of corporate subsidies.
The bottom line: we are systematically driving people out of TM by subsidizing the more expensive MA.
Inflated benchmarks from these three sources start MA plans “on third base” and risk score gaming gives them a free walk home. For MA plan owners it feels like they hit a home run. For taxpayers and Medicare beneficiaries footing the bill it feels like a series of major errors. CMS in its 2024 Medicare Advantage Advance Notice has proposed significant changes to the Risk Adjustment system. We believe this would be an important step forward towards addressing the vast overpayments to MA plans and deserves our support.