After a 9 percent increase from 2021 to 2022, enrollment in the Medicare Advantage (MA) program is expected to surpass 50 percent of the eligible Medicare population within the next year. At its current rate of growth, MA is on track to reach 69 percent of the Medicare population by the end of 2030.

The subsidies fueling MA’s rapid expansion are well-documented. They include: benchmarks set explicitly above fee-for-service (FFS) spending in the traditional Medicare (TM) program; quality bonuses that add to total payments (in contrast to budget neutral bonuses for accountable care organizations [ACOs]); payment increases from diagnosis coding that elevates risk scores above 105.9 percent of TM levels, thus exceeding the current 5.9 percent coding adjustment designed to adjust for more aggressive coding in MA; and likely some residual favorable selection after risk adjustment. With these subsidies and any efficiencies generated from care management, MA plans have been able to attract an increasing share of beneficiaries by offering supplemental benefits (e.g., dental, hearing, and vision) and low (if not zero-dollar) premiums that distinguish MA as an increasingly appealing alternative to the combination of TM plus Medigap coverage.

MA’s rapid growth has stirred controversy, focused primarily on its exploitation of a manipulable risk-adjustment system. This has allowed MA to appropriate growing, unintended subsidies at a time when the Medicare Part A trust fund is projected to be depleted in four years. Debate has pitted concerns about fiscal sustainability and corporate influence against the value and popularity of benefits, which are politically challenging to contract once expanded. Indeed, the Centers for Medicare and Medicaid Services (CMS) recently proposed to maintain the status quo in MA payment for 2023, which would allow an 8 percent rate increase that is well in excess of forecasted spending growth in TM; the excess is primarily the result of continued divergence in risk scores between MA and TM.

Even if the coding adjustment was increased to maintain subsidies at their current levels, we would still expect MA’s expansion to continue; if subsidies are permitted to grow as proposed, we could see MA’s trajectory accelerate. MA expansion might still accelerate under current or somewhat lower levels of subsidies for other reasons. For example, employers have been shifting an increasing share of their covered retirees into MA to take advantage of the subsidies and relaxed treatment of network adequacy specific to employer-sponsored plans. In addition, growth may beget growth as MA becomes normalized as a beneficiary choice and Medigap markets contract.

Stepping back, the prospect of a rapidly diminishing TM raises other, broader questions about Medicare’s structure and future that deserve a national discussion we seem reluctant to have. No longer a mere annex to the TM edifice, MA is a massive wing, but it remains attached by statute, dependent on TM for its operation. If the purpose of subsidizing MA is to engineer a de facto restructuring of benefits and financing in Medicare, a codifying act of Congress will be needed soon to make the objectives explicit, establish a new basis for MA payments that is independent of TM, and set a sustainable course. However, the expected gains that would motivate an explicit move toward MA as the dominant, if not only, form of Medicare have been muddied by subsidies, and any deliberate restructuring must also consider the value of a viable TM as a competitor and constraint.

At this juncture, short of comprehensive reform by Congress, CMS may find it challenging to build value in Medicare over this decade if TM’s scaffolding erodes. Much can be done under CMS’s existing authorities to promote efficiency and equity, but, under Medicare’s present configuration, that requires preservation of TM. Without substantive legislative reform on the horizon, regulatory policy will thus need to keep the long view in mind, lest several years of inertia set in motion an unalterable course to a lesser outcome.

With these structural dynamics and Medicare’s long-term goals in mind, this article attempts a dispassionate policy analysis of three currently debated questions: 1) has MA been successful; 2) should we be concerned about its current trajectory; and, if so, 3) how should we think about a course correction?

Has MA Been Successful?

It depends on your definition of success. MA has been clearly successful in managing utilization more tightly than TM, as rigorous studies that account for non-random sorting of beneficiaries into MA have shown. Plans limit utilization via provider network restrictions, direct utilization review such as prior authorization, and strategic cost-sharing; changes in utilization due to redesigns of care delivery are less clear. MA has also been successful in sidestepping the need for legislation to expand benefits, and this has disproportionately benefited historically marginalized groups as the proportion of these groups in MA has grown, thereby advancing equity. However, the substantial subsidies MA receives are largely responsible for the extra benefits and have more than offset savings from any efficiencies, posing a net cost to Medicare and complicating assessments of MA’s added value.

Critiques of MA payment policy typically characterize MA subsidies as “overpayments” relative to what was intended by the program’s design at its inception or in its current form. Accordingly, FFS spending or benchmarks inclusive of quality bonuses (i.e., payments net of coding effects) are used as yardsticks of program integrity. But payments above these levels do not necessarily constitute overpayments, as MA’s net benefits could exceed its net costs—that is, the additional spending may be worth it.

Such a valuation, however, is very challenging for at least four reasons. First, evidence on the impact of MA on quality of care and health outcomes is less clear than its impact on utilization; much of the evidence is subject to selection bias. Second, the average value of the additional benefits may differ greatly from the incremental value. Rebates, the portion of the plan payment that is intended to finance extra benefits (equal to 50-70 percent of the difference between a plan’s bid and its benchmark), have grown rapidly as subsidies have grown. Over time, the rebate dollars allocated to directly measurable premium and cost-sharing reductions have plateaued, raising concerns that an increasing share of the rebate is retained by plans—e.g., allocated to supplemental benefits that count actuarially toward the rebate but may go unused by enrollees.

Third, the total size of the subsidies is unclear because of methodological challenges in estimating differences in coding intensity between MA and TM net of true differences in risk. MedPAC estimates this differential to be 3 percentage points greater than the 5.9 percent coding adjustment, whereas Kronick et al. estimate a much greater difference (~14 percentage points). Fourth, possible spillovers of MA onto FFS spending further complicate the calculus. In assessing MA’s relative costs, ideally we could compare MA payments to what FFS spending would be in the absence of MA rather than to observed FFS spending.

A virtue of neutral payments (an “even playing field”) would be that we could more directly judge the comparative advantage of MA based on enrollment (its relative attractiveness as revealed by beneficiary choices) at an equal—or lower—level of payment. But we do not have that luxury at present. At the very least, MA seems to have offered an option preferred by beneficiaries who are willing to accept some limitations on provider choice and utilization in favor of lower out-of-pocket costs—a mechanism for converting efficiencies into enrollee benefits, to some extent. But it also seems clear that recent growth in subsidies from coding intensity, and any further subsidy growth, could be put to better use.

All that said, success in policy is a relative concept, and what’s past is prologue. Even if one believes that MA outperforms TM now, subsidies notwithstanding, we must ponder how it would perform under alternative or future scenarios. For example, how well would MA compete without subsidies against a TM with an updated benefits package and robust ACO program? Without differences in benefit generosity, we might not see the salutary effects of MA on health outcomes suggested by some research. Also, studies of plan bids have consistently shown that MA plans only partially pass through payment increases to enrollees as extra benefits. Increasing the generosity of Medicare benefits directly would be an unambiguously cheaper way to add the same benefits if it were not for MA’s ability to control utilization.

But what if TM became more efficient than it is now? In contrast to the generous and growing subsidies for MA plans, benchmarks for ACOs in the Medicare Shared Savings Program (MSSP) have been designed to claw back savings as they are produced by ACOs individually or collectively, thereby weakening incentives to participate and save. While MA has grown from 39 percent to nearly 50 percent of eligible beneficiaries from 2018 to 2022, the size of the MSSP has stagnated over that time, constituting a falling share of the Medicare population. Essentially, conditions were set for MA to win, so it won.

Similarly, we must consider how successful MA would be as the dominant program, paired with a withered or extinct TM. To date, MA’s advantages have been achieved in competition against a “public” option. Thus, the debate over MA’s success so far, while instructive, may bear less than one may think on policy change if policy must change. Which brings us to the next questions.

Should We Be Concerned About MA’s Current Trajectory?

No matter what you think about MA’s success, its trajectory under current payment policy should give you pause for at least three reasons.

First, we cannot afford 8 percent annual spending growth in Medicare. If the Part A Trust Fund reserves constitute a true budget constraint, MA payments will need to be cut to avoid a 9 percent benefit reduction upon the Trust Fund’s depletion in 2026. Although Congress may act to replenish the Trust Fund through a limited appropriation to continue support of 100 percent of scheduled benefits, permitting health care spending growth to exceed GDP growth has consequences. As Skinner et al. recently reminded us, devoting an increasing share of GDP to health care means that we have less to spend on other things, some of which may be more important to health than health care. The impact is insidious but unmistakable and inequitable, for example eliminating an entire decade of wage growth.

MA subsidies may be disproportionately distributed to historically marginalized groups. But unless the subsidies are financed by a new tax on the rich, continuing them as MA grows will necessitate other inequitable actions—whether drawing down benefits for future beneficiaries, increasing regressive FICA (Social Security and Medicare) taxes, or reducing public spending on important social services such as housing and education. Efficiency in health care is critical for equity. Ideally, Medicare could lead the way by indexing its spending growth to GDP growth and encouraging more efficient and equitable use of the (already substantial) resources allocated.

Second, to the extent we care about evidence-based health policy, we should prefer a horse race that isn’t rigged. It may be the case that a MA-for-all restructuring of Medicare is sound policy, but ideally MA could be put to the test against a legitimate competitor and without a head start. One concern about MA is that insurer intermediaries add administrative costs. Another is that MA may not be as well-suited to transforming care delivery as population-based payment models in which Medicare contracts directly with providers (thereby ensuring integration of payment and delivery). Financial risk and the accompanying flexibilities may not be transmitted to providers as well through insurers, in part because successful risk-contracting with providers by one insurer may spill over to benefit its competitors (a free rider problem). More generally, MA arguably exacerbates the multi-payer problem in provider payment reform.

Third, and most elementary, today’s Medicare is not structured to support a dominant MA program. By statute, MA is entirely dependent on TM for establishing its payment rates. As MA grows, local FFS spending will no longer provide a reliable external benchmark. This is not a distant problem but an impending one. Already, some large urban counties are nearing or eclipsing 70 percent MA enrollment. While the ensuing selection pattern is hard to predict, it stands to reason that residual holdouts in TM will be those with higher demand for unrestricted provider networks and higher incomes, and thus higher utilization—and that those with lower demand who currently forgo supplemental insurance will increasingly take up zero-dollar-premium MA plans. If so, MA benchmarks could rise further above enrollee costs, acting to accelerate MA growth.

In the short-term, steps could be taken to expand the geographic basis for calculating MA benchmarks, but selection concerns will remain, and this will require reaching into increasingly distant areas. In Florida, for example, approximately 58 percent of eligible beneficiaries are already enrolled in MA (based on its raw percent in 2021 trended forward by the national MA growth and adjusted for the proportion of beneficiaries enrolled in Part A and B [91 percent]).  

A diminishing TM has other structural consequences. The MSSP may unravel as decreasing proportions of ACO patients remain in TM, introducing noise and complexities into benchmark and savings calculations, as well as increasing the effective size organizations must have to meet statutory requirements for participation. At the start of the MSSP in 2012, an ACO had to serve, on average, approximately 7,000 Medicare patients in total to meet the 5,000 TM beneficiary requirement. Now the necessary size is approximately 10,000 Medicare patients.

A receding MSSP would not be a concern if the loss was merely an inferior program, but the MSSP provides the Medicare program a unique regulatory opportunity to index benchmark growth in both TM and MA to an administratively set trend. Doing so not only greatly strengthens incentives for providers to participate and save in the MSSP, as it allows benchmarks to rise above FFS spending as ACOs slow FFS spending growth; in addition, the allowable benchmark trend could eventually be tied to GDP growth. Since MA benchmarks are already based on FFS spending, this could allow Medicare to set program-wide spending growth to a socially desired rate. A defunct MSSP would likely foreclose this option, as well as the opportunity to understand the relative merits of risk contracting directly with providers.

Likewise, the Medigap market could destabilize, if not collapse, in high-MA states. Several of the major insurers in the Medigap market are also major insurers in MA, which might hasten Medigap premium increases as other insurers exit in response to a shrinking TM. Not only would this accelerate MA growth, but it would also greatly weaken TM as a competitor, as TM benefits without Medigap are limited.

Already, the MA market is becoming increasingly concentrated, with the top 4 insurers accounting for 70 percent market share in 2021. The same studies demonstrating limited pass-through of payment increases to enrollees in MA also consistently show that pass-throughs are less complete in less competitive markets. As economic theory would predict, competition is critical to the functioning of MA. Continued consolidation coupled with a weakened TM could severely undermine competition in MA and thus its effectiveness. With a viable TM in the mix, at least there would always be one competitor. Finally, as MA grows and consolidates, so do concerns about regulatory and legislative “capture.”

Regardless of one’s politics or views about the motives of insurers versus providers, their relative influence as interest groups, or the sanctity of Medicare as an entitlement program guaranteeing access to any willing provider, it should be evident that the current trajectory is fraught. Even if we assume that an MA-for-all model for Medicare is the right policy direction, a passive back door strategy that allows these dynamics to play out could undermine the program’s goals and make successful reform harder.

How Should We Think About A Course Correction?

This is as much a political question as an economic one, but the key is to think and correct now. Ideally, Congress would act. Generally, there are two legislative paths. The first would strengthen TM by updating the benefits package and restructure MA to eliminate subsidies; establish a basis for setting and controlling MA payments when MA grows beyond some threshold in a region (e.g., indexing them to GDP growth); and ensure competition in such regions. MA benchmarks would not necessarily fall substantially under this option, as TM spending would rise as a consequence of the additional benefits.

How such a leveling up of TM would be scored is an interesting question; the answer would depend on estimates of how much of the additional spending would happen anyway under current law as MA (currently the costlier option) grows. Presumably, the cost of leveling up TM would be financed in part by eliminating MA overpayments related to coding intensity, but could also be financed in part by constraining MA spending growth to an external index should MA continue to grow. In concert, CMS and CMMI could continue to strengthen their portfolio of alternative payment models. This path would make TM a more viable program and competitor, establish an even playing field for MA to improve upon TM, and introduce explicit checks on spending growth and consolidation.

The second path would be to embrace MA without fortifying TM, sustain current MA benefits with only modest payment cuts to start, but enact strong mechanisms for controlling subsequent payment growth and ensuring adequate competition (and thus pass-throughs) in MA. This path would lead Medicare into uncharted territory. Recognizing that the U.S. has not excelled in legislating limits on health care spending or devising and enforcing stiff competition policy, we should think twice before discarding constraints availed by a public option whose absence is bemoaned in other settings. But either path would be an improvement over the status quo and could effectively head off the prospect of an increasingly bloated MA program without effective controls over competition or spending.

The likelihood of Congressional action in the near term, however, seems low. This leaves regulatory options. The administration, specifically CMS, can do much but also faces challenging constraints. Since CMS cannot change the basis of payment for MA or regulate MA market structure, the most viable regulatory path must attempt  the undertaking of the first legislative path described above but without the capacity to expand TM benefits directly—that is, preserve and strengthen TM enough to make it and, by extension, MA perform as well as possible until more definitive reform can occur.

Although MA benchmarks and quality bonuses are set in statute, CMS can increase the coding adjustment to claw back more of the payments that plans receive as a result of greater coding intensity in MA. This will need to go further than merely negating growth in the coding subsidy and begin to progressively reduce it while monitoring benefits. Doing so faces strong political headwinds to the extent that payment cuts amount to benefit cuts, but evidence of partial pass-throughs and lower incremental value of benefits in MA should ease this concern. Based on the popularity of MA two years ago, a 5-6 percent cut relative to proposed rates for 2023 (negating the scheduled subsidy growth plus a 2-3 percent reduction) should have a limited impact on valued benefits.

Further payment cuts may be tougher but necessary to preserve legislative paths to long-run success. Competition and spending controls are harder to introduce once lost. Conversely, continued growth in MA subsidies would only make precedents stickier as benefit differences widen and an increasing share of the Medicare population face the prospect of medical underwriting in the Medigap market if they switch from MA to TM. (Such medical underwriting is allowed in most states for those switching out of MA.)

CMS can also do much to strengthen the MSSP. Greater savings in the MSSP exert some downward pressure on MA benchmarks and frees up resources that ACOs can share with patients as enhancements. This pass-through mechanism is currently quite limited but could potentially be strengthened. Ideally, a cut of Medicare’s savings from an ACO could be devoted to buying down Part B and D premiums for the ACO’s patients. Other areas of regulatory discretion include reforming risk adjustment and addressing benchmarks in high-MA counties. Any system of risk contracting with providers or plans will require a risk-adjustment system that is not manipulable and thus does not invite costly gaming that poses a deadweight loss to society.

Such regulatory steps have been suggested by many. The purpose of this article is not to lay out new recommendations but rather to lay out why they are important and arguably urgent. At this point, regulatory options do not present a viable long-term strategy—MA growth will continue to force a reckoning—but they can avert undesirable developments by buying time for more decisive action. Even modest corrections can meaningfully alter trajectories, but only if they are made early enough in the course.

Author’s note

Dr. McWilliams reports serving as a senior adviser to the Center for Medicare and Medicaid Innovation and a consultant to RTI International and Blue Cross Blue Shield of North Carolina. He also reports receiving research funding from the NIA, AHRQ, Arnold Ventures, and the Commonwealth Fund. The content of this article solely reflect’s the author’s views and not necessarily those of any of the listed organizations with which the author is affiliated.