Policymakers have long sought to move traditional Medicare (TM) away from its fee-for-service financing model, which rewards volume of care over quality, relies on complex and politicized administrative pricing, and inhibits care innovations that do not fit neatly into its framework. The two main alternatives that have emerged over the years are Medicare Advantage (MA) and accountable care organizations (ACOs). MA has grown to over half of Medicare enrollment, and officials at the Center for Medicare and Medicaid Services (CMS) have set the goal of having every TM beneficiary in an accountable care relationship by 2030. This brief explains each of these models and highlights their structural differences and similar policy challenges.
Key Takeaways
- Medicare Advantage (MA) and accountable care organizations (ACOs) offer population risk-based payment methods for health care insurers and providers, respectively, as alternatives to traditional fee-for-service Medicare.
- MA and ACO models have different designs in terms of enrollment, financing, and care management rules, but they face common policy concerns related to payment benchmarks, risk adjustment, quality measurement, and selection biases. These factors in turn affect savings, program participation, quality of care, and access.
- These policy factors have implications for the levels of savings that each program can achieve for Medicare, which evidence suggests can be improved.
Background
MA is a program within Medicare where private insurance plans provide coverage to Medicare beneficiaries and pay health care providers for their services. ACOs are groups of health care providers that agree to be held accountable for the cost and quality of care for groups of beneficiaries in TM.
Managed care companies began to participate in Medicare at a limited scale in the 1970s. Congress formally incorporated them into Medicare Part C in 1997 under the name “Medicare+Choice.” In 2003, it updated Part C’s design and renamed it Medicare Advantage. Medicare pays MA organizations (MAOs) based on a process where their plans submit bids against TM spending targets called benchmarks. In 2023, 184 MAOs offered about 5,600 plans to 31.6 million MA enrollees, with total non-drug payments of about $455 billion.1 MA enrollment has grown rapidly to over half of the Medicare population, driven by its ability to innovate in benefit offerings and manage the cost of care. As such, it offers the clearest alternative to the shortcomings of TM and is a vehicle for greater coverage choice and market competition in Medicare.2
Congress first included ACOs in Medicare through the Affordable Care Act, although the concept predates that law. Congress directed CMS to establish a voluntary payment model starting in 2012 called the Medicare Shared Savings Program (MSSP), where participating health care providers and suppliers establish ACOs through networking agreements. ACOs must include primary care providers and may include specialists and hospitals, but there is typically a delineation between physician-group-led and hospital-led ACOs. Providers within ACOs receive payment for furnishing health care services but keep a portion of “shared savings” if they reduce costs below a benchmark spending target calculated from historical spending data. There have also been several other ACO models with different rules, such as ACO Realizing Equity, Access, and Community Health (REACH), which relies on direct contracting between CMS and providers as risk-bearing entities.3 In 2024, there were 480 ACOs (roughly 635,000 providers) and 10.8 million people in MSSP as well as 122 ACOs (roughly 173,000 providers) and 2.6 million people in REACH.4 CMS officials have suggested moving all TM enrollees to an accountable care relationship by 2030.
ACOs and MAOs do not directly compete against each other, as ACOs receive shared savings for TM beneficiaries and MAOs pay for services received by MA beneficiaries. However, ACOs indirectly affect competition for enrollees between TM and MA.
Design Differences
Enrollment
A key difference between MA and ACOs is that Medicare beneficiaries may choose to enroll in MA plans (new beneficiaries are enrolled in TM by default), while Medicare generally assigns TM beneficiaries to ACOs if they receive services from ACO-aligned primary care providers. However, TM beneficiaries may also voluntarily join ACOs or, conversely, opt out of sharing certain information with ACOs to which they are assigned. An ACO is still responsible for the cost of Medicare services its assigned beneficiaries choose to receive from other providers outside of the ACO.
Because joining an MA plan is voluntary, there is more competitive pressure to attract and retain enrollees (for example, plans must devote more resources to marketing). An ACO must provide written notice to enrollees who become aligned with the ACO, according to federal guidelines. ACOs also have some incentive to advertise—for example, to encourage beneficiaries to see aligned providers or to gain voluntary enrollees. MA plans must submit their marketing communications to CMS for review, a requirement that CMS removed for ACOs starting 2023.5
Financing
Medicare pays MA plans and ACOs in a manner intended to incentivize more efficient care. MA payments are fully capitated, which means that they are set on a per-member per-month basis, rather than per service or based on actual costs incurred. If a plan’s payments exceed its costs, it keeps the full difference, but likewise it bears the full loss if the opposite is true. CMS also requires MA plans to meet a medical-loss ratio where at least 85 percent of revenue goes toward health care and quality improvement, as opposed to administrative costs or profits. This creates a strong incentive to hold down costs, although plan competition and CMS regulations discourage cost reductions that undermine the quality of coverage. MA plans can implement various methods for payment, including adopting a similar fee-for-service model as TM.
ACOs face difference levels of financial risk based on the models they participate in. ACO models may allow for upside-only arrangements, particularly for new participants, but require them to gradually bear more downside risk—alongside higher upside reward—over time. In MSSP in 2024, there were five levels of risk. At the low end, ACOs keep 40 percent of cost savings up to a limit of 10 percent of the benchmark without sharing in any losses. At the high end, they keep 75 percent of savings up to 20 percent of the benchmark while bearing up to 75 percent of losses up to 15 percent of the benchmark.6 REACH has two risk levels: 50 percent of savings and losses and 100 percent of savings and losses. REACH also offers the option for capitated primary care payments.7
Care Management
Because MA plans are payers of health care services, they have greater visibility into beneficiaries’ total cost of care. Unlike TM, they may contract with networks of providers—including health maintenance organization (HMO) MA plans—and conduct utilization management. Thus, MA beneficiaries have higher cost-sharing when visiting out-of-network providers and may be subject to explicit prior authorization for care. MA plans often integrate Part D prescription drug coverage into their plans and may develop specialized plans for particular populations. MA plans are also required to cap out-of-pocket expenses for their enrollees, while TM beneficiaries must purchase supplemental coverage to do this.
ACOs can only directly manage the cost of care provided by its aligned providers and are subject to TM rules limiting the establishment of networks or utilization management. This arrangement gives beneficiaries aligned with ACOs more flexibility in choosing providers or obtaining care, but it constrains ACOs’ ability to police cost, quality, and program integrity. Therefore, they rely heavily on prescribing specific standards of treatment and coordinating care among aligned providers, and encouraging use of such providers, to contain cost. One intended result may be early detection and prevention of medical conditions through primary care.
Payment Issues
Despite differing structures and roles in the health care system, MA and ACOs share some design features, as well as concerns about their payment policies.
Benchmarks
MA benchmarks are bidding targets calculated based on local TM spending, including spending by non-MA-eligible beneficiaries who do not have both Part A and Part B coverage. Counties are divided into quartiles by their levels of TM spending and assigned benchmarks of 95, 100, 107.5, and 115 percent of local TM spending. Counties with more spending get lower benchmarks and vice versa. This is intended to encourage savings in high-cost counties and greater plan participation in low-cost counties. MA base payments are the lesser of a plan’s bid or benchmark. When the bid is below the benchmark, the plan keeps a portion of the difference as a rebate it must use to reduce premiums, lower cost-sharing, or offer supplemental benefits such as dental, vision, hearing, and gym memberships.
In MSSP, benchmarks are a target for earning shared savings. CMS calculates ACOs’ benchmarks for a given agreement period using historical TM expenditure data for their assignment populations during a baseline period prior to the start of its model and blends it with average TM spending for local ACO-eligible beneficiaries. It updates benchmarks annually by blending a projected expenditure growth rate with actual national and regional growth rates, changes in enrollee severity and case mix (i.e., risk adjustment), and changes to the model’s ACO participants.8 REACH benchmarks also include a combination of historical spending data, projected and actual growth rates, and risk adjustment.9 CMS may also modify benchmarks in other ways, such as with health equity adjustments that reward ACOs with high proportions of what CMS refers to as underserved enrollees.10 ACOs generally provide core TM benefits—however, starting in 2025, CMS allows ACOs with a history of shared savings to receive prepaid shared savings, which they may use to provide “direct beneficiary services” not usually available in TM, such as cost-sharing support, beneficiary meals, or tenancy support.11
Policy researchers have criticized the benchmark methodologies for each program. For MA, setting statutory benchmarks explicitly above TM expenditures increases the program’s cost. MA benchmarks for 2024 were estimated at 108 percent of TM spending on average.12 Likewise, ACO benchmarks do not accurately estimate counterfactual spending, because they are designed to balance other goals such as achieving cost savings and ACO participation. Updating benchmarks based on actual spending trends makes continued participation less attractive for ACOs that successfully save money, because doing so would lower their benchmarks. Blending growth rates is meant to reduce this “ratchet effect,” but it may also allow some ACOs to earn higher payments without achieving greater savings.13
Risk Adjustment
Payments to both MA plans and ACOs are risk adjusted to account for enrollees’ predicted expenditures based on demographics and health status. MA and most ACO beneficiaries use the same risk model that estimates costs for specific medical diagnoses collected from medical claims across the entire Medicare population.14 CMS assigns each beneficiary a “risk score” measuring these costs relative to a normalized average of 1.0. This incentivizes MA plans and ACOs to report diagnoses more thoroughly to obtain higher payments (or, in some cases, exaggerate diagnoses), which leads to “coding intensity.” MA plans’ role as payers and access to enrollee claims across providers affords them different mechanisms for coding compared to ACOs, such as the ability to conduct chart reviews.
In MA, plans submit data on patient encounters with providers to CMS. CMS multiplies a plan’s base payment rate for each enrollee by their risk scores. To account for coding intensity relative to non-risk-adjusted TM payments, CMS reduces MA risk scores by a statutory minimum amount of 5.9 percent. (Although Congress has given CMS the ability to exceed the statutory minimum, CMS has never done so.) MA plans are also subject to audits of risk adjustment data to confirm their alignment with medical records.
CMS calculates ACO risk scores from ACOs’ medical claims. It applies risk adjustment to the expenditure data used to calculate ACOs’ benchmarks, which impacts shared savings. Capitated payments available under REACH are also risk adjusted. To account for coding intensity in MSSP and REACH, CMS limits changes in the average risk score between the final baseline year in the benchmark and the current performance year to 3 percent. In REACH, CMS also applies a retrospective coding intensity factor to reduce by 1 percent the individual-level risk scores that grow faster than average year over year.15
Quality Measurement
Quality measurement impacts MA payment levels and ACO eligibility for shared savings. As with other such programs in Medicare, these tend to increase administrative burden without improving quality of care, including by reliance on ineffective metrics.16
CMS assigns an MA plan a star rating between one and five (in half-star increments) based on about 30 metrics for clinical processes, health outcomes, and customer experience. It modifies star by other factors related to plans’ performance on the quality metrics, such as improvement over time and performance disparities between plans for enrollees considered underserved.17 Plans with four or more stars receive “quality bonuses” through higher benchmarks and keep larger shares of bid-benchmark differences as rebates. Quality bonuses cost roughly $11.8 billion in 2024—however, regulatory changes in the star calculation methodology will likely lead to a decrease in 2025.18
ACOs in MSSP must meet a minimum quality performance threshold to receive shared savings, and scoring higher reduces their downside risk. CMS also applies a health equity adjustment to an ACO’s performance results based on the share of its assigned beneficiaries that is considered underserved.19 In 2025, there are about six MSSP measures: consumer surveys, unplanned hospital readmissions, breast cancer and depression screenings, and management of diabetes and high blood pressure.20 In REACH, rather than basing eligibility for shared savings on quality performance, CMS withholds 2 percent of benchmarks for ACOs to earn back. REACH uses measures for unplanned readmissions, consumer surveys, and patients with chronic conditions (i.e., unplanned admissions, days at home, and follow-up care).21
Selection Biases
Enrollment in MA and ACO participation in a model are both voluntary and therefore non-random. This introduces potential selection biases into their payment methodologies that raises the question of whether risk adjustment policies are effective.
MA benchmarks are based on average TM costs. However, the Medicare Payment Advisory Commission (MedPAC) suggests that individuals who enroll in MA have below-average spending, even when accounting for risk scores. This would mean MA plans have systemically lower-than-expected costs partly due to their enrollee population rather than cost-saving efforts. This likely occurs due to beneficiary preferences (e.g., prioritizing cost-sharing protections in MA over a more expansive provider network in TM). MedPAC estimates that “favorable selection” led to excessive MA payments of roughly 9 percent in 2024.22
Favorable selection may also impact providers’ decisions to form ACOs. Although new participants may receive more flexible contract terms, such as a gradual transition to downside risk, providers that do not expect to benefit from a model may exit or refrain from joining at all. The result is that providers that already have low costs may be more likely to join ACO models than high-cost ones with more savings potential, because the former may find it easier to earn shared savings than the latter.23 MedPAC has found that ACOs can influence beneficiary assignment through annual wellness visits, leading to “unwarranted” shared savings in MSSP.24 In recent years, CMS has attempted to attract ACOs serving high-needs populations with health equity adjustments in benchmarks and quality measurement, as well as by offering advance payments in certain cases.
Cost Savings
A key component of the value proposition for MA and ACOs is the potential for cost savings relative to fee-for-service TM. As evidenced by the critiques of MA and ACO payment methodologies, there are opportunities to increase their cost effectiveness.
MA plans usually provide core Part A and B benefits more cheaply than TM. MA plan bids, which represent their estimated cost to cover those benefits, were roughly 80 percent of TM costs in 2024. Yet various payment policies—county benchmark quartiles, rebates, quality bonuses, and risk adjustment—leave plan payments roughly equal to per-enrollee TM spending. Other factors, such as coding intensity and favorable selection, lead some researchers to conclude that MA is costlier to taxpayers than TM on a per-beneficiary apples-to-apples basis. Various studies estimate MA payments relative to TM spending, but offer differing estimates of factors such as comparable enrollee populations or how MA growth can reduce costs in TM through “spillover” effects.25
Each year, CMS touts significant Medicare savings within MSSP—for 2023, it announced $2.1 billion net of shared savings payments.26 Since 2013, these net savings sum to over $10 billion.27 However, the Congressional Budget Office (CBO) suggests that there are cases where ACOs only appear to yield savings when measured against their benchmarks, but may actually generate net costs to Medicare. In a review of more rigorous studies that compare ACO spending with a counterfactual estimate of costs for beneficiaries if they had not joined an ACO, CBO found “mixed results,” with two ACO models achieving small net savings and three with small net costs. The growth of alternative payment models has made such evaluations rarer and more difficult to conduct in recent years. CBO concludes that ACOs are associated with “small net budgetary savings”—particularly in the early years of these models and for ACOs that are physician-led, more primary care-heavy, and take on more downside risk. As with MA, ACO savings figures depend on whether they account for selection biases or spillover effects.28 One recent report (a correction of a previous version), using MSSP data from two prior studies, found that it was associated with higher MA benchmarks and net losses to TM and CMS overall.29
Only a few studies compare the spending impacts of MA and ACOs directly against each other. One study examined the spillover effects of both programs and found that greater MA and ACO enrollment in the same geographic area led to higher combined spillover effects in terms of reducing post-acute care expenditures.30 However, one study of a health system in the southern United States found that MSSP had higher total health care spending (by 22-26 percent) and utilization than MA had after controlling for demographic and clinical risk factors.31
MA, ACOs, and Medicare’s Future
Overall, MA and ACOs show different approaches to transforming Medicare from a fee-for-service health program, but there are also areas for improvement. Each has opportunities for improved payment policies: more credible and savings-focused benchmarks, less gameable risk adjustment, and less burdensome quality measurement that is more meaningful for patients. Some problems, such as selection biases, demand different approaches, since enrollment is based on choice in MA and on attribution for ACOs.
Others, such as coding intensity, have been studied more thoroughly in MA than in ACOs (although recent research finds that under-coding in TM contributes to about 22 percent of MA coding intensity.32
Third-party estimates of MA or ACO savings tend to show mixed results. Although one direct comparison suggests that MA has more potential for reducing health care costs, some argue that ACOs save more money despite more favorable policy treatment towards MA.33 One proposed pathway is to have ACOs bid to provide Part A and B benefits, allowing beneficiaries to choose ACO-managed TM or MA on the basis of premiums.34 Given that beneficiary choice helps MA to balance incentives for cost containment and accountability for the value of coverage, adopting a similar model across Medicare would allow for MA plans and ACOs to directly compete.
Even absent this, more modest changes can improve each program. MA’s growth offers a pathway to transforming Medicare as a whole and ACOs can pass
more financial risk onto health care providers in TM.
