The Evolution of Medicare: Challenges, Responses, and Prospects Free

Abstract

The Medicare program has provided a near-universal source of health care coverage for America's elderly since 1965. Over its 60-year history, the program has evolved to cover a greater share of the population and to pay for an increasing share of the nation's health care bills. As Medicare has grown, so too have its challenges. The traditional Medicare program has failed to keep pace with a rapidly changing health care sector and demographic shifts. Constrained by its own benefit design, Medicare has allowed privately contracted health plans (Medigap, Medicare Advantage) to provide much needed yet inadequate remedies to the program's shortcomings. After briefly recounting Medicare's origins, we discuss how the program's founding statutes have hindered its ability to respond to new and growing challenges along the dimensions of cost sharing, cost containment, and benefit design. We then propose a three-pronged approach to reforming Medicare's benefit structure. We argue that a simplified enrollment process, a single benefit that brings together the program's constituent parts (Part A, Part B, and Part D), and a new organizational structure for care delivery based on the program's experience with Accountable Care Organizations will together create a robust foundation that can sustain the Medicare program into the future.

The Medicare program is 60 years old. It is popularly beloved and politically strong, but most analysts examining it believe it is due for a refresh. From a small program that covered just 10% of the population and paid 10% of health care bills in 1968, Medicare has expanded in size, complexity, and influence and become the dominant force in the US health system (CMS 2014; Gornick 1976). It provides vital financial protection and access to care to nearly one in five Americans. It pays 26% of all hospital costs, 26% of outpatient costs, and 32% of pharmacy costs in the United States (CMS 2023b). At the same time, Medicare faces serious challenges. Trust fund solvency is threatened by the aging of baby boomers and continued growth in health care costs relative to inflation, which also intensify federal budget impacts. The Hospital Insurance Trust Fund is currently expected to become insolvent by 2036 (Board of Trustees 2024). Medicare is forecast to constitute 18% of the federal budget and 3.9% of gross domestic product (GDP) by 2032 (Cubanski and Neuman 2023).

There is limited scope for further cost containment within the traditional Medicare (TM) program. Fee-for-service payment rates for doctors and per-case prospective payments based on diagnosis-related groups (DRGs) for hospitals are already well below commercial levels (Cohen, Maeda, and Pelech 2022). Placing additional limits on payment increases might constrain the set of physicians and hospitals who will be willing to provide services to beneficiaries. The lack of political appetite to reduce Medicare payment growth rates further is well documented in the congressional record. For example, Congress overrode all but one scheduled cut to physician payments under the Sustainable Growth Rate system until it was repealed altogether in 2015 (Hahn 2015; Steinbrook 2015). Similarly, Congress backtracked on the substantial payment reductions under the 1997 Balanced Budget Act (O'Sullivan et al. 1999). Other cost-containment efforts not directly focused on Medicare prices, including value-based payment reforms included in the Affordable Care Act (ACA), have produced only modest net savings (Lewis et al. 2022). The Inflation Reduction Act has opened the door to drug price negotiation, but the scope of additional savings in the pharmacy program is dwarfed by the overall cost of the program.

By contrast, the growth of Medicare Advantage (MA), the program component that allows beneficiaries to opt for coverage through a private insurance plan paid a set amount per enrollee by the federal government, makes it an important target for cost containment. MA has ballooned as a share of Medicare over the past two decades. Program members now account for 54% of the Medicare-eligible population, and payments to MA plans account for 54% of the Medicare budget (Freed et al. 2024a). Despite decades of attempted fixes, most analysts believe that plans continue to be overpaid and that they continue to select healthier-than-average patients, driving up overall Medicare spending. Yet, while MA plans are overpaid, considerable research suggests that they control the use of medical services more effectively than TM through reduced rates of hospital and emergency department visits and shorter hospital and postacute facility stays (Agarwal et al. 2021).

The positives and negatives of MA have led to two very different sets of policy proposals for addressing Medicare. Progressive analysts would prefer to, in their words, level the playing field. That is, they want to shrink the overpayments to MA plans so plans have fewer resources with which to attract beneficiaries. To do so, analysts have proposed replacing MA's risk scoring system, addressing inflated payment benchmarks, and reforming administered payments or replacing them with competitive bidding (Gilfillan and Berwick 2021b, 2023; Lieberman, Ginsburg, and Valdez 2023; Lieberman, Ginsburg, and Lin 2023). They would also like to improve TM by adding the extra benefits that MA plans currently use to entice beneficiaries, particularly dental, hearing, and vision care as well as catastrophic coverage for Part A and Part B services (Freed et al. 2024b). More recently, former Vice President Kamala Harris suggested adding a home care benefit to Medicare that would provide long-term services and supports.

Conservative analysts would prefer to promote MA and move toward a system of premium support, in which MA, not TM, would be the default beneficiary option in the Medicare program. Rather than the current process under which MA payment rates are tied to TM-based benchmarks, government payments to MA plans would increase according to a legislated formula (for example, tied to GDP growth). Plans would compete for customers under regulations governing covered benefits (MedPAC 2017).

Yet seeking to save Medicare by either shrinking MA or promoting its continued growth carries with it a variety of complexities that will make that job difficult. On the shrinking side, for many beneficiaries MA has become an essential element of accessing care within the Medicare system. Among MA beneficiaries in mainstream plans (as opposed to Special Needs Plans or preferred provider organizations), 97% have prescription drug (MA-PD) coverage. Three quarters of these enrollees pay no premium above the Part B premium, and the average MA-PD premium is just $14 per month. Meanwhile, most TM beneficiaries (89%) have some form of supplemental insurance that covers additional services (hearing, vision, dental) but requires cost sharing that, for many retirees, is cost-prohibitive for accessing care. Those with supplemental insurance are typically covered either through an employer (employee benefit or retiree plan) (32%) or individually purchased Medigap plans (41%) (Freed et al. 2024b; Ochieng et al. 2023). Lower-income people without retiree coverage in TM have few options for addressing the program's prohibitively high cost sharing, such as the Part A deductible for 2024 of $1,632 for 60 days followed by copayments of $408 per day after 60 days, the 20% cost sharing for physician services, and the absence of stop-loss protection against catastrophic expenditures (CMS 2023c). Private Medigap coverage has very high premiums and, in most places, limits enrollment for beneficiaries with preexisting conditions who do not enroll within their first six months of Part B coverage (Boccuti et al. 2018). The prevalence of employer-sponsored supplemental insurance has been declining over time (Neuman and Damico 2024). The various Medicaid-administered low-income subsidy programs provide cost-sharing assistance for beneficiaries with incomes just above the Medicaid threshold (as well as Medicaid-eligible individuals), but these programs are administratively burdensome and have had relatively low take-up as a result of strict assets tests, complex administrative processes run by state Medicaid programs, and limited availability of information about the programs (Erzouki 2024; Primus and Bingham Rich 2024).

On the expanding side, while Medicare overpayments are a growing target for cost control, they are not the only challenges posed by the growth of MA. The strategies MA plans use to attract the most profitable patients pose risks for the quality of care available to the sickest and most complex patients. The added benefits that MA plans use to attract patients, such as limited front-end coverage of vision, dental, and hearing services, pump up public costs while generating very modest gains in beneficiary financial protection. The strategies plans use to contain spending, such as aggressive utilization management including prior authorization, concurrent reviews of treatment regimens, and retrospective reviews of claims, raise administrative costs across the system and place undue burdens on providers. These are mechanisms that can reduce inappropriate utilization by enlisting clinical judgement in rationing, but they can also be applied haphazardly to increase hassle factors that discourage beneficiaries from obtaining care and reduce utilization in an arbitrary fashion. Importantly, the ability of MA plans to contain spending depends critically on the availability of publicly regulated TM payment rates for hospitals and physicians (Baranoff et al. 2021; MedPAC 2021; Ramsay and Jacobson 2024). Finally, moving to an MA-dominated model and tying MA payment rates to a metric outside the health care system would reduce the Medicare program's ability to ensure that seniors continue to have access to new, costly, and effective health care technologies (Glied 2018).

To chart a path forward, it is necessary to recognize that MA's strengths and weaknesses are a direct consequence of the program's evolution as a graft onto the original structure of Medicare. TM's fundamental benefit structure remains largely a vestige of the 1960s, as it has sluggishly moved toward a modern infrastructure for cost containment beyond price regulation, sensible protection against catastrophic costs, and an adequate scope of benefits. Because MA grew in the shadow of TM, its features are not optimal either; rather, they are conditioned by TM. As we illustrate below across the dimensions of cost sharing, cost containment, and benefit design, the principles and provisions baked into Medicare's original structure have impeded the program's ability to respond to changes in population needs and in the health care environment, creating an opening for alternative plans that are themselves not optimally designed. Preparing Medicare for the next 60 years will require revising the program's fundamental structure.

I. Traditional Medicare

At its inception, Medicare was deliberately designed to rock as few boats as possible. Its design was conservative: it simply layered the most prevalent private-sector health insurance policy design over the existing Social Security structure. The hospital component, modeled on the standard Blue Cross policy, covered inpatient hospital stays (limited to 90 days per stay), postacute facility services, home care (limited to 100 days per episode), and outpatient diagnostic services. It featured a separate deductible for each hospitalization episode, similar to the design of an automobile or home insurance policy, as well as per-day cost sharing. The Supplementary Medical Insurance program, modeled on the Federal Employees Health Benefits (FEHB) major medical program, covered medical and other health services and limited home health services. The split between hospital and supplemental medical insurance mirrored the Blue Cross/Blue Shield design (Ball 1995; Oberlander 2003: 31).

Medicare's administrative design was also deliberately restrained. The first three provisions of the Medicare statute, Title XVIII, prohibit the federal government from interfering in the practice of medicine, from limiting beneficiary choice of provider, and from constraining the sale of additional private insurance to beneficiaries. While many of its proponents at the time had more ambitious goals—namely, national health insurance—the early Medicare program sought merely to plug a hole in an existing health insurance and health care financing ecosystem.

Over time, TM has had bursts of innovation in adopting new models of provider payment, such as DRGs, but it has been much slower to change in other important dimensions. Rather than changing the program as the environment changed, policymakers have often opted to address bigger challenges by opening new offshoots of Medicare—low-income subsidy programs tied to Medicaid, pharmaceutical benefits through a new Part D, and, most importantly, Medicare Part C (now MA)—rather than dismantling and reconfiguring its initial structure.

II. Cost Sharing

Medicare's initial proponents saw it as an enhancement to the existing Social Security system, which provides a progressive cash benefit (Ball 1995). But health insurance is fundamentally different from a cash benefit program. Cash layers over everything else; more cash is always better. Health insurance is peskier. Rather than layering over the status quo, the new health insurance benefit intersected with the existing environment in several ways. First, about 54% of seniors newly covered by Medicare already had some form of hospital insurance (although only 25% had substantial coverage through Blue Cross), and about 46% had at least nominal surgical insurance (Cohen 2009; Finkelstein 2007). Many obtained those benefits through retiree programs provided by their employers or unions that they had “paid for” through lower wages while they were working (Follmann Jr. 1965). Adding a new universal benefit, especially one with premiums, would not have been acceptable if it had made them worse off.

The provision that prohibited the government from constraining the sale of private insurance solved this problem (see section 1882 of the Social Security Act). Although Medicare's design incorporated the high cost sharing that was the primary strategy for cost containment in private insurance of the time, that design did not reflect the coverage facing those newly enrolled—at least not for the majority who had held private coverage before Medicare began. That group could use their existing coverage to fill in the gaps in their new Medicare coverage (Gornick et al. 1985). Under TM's indemnity insurance model, patients could seek care from any provider, have Medicare pay its share of expenses, and use their private coverage to pay any remaining cost sharing.

This design feature meant that Medicare's introduction was a boon to seniors with private insurance (and a windfall to the firms that paid for much of it). Per capita inflation-adjusted private insurance spending fell by 3.6% between 1965 and 1967 as Medicare phased in; private insurance-paid hospital spending fell by 7% (CMS 2023a). Retiree health insurance plans quickly adapted. Rather than providing catastrophic financial protection, they offered coverage specifically designed to fill in the cost sharing of the new Medicare program. During 1966–1975, private insurance covered some or all of the cost-sharing amount for half of all beneficiaries, and less than one third were paying for the full cost-sharing amount out of pocket (Kittner 1968, 1970). The availability of supplemental front-end coverage meant that Medicare was a particularly valuable program for those higher-income seniors who had previously held retiree coverage. They now faced much lower effective cost sharing because they could layer their existing coverage over Medicare.

The introduction of Medicare also spawned a new individual market for Medigap policies, which typically paid for out-of-pocket costs stemming from the TM benefit design, prescription drugs in many cases, and the coverage of catastrophic costs. Those products were aimed at those without retiree benefits who could afford to pay for premiums for supplemental coverage. From a beneficiary perspective, Medigap coverage is an excellent deal. Much of the additional use induced by eliminating Medicare cost sharing falls onto Medicare, so a Medigap premium buys much more utilization than an equivalent premium for a comprehensive insurance policy. Estimates suggest that seniors with Medigap coverage use 22%–24% more Medicare-paid services than do those without this added coverage (Cabral and Mahoney 2019). The introduction of Medicare provided coverage to lower-income seniors without either retiree benefits or the means to buy Medigap-provided protection against catastrophic financial risk, but their access to services was constrained by the program's cost sharing. In stark contrast to Social Security, Medicare's initial cost-sharing design, when combined with the availability of retiree benefits and Medigap, was quite regressive. High-income people with Medigap have very low out-of-pocket costs and make more use of Medicare, while middle-income people (before MA) faced high out-of-pocket costs and made less use of taxpayer-funded Medicare.

Meanwhile, changes in health care technologies, information technologies, and labor market dynamics were transforming the structure of health insurance for those younger than 65. As new health care technologies came online, the overall cost of medical care increased, and the concentration of medical expenses increased even further. The combination of these two trends generated a much greater demand for health insurance. In 1963, a person in the top 1% of health spenders could expect to spend an amount equivalent to about 120% of median annual income. By 1977, a person in the top 1% of health spenders could expect to spend an amount equivalent to 280% of annual income (Berk, Monheit, and Hagan 1988).¹ New technologies also changed the composition of spending, weakening the delineation of care between hospitals and physicians’ offices and between surgical interventions and pharmaceutical interventions. Meanwhile, improvements in computer technologies made it much easier to manage complex benefit designs and monitor physician behavior.

These changes fostered shifts in private insurance design. The distinction between hospital and medical insurance, reflected in the original separation between Blue Cross and Blue Shield then enshrined in Medicare, began to disappear by the 1980s, replaced by comprehensive managed care policies that often included pharmaceutical benefits. In place of the per-episode hospital deductible, plans incorporated a single deductible that spanned all services across the year and an out-of-pocket maximum that capped total spending, addressing the growing burden faced by those who had catastrophic illnesses. While Medicare-insured seniors with supplemental coverage faced less cost sharing than most privately insured workers, seniors without supplemental coverage were now much less protected than their privately insured counterparts. Their coverage had no out-of-pocket maximum, and if they were very unlucky and had multiple (five or more) hospitalizations in a year, they might incur expenses of as much as $2,600 associated with hospital deductibles alone (based on the 1987 inpatient hospital deductible) (SSA 2011).

This situation provided the impetus for the Medicare Catastrophic Coverage Act (MCCA) debacle of 1988. Congress and the Reagan administration sought to provide financial protection against pharmaceutical costs by adding catastrophic drug insurance to Part B, to put an out-of-pocket maximum on Part B cost sharing, and to reform the cost-sharing structure of Part A, including eliminating inpatient hospital copayments above the deductible. Most of the cost (63%) was to be paid by Medicare beneficiaries through progressive, income-based supplemental premiums, but Part B premiums for all enrollees were to increase by $4 in 1989 to finance the remaining 37% of the costs of the catastrophic coverage policies. In effect, the MCCA would have made about 22% of seniors (mainly those without Medicaid, Medigap, or retiree insurance) better off, but it would have made about 30% of seniors (those with retiree coverage) worse off (Christensen and Kasten 1988; Morrisey, Jensen, and Henderlite 1990). Moreover, the MCCA's design required that Medicare beneficiaries pay the entire cost of its benefit expansion. These arrangements resulted in an outpouring of anger from Medicare beneficiaries that proved to be politically unsustainable. Amid rising controversy and a backlash that also reflected beneficiaries’ confusion about the act's financing provisions, Congress repealed the catastrophic coverage provisions in 1989, only 16 months after their passage (Himelfarb 1995; Oberlander 2003: 66–69).

Meanwhile, private health insurance was undergoing a period of tremendous change. As care continued to move out of hospitals and hospital occupancy fell (spurred by Medicare and private payment changes), private insurers began to negotiate contracts with hospitals, promising a flow of patients in exchange for lower payment rates. The managed care revolution that ensued fundamentally changed how health insurance worked, in direct contravention of the principles in TM (Oberlander 2003: 167–69). Private insurers deliberately intervened in the practice of medicine through utilization review, and they constrained consumer choice of provider (Glied 2000). While they did not necessarily prohibit duplicate coverage, limited provider networks and changes in employer contribution rules that made it costly to cover an already insured spouse made such coverage increasingly rare. Insurers could and did use cost sharing to incentivize patient choice of service type or provider. Additional improvements in information technology made all this monitoring and management a more effective (and potentially fairer) way to constrain utilization than the blunt cost-sharing design of TM.

Medicare had allowed health maintenance organizations (HMOs) to participate in the program from the start, although initially they were required to bill as if providing services on a fee-for-service reasonable-cost basis. Beginning in 1972, plans could be paid using capitated premiums; a capitated payment system arrived in 1982 under the Tax Equity and Fiscal Responsibility Act. But most managed care enrollees were people who had previously held HMO coverage, usually through a staff model plan (McGuire, Newhouse, and Sinaiko 2011; Patel and Guterman 2017). Few new beneficiaries coming from existing indemnity plans were interested in constraining their choice of provider to a managed care network. The growth of managed care in the younger-than-65 population changed the dynamic. Medicare managed care enrollment grew sixfold during the 1990s, slowed with a change in payment rates in 1999, and then began accelerating rapidly in 2003 (Zarabozo 2000). Medicare payment rates (and the additional benefits plans passed on to enrollees) drove much of the change, but structural factors also played a large role. On the demand side, beneficiaries who had become accustomed to managed care were also accustomed to narrower network breadth. On the supply side, managed care utilization strategies allowed companies to sell MA plans that insulated people from the antiquated TM cost-sharing structure at far lower prices than existing Medicare supplemental plans could manage. While beneficiaries who were set on having access to the full range of Medicare providers (typically beneficiaries who were richer and sicker than average) might be willing to pay more for a conventional supplemental plan, lower-income beneficiaries with fewer ties to existing providers were willing to make the switch to MA (Brennan 2024: 138).

Under MA's cost-sharing structure today, plans may adopt their own cost-sharing rules, but they are required to have an out-of-pocket maximum of no more than $8,850 for in-network care under Parts A and B, and an additional $2,000 out-of-pocket limit for care under Part D beginning in 2025. HMO plans generally have much lower out-of-pocket maximums (and generally low cost sharing). The average among non-HMO plans is close to the legislative maximum, although less than half of all MA beneficiaries are enrolled in a non-HMO plan (Freed et al. 2024b). That is certainly better than the open-ended exposure under TM, but even for middle-income seniors, it is very high relative to their incomes, and it is much higher than the typical out-of-pocket maximums in employer plans (which combine hospital, physician, and pharmaceutical spending) (Claxton et al. 2023). This cost-sharing structure generates two problems. First, high out-of-pocket maximums affect the small share of beneficiaries with the highest costs. Despite their much higher premiums, Medigap supplemental plans will provide better financial protection than MA for beneficiaries with serious health conditions who anticipate such costs. Designs with low premiums combined with high out-of-pocket maximum limits contribute to favorable selection among MA plans.

Second, because MA plans are an outgrowth of TM plans, they do not incorporate progressive benefit designs. Unlike the arrangements under the ACA Marketplace, and similar to standards in employer-sponsored insurance, deductibles, coinsurance rates, and out-of-pocket exposures under MA do not vary with income. To get that protection, beneficiaries need to participate in a separate low-income subsidy program (Dual-Eligible Special Needs Plans). While Medicare's financing is progressive and has become more so over time with the addition of income-related premiums in Parts B and D, the removal of the cap on taxable wages, and the taxation of investment income, benefits in the TM program, when combined with the availability of supplemental coverage, remain regressive.²

III. Cost Containment

The Medicare provision that permitted supplemental insurance has made redesigning TM to modernize cost sharing particularly difficult. The provisions that prohibit Medicare from intervening in the practice of medicine and allow beneficiaries free choice of provider have similarly made it difficult for the program to contain costs in some ways. Here too, MA plans have filled a gap, but not always in ways they might have had they evolved in the absence of TM.

Originally, Medicare's small market share and the restraints on design meant that unlike universal single-payer systems, Medicare had limited control over the aggregate supply of medical services or the diffusion of technologies and, initially, prices were established by providers (Oberlander 2003: 33). Medicare and private insurance payment rates were not far apart through the mid-1980s. Research (famously, the Dartmouth Atlas) consistently documented that Medicare costs were highest in areas where provider competition was most intense (Cromwell and Burstein 1985; Feldman and Dowd 1986). In these markets, providers engaged in a medical care arms race, adding services and amenities to attract patients. While Medicare tried to address inappropriate utilization through a network of peer review organizations, these ultimately proved to be largely toothless (Oberlander 2003: 116–20). Lack of control over supply combined with limits on demand-side cost sharing led to predictable escalation in costs.

Medicare responded to cost escalation and rising federal budget deficits in the 1980s and 1990s by reforming the methods it used to pay providers, shifting from provider-determined rates to administratively established rates (a prospective payment system based on DRGs for hospital care, and a national fee schedule based on the Resource-Based Relative Value Scale for physician services) (Oberlander 2003: 120–29). Where possible, as in hospital payment, payment shifted to broader bundles of services to limit providers’ ability to manipulate volume in response to changes in price. Private insurers adopted many, but not all, of these payment innovations.

All that changed with the rise of managed care during the 1980s and 1990s. Initially, private insurers found that in competitive markets with substantial unused capacity they could undercut regulated rates by bargaining with providers. In addition, they could use utilization management and network design to keep costs down. While Medicare spending varied positively with the extent of competition, researchers found that private insurance spending fell as market competitiveness increased (Cohen, Maeda, and Pelech 2022).

Over time, the cost-containment strategies of managed care plans and TM diverged. Managed care plans, buoyed by information technology, relied ever more heavily on utilization review. A new arms race developed within MA, centered on coding and documentation strategies rather than the introduction of new technologies. That served to add administrative complexity and cost across the system as new “systems” and coding staff were introduced. TM, which had very limited utilization management abilities, could only respond by constraining growth in prices. In the 2000s, the divergence in prices accelerated. On average today, private insurers pay 29% more than TM for physician services and more than twice as much for hospital services (Cohen, Maeda, and Pelech 2022). Notably, however, private (non-MA) prices vary within and across markets much more than TM prices do. “Must have” providers preferred by privately insured patients can command rates 4 times as high as TM, while non-MA private insurers pay rates below TM to hospitals disfavored by privately insured customers (White and Whaley 2021). Analyses in the late 2000s comparing variations in spending across the country among the Medicare population and the privately insured population found very different patterns in the two groups. In competitive markets, TM's lack of utilization control drove higher spending. Private insurers, by contrast, could contain utilization and bargain hard for lower prices in those markets. In concentrated markets, TM benefited from lower utilization, but private insurers suffered from high prices (Newhouse, Garber, and Graham 2013).

MA plans have the best of both worlds. Like private managed care plans, they control utilization much more effectively than TM does. They are particularly effective in reducing hospital and post acute service use and in making substitutions among sites of care. But they can also do much better than private insurers because they can take advantage of TM's regulatory rates (Luft 1998; Trish et al. 2017). Currently, hospitals and physicians, even in concentrated markets, are willing to accept TM rates from MA plans, giving MA plans a significant advantage over other products offered by the same private insurers. Because the differential between Medicare and private rates is so large, particularly in concentrated markets, MA plans could not remain competitive with TM if they had to pay private insurer rates, despite their success at utilization control (Pelech and Nelson 2017).

Analysts have puzzled over why providers have been willing to accept TM prices from MA plans, especially since MA plan utilization management and claims denial practices place much more significant burdens on providers than TM plans do. One reason may be section 1866 of Medicare and Centers for Medicare and Medicaid Services (CMS) rule 42 CFR 422.214, which sets out-of-network payment for services to MA plans at TM rates. In other words, if providers do not accept MA prices that are close to but typically higher than TM, their out-of-network status would otherwise leave them with lower TM rates (Berenson et al. 2015). Another explanation may be that competition from TM leads MA plans to negotiate lower rates from providers (McWilliams 2024). Either explanation, however, suggests that as MA plans come to dominate the market their ability to rely on TM-administered payment rates will erode because the differential between commercial market rates and TM will lead providers to balk at deciding to supply MA plans. This inability to maintain administered prices has led the Congressional Budget Office to question the extent of potential savings from an all-MA program (Pelech and Nelson 2017).

Analysts have long bemoaned TM's inability to engage in utilization-focused cost containment (Lave 1984). Finally, under the ACA, Congress somewhat successfully added new cost-saving methods of utilization management onto the program's existing structure. Accountable Care Organizations (ACOs) are groups of providers that voluntarily coordinate care for a defined population of TM beneficiaries. If the providers keep costs below a financial benchmark and adhere to quality performance standards, they can share in Medicare's savings. The expectation is that this incentive will encourage providers within an ACO to avoid overproviding services while maintaining quality of care. ACOs began in one-sided risk programs and were then shifted toward two-sided risk agreements, where they bear a portion of excess costs if spending exceeds a benchmark (Spivack, Murray, and Lewis 2023). The upside potential of savings and the selection incentives in ACO contracts are both limited. Participating providers are paid fee for service at TM rates, and beneficiaries are not limited to providers within their ACO network (so the program remains in compliance with the original statute). The historical financial benchmarks set for ACOs entering the program are based on risk-adjusted average per capita Part A and Part B expenditures for original fee-for-service beneficiaries (CMS 2017; McWilliams et al. 2016). Evidence to date suggests that ACOs with two-sided risk contracts have delivered modest savings and have maintained or improved quality of care (Gonzalez-Smith et al. 2019). Savings have been greatest among physician-led ACOs, which have the greatest opportunity to maximize fee-for-service payment to ACO members while reducing costs by cutting back on hospital and postacute service use.

IV. Benefits, Networks, Formularies, and Selection

MA's ability to combine administered pricing and managed care utilization control give it an inherent competitive advantage over TM (Glied 2000). It has expanded on that advantage by attracting healthier-than-average beneficiaries and exploiting loopholes in the risk-adjustment system designed to mitigate those selection effects (Garrett 2024).

The staff-model HMOs that participated in Medicare at its inception, such as Kaiser Permanente, required members to use a defined set of providers, and they offered generous front-end benefits, including preventive services. The combination of closed networks and front-end benefits made HMOs attractive to young families—and to relatively healthy Medicare beneficiaries who had not already established relationships with particular providers. Early studies of selection into Medicare HMOs found that HMO members were healthier than their counterparts in TM based on preenrollment utilization, postenrollment mortality, self-reported health status, and the number of difficulties with activities of daily living/instrumental activities of daily living, with HMOs particularly likely to attract those who self-reported their health as excellent (Mello et al. 2003).

By 1972, the selection incentives prompted the introduction of risk adjustment into the Medicare program, in conjunction with the expansion of MA plan options. Risk-adjustment programs reduced payments to plans with healthier-than-average enrollees and increased payment to those with sicker-than-average enrollees. Medicare adjusted its risk-adjustment program in the face of continued selection several times (Patel and Guterman 2017). Capitated plans, in turn, modified their practices to take best advantage of the risk-adjustment systems in place. Currently, Medicare risk adjustment is based on recorded diagnoses, and plans invest in identifying and coding diagnoses that make their enrollees appear sicker relative to how they would be classified in TM (Gilfillan and Berwick 2021a; Jacobs and Kronick 2021).

Risk-adjustment mechanisms can be modified to avoid incentives for upcoding, but the potential for selection across types of plans remains substantial. One way to attract healthier people is through benefit design. Here the challenge is that differences in benefit design, whatever their intention, can generate incentives for selection. This has become an increasing challenge as private-sector benefit design has changed over time.

Medicare's initial benefit design matched the design of commercial insurance plans of the time. Like the standard Blue Cross plan, it included hospital and postacute institutional and home care. Like the FEHB major medical plan, it included physician services in hospitals and offices and at home. Medicare considered but did not incorporate a pharmaceutical benefit (Oliver, Lee, and Lipton 2004). Medicare also did not include preventive services, which were not typically covered by private insurance at the time because they are predictable and not very costly (Schauffler and Rodriguez 1993).

Prevention was always a selling point for HMOs, which had grown out of company-provided health insurance plans where there was little opportunity or incentive to engage in selection (Morrison and Luft 1990). But as managed care in its various forms grew in the 1990s, benefit design played an increasing role in contributing to selection. Preventive services might, in principle, lead to reduced overall health spending. They were also valuable in attracting plan members who did not anticipate needing acute care services at all (Greene, Blustein, and Laflamme 2001). Incentives to include preventive services increased further as they became an important component of plan quality measurement (Corrigan and Nielsen 1993). By 1989, nearly all HMO plans covered physical exams and preventive diagnostic care. Under pressure to compete with their HMO counterparts, traditional insurance carriers and Blue Cross/Blue Shield plans began covering preventive services as well (Gabel et al. 1990; Fox and Kongstvedt 2013: 9).

Traditional Medicare could not simply add preventive services. The only way to cover preventive care under Medicare's original formulation was to argue that it saved the program money. In 1979, analysts concluded that coverage of the pneumococcal vaccine would have that effect, and in 1980 it became the first preventive service covered under TM (OTA 1984). Unfortunately, few other preventive services, whatever their benefits for health, could be shown to save money for the Medicare program, and that avenue for adding prevention proved broadly unsuccessful (Russell 1984). Instead, Congress added specific preventive service benefits to Medicare from time to time, concluding with the ACA mandate that eliminated cost sharing for high-value preventive services and vaccinations (CMS 2010).

The interplay between plan design and selection became increasingly complicated as the scope of covered benefits expanded and information technology improved. During the 1960s and 1970s, private insurers and the Medigap supplemental policies that grew up around Medicare gradually began to add pharmaceutical coverage to their plans. By the mid-1980s, the high prevalence of drug coverage among retirees with supplemental insurance helped torpedo the Catastrophic Coverage Act, which would have added a pharmaceutical benefit to Medicare Part B, financed by current beneficiaries (Oliver, Lee, and Lipton 2004). Ultimately, passage of the Medicare Modernization Act (MMA) extended pharmaceutical coverage to Medicare benefits. However, unlike the 1988 proposal, pharmaceutical benefits were assigned to a new, voluntary program, Medicare Part D, that would be administered by a distinct set of private insurers and would have its own separate benefit design, requiring plans to cover specific classes of drugs but allowing them to devise their own formularies within those parameters (Megellas 2006). Part D could be combined with an MA plan, and today, more than half (57%) of Part D recipients are in MA plans (Cubanski and Damico 2024).

Medicare Advantage plans, particularly those that also include Part D benefits, now have multiple additional dimensions through which they can attract—or deter—potential enrollees. They can choose which specialists to include in their networks, which drugs to include in their formularies, and which supplemental benefits to include. The evidence suggests they use all of these mechanisms to deter people who are likely to be heavy utilizers (beyond the utilization addressed in the current risk-adjustment system) from joining (Garrett 2024).

Plans also engage in practices to attract the healthiest beneficiaries. They must use a share of rebate dollars (the difference between plan bids and the payment benchmark) for the direct welfare of beneficiaries by enhancing benefits, reducing cost sharing, or lowering premiums. Most MA plans use part of the rebate to offer front-end dental, vision, and hearing benefits, which cover initial or prophylactic visits but rarely include coverage for costly restorative procedures, and nearly all offer gym memberships or other fitness benefits (Fuglesten Biniek et al. 2022). Like the preventive services offered by HMOs, these front-end benefits offer little insurance value. Instead, healthy customers—the sort who use gyms and get their teeth cleaned and their glasses replaced—are attracted by benefits that are nearly fungible to cash.

Selection corrupts MA in many ways. Most obviously, it leads to gaming of the risk-adjustment system through upcoding—the practice of adding codes to clinical cases, even if they are not directly related to the condition being treated, to obtain an upward payment adjustment—and other strategies. These gaming strategies add administrative costs that generate no actual health or financial protection benefits. More insidiously, selection distorts benefit design toward front-end services whose costs can be anticipated. Efforts to reduce MA growth by adding the same benefits to TM cuts into the available public funding for services where financial risks are significant, such as long-term services and supports. Finally, selection through formulary or network design may leave the most vulnerable Medicare beneficiaries, those with rare or complex conditions, with severely limited treatment options for the care that is likely to be most useful to them. In almost all states, preexisting condition exclusions can continue to be applied to those who purchase Medigap coverage after the initial enrollment period, so these beneficiaries may be left choosing between a TM-only policy with a broad network but unaffordable cost sharing and MA plans that exclude the providers that would serve them best.

V. A Proposal for Reform

The fundamental structure of TM is based on outdated insurance design and is incomplete, regressive, and limited in its ability to control costs. Supplemental insurance, including Medigap, which has existed since TM's inception, reinforces these problems, making TM more regressive and costlier while acting as a roadblock to program reforms. Over the years, MA has responded to TM by filling in the gaps and crevices in the original program for those without supplemental coverage while nevertheless remaining tethered to TM both through the premium bidding benchmark system and through its reliance on TM's administratively set service prices. The program that has resulted is rife with distortions that promote positive selection into MA, leaves people who become severely ill after joining MA without affordable alternatives, and saddles the system with high administrative costs. In addition, the MA program, which in principle relies on competition rather than government bureaucracy to generate efficiency, largely operates in highly concentrated markets. The inherent flaws of TM, compounded by Medigap and distorted further by MA, means that incremental reforms to risk adjustments and altered policies toward upcoding will not address the structural problems that bedevil the overall Medicare structure today. Turning Medicare into a program that can be efficient, progressive, and sustainable over the next 60 years will require revisiting the program's overall structure. A good place to start is the recent evolution in TM that has introduced some supply-side policies aimed at utilization management and provider-side incentives for quality and cost control.

The first step in any reform proposal would be to simplify how beneficiaries enter the Medicare program. That depends on both the enrollment process and the complexity of the structure of Medicare. The complexity and baroque structure of traditional Medicare, with its multiple parts (A, B, and D)—each with separate enrollment processes and subsidy structures, and each incorporating substantial variation in benefit design—serve to complicate enrollment, promote positive selection into MA, and inhibit consumer choice and competition (Ochieng et al. 2024).³

The second step will require establishing a modern benefit design that encompasses all medical care services under one policy. That would eliminate the distinctions between Parts A, B and D and would in turn force a rethinking of the enrollment process and the trust funds (Davis, Schoen, and Guterman 2013). The separation of Medicare components also creates “wrong bucket” problems, where savings achieved in one part of the program do not automatically flow into a different bucket (Cutler 2024). Pharmaceutical innovations such as Ozempic, for example, place new stresses on Part D but might reduce costs in Part A. The terms of that trade-off are challenging to negotiate in the current multipart system.⁴

A new benefit design should incorporate these three principles: (1) a single uniform benefit design that would apply to all Medicare beneficiaries; (2) premiums and cost sharing that are based on income; and (3) a catastrophic stop-loss. We anticipate that the catastrophic threshold would be set based on existing data from private insurance. For example, a catastrophic trigger for individual beneficiaries might be set at between $3,000 and $5,000 in 2023 dollars (the 57th to the 68th percentile in private health insurance plans) (Claxton et al. 2023). The impact of applying these principles would be to attenuate selection incentives, enhance the effectiveness of consumer choice and competition, and ensure affordability of coverage and care for all beneficiaries. Adhering to these principles would also provide the most valuable form of insurance protection to all Medicare beneficiaries: protection against large losses. In addition, modernization of the benefit design offers the opportunity to eliminate supplemental coverage (Medigap). That would have the effect of reducing the extra costs generated by the wraparound coverage and would make the program more progressive.

The third key element would introduce a consistent, modern approach to care management and cost control. Lessons from the Medicare Shared Savings Program and ACO demonstrations suggest a platform for a new organizational structure for care delivery in Medicare. The system would be built around competing organized delivery systems, such as ACOs, preferably physician-driven, that are equipped with care management tools and subject to financial incentives that promote both efficiency and quality care. These organizations would be paid through a gain-sharing program that features two-sided risk around a spending benchmark, meaning that cost savings and excess costs would be treated symmetrically so ACOs would face risks of gains and losses.

The details of how the spending benchmark is established are important for generating incentives to deliver efficient care. For example, the spending target or benchmark could be set according to a mix of the provider organization's own providers and regional providers for multiple years. In all cases, the provider organizations bear the risk of excess costs (above a payment benchmark), rather than beneficiaries. The gain sharing would continue to require organizations to meet quality targets as a precondition for collecting any savings. The payment or cost target would be risk-adjusted according to the composition of the ACO beneficiary population. There would be risk sharing for losses by the ACO above a predetermined level to account for shortcomings in risk adjustment. This combined design would mitigate some of the risk selection and coding intensity incentives in the current program.

Finally, Medicare would continue to establish administrative payment rate maximums. As in the current ACO program, beneficiaries would continue to have free choice of provider, with payment at these rates. In areas of the country without sufficient population to support multiple organized delivery systems, CMS would continue to use fee-for-service payment system incentives and quality metrics to ensure performance.

Conclusions

The American political system favors incremental reforms to public programs. The last 40 years of Medicare policy fit squarely in that tradition, yet we are left poorly prepared for the rest of 21st-century health care. The US population is aging rapidly and becoming more unequal in income and wealth, and it faces the benefits and financial requirements stemming from continued advances in technology. Addressing spending growth in Medicare along with the economics of Social Security are central to putting the nation's fiscal house in order. For these reasons, incremental reforms to Medicare will not support the demands on the program. Because MA serves as a patch to the outdated TM program, it is unlikely to successfully stand alone as the modernized Medicare program. Our vision is for a program overhaul that is grounded in the original purpose of Medicare: to expand access to health care while at the same time reducing the financial risk to older adults. Our preferred program design would recognize the heterogeneity in economic circumstances of older adults and incorporate lessons about how payment and administrative mechanisms can be put in place to manage the costs and quality of care. A simplified program structure is key and could make program entry easy and coverage understandable and fair, and it would move accountability for management of care and quality closer to the clinical encounter. It is past time to comprehensively modernize Medicare.

Acknowledgments

This work was funded by the Commonwealth Fund under grant number 23–23526.

Notes

References