In a 2022 book, Seemed Like a Good Idea, health economist Mark Pauly and his University of Pennsylvania coauthors describe how health-care policymakers “often rely on hope or conjecture, not rigorous evidence of effectiveness” when proposing and implementing new programs. Even when those programs fail, as they usually do, the authors argue that policymakers often persist in supporting them. Look no further than the Affordable Care Act, aka Obamacare or the ACA.

Obamacare’s proponents claimed that the American health-care system could and should be transformed from a fee-for-service model to one based on “value.” To make this transition, the law would promote the formation of accountable-care organizations, or ACOs—groups of health-care providers that assume responsibility and financial risk for the quality and costs of care for a defined group of patients. These groups would presumably yield superior care coordination, enhanced quality, and ultimately lower costs.

When Obamacare was enacted, commentators noted that the proposed ACOs looked like the integrated-delivery networks tried in the 1990s and the Medicare Coordinated Care Demonstration projects funded in 2002, both of which failed to lower costs and improve quality. But the health-policy community insisted that it knew better. A recent publication from the Congressional Budget Office (CBO) suggests that, as usual, they were wrong.

The federal government’s Medicare program operates ACOs through the Centers for Medicare and Medicaid Services (CMS) under two parallel tracks: the Center for Medicare and Medicaid Innovation (CMMI) and the Medicare Shared Savings Program (MSSP). Thirty-nine percent of Medicare’s fee-for-service beneficiaries are enrolled in an ACO.

CMMI was created by the Affordable Care Act in 2010 to test “innovative payment and service delivery models to reduce program expenditures . . . while preserving or enhancing the quality of care furnished to individuals” on Medicare, Medicaid, or the Children’s Health Insurance Program. CMMI operates ACOs and other value-based-care programs, such as bundled payments and medical-home models.

Obamacare appropriated $10 billion in mandatory funding for its first decade of operation, fiscal years 2011–2019—funding not provided through the annual appropriations process—to identify, develop, test, and evaluate payment models. In addition, it appropriated $10 billion for each subsequent decade. Payments for the medical services themselves, as well as for bonus payments for meeting quality and other metrics, still come out of the Medicare trust funds. These mandatory, ten-year appropriations ceded broad powers to CMMI to create programs without limits on scope and duration, free from congressional oversight.

The CBO originally projected that CMMI would save $2.8 billion from 2011 to 2020, and $77.5 billion over the next ten years. CMMI ran and published evaluations of the 49 health-care models operated by the center, nearly all of which focused on Medicare beneficiaries, during those first ten years. Only six saved money.

The CBO has found that, instead of saving money, “CMMI’s activities increased direct spending by $5.4 billion, or 0.1 percent of net spending on Medicare, between 2011 and 2020.” CMMI spent $7.9 billion to operate 49 models that reduced spending on health-care benefits by only $2.6 billion.

Why the discrepancy? The CBO’s estimate of CMMI’s expenditures on the models ($7.5 billion) was close to what the center actually spent ($7.9 billion). But the CBO’s savings estimate was way off; it predicted the center’s models would yield a $10.3 billion reduction in benefits spending, versus the actual $2.6 billion. The CBO assumed, without any detail about specific projects, that “CMMI would tend to identify models that reduced saving over time and then expand those models to produce greater savings over time.”

In fact, CMMI has certified only four models for expansion, at least one of which—the Medicare Diabetes Prevention Program—has shown no evidence of savings in Medicare expenditures in the expanded program. Another, the Pioneer Accountable Care Organization Model, began in 2012 with 32 participating ACOs and concluded with only nine participants in December 2016, after which it was incorporated into MSSP. The jury is still out on the others. Contrary to the CBO’s expectation that savings would grow as the CMMI identified and expanded the models that saved money, the CBO found that after the CMMI’s first two model certifications, its certification rates declined.

The CBO also has now updated its previous estimate that the center would save $77.5 billion over its second decade, 2021–2030. It now projects that CMMI will increase net federal spending by $1.3 billion over that period and acknowledges that CMMI might “achieve smaller savings than projected in its second decade” since the models CMMI has thus far implemented “may have already targeted the most easily addressable drivers of health care spending for providers, therefore making similar savings over the next decade more challenging to achieve.”

Other researchers have reached similar conclusions. Brad Smith, former head of CMMI, wrote in the New England Journal of Medicine that “the vast majority of the Center’s models have not saved money, with several on pace to lose billions of dollars.” Avalere, a health-care consultancy, estimates CMMI will produce $9.4 billion in net losses over the 2017–2026 budget window as “a result of models that did not generate expected savings and model savings that did not outweigh the cost of shared savings payments to participants.” Similarly, Health Management Associates, a health-care research and consulting firm, found that CMMI in its first decade showed “minimal success in fulfilling its . . . statutory criteria of lower spending or improved quality.”

The second ACO track, the Medicare Shared Savings Program, was a demonstration project that the ACA expanded and made permanent. Currently, 456 ACOs in the MSSP track provide care to 10.9 million beneficiaries.

CMS has reported savings compared with MSSP spending targets over the past six years. But as the CBO points out, these spending targets, known as benchmarks, do not approximate the counterfactual spending that ACO beneficiaries would otherwise have spent. Instead, they are administratively set objectives

for encouraging providers to participate and to reduce spending. That means that an ACO could generate savings as measured against a benchmark but generate no savings—or even have net costs—relative to an estimate of how spending would have evolved among an ACO’s patient population if the ACO had never formed.

When the CBO reviewed evidence of studies with counterfactual control groups, it found “mixed results.” While several studies identified small net savings, some failed to incorporate the costs of performance payments to the ACOs. Others found savings by assuming spillover savings from changes in the delivery of services to beneficiaries not enrolled in ACOs. But spillover effects can be positive (cost saving) or negative (cost increasing). Evidence of positive spillovers is largely confined to Medicare’s bundled payment-demonstration projects for hip- and knee-replacement surgeries, where a set amount is paid to cover all costs associated with the procedure.

Moreover, it is hard to estimate savings generated by MSSP and extrapolate them to non-MSSP participants, since providers participate on a voluntary basis—only if they think they are likely to achieve savings. If they fail to see savings, they withdraw. When the CBO reviewed three studies that tried to adjust for this provider-selection effect, it found that adjustments in two either wiped out or diminished the savings. The third study found that the adjustments had no impact on the “modest savings.”

Undeterred by these findings, CMS officials remain committed to plans that would move all Medicare fee-for-service beneficiaries into accountable-care plans. CMMI director Liz Fowler told attendees at the America’s Physician Groups conference on October 30 that “The most important driver in achieving broader health system transformation is [moving] 100 percent of beneficiaries in accountable care relationships by 2030.”

One should not be surprised by this determination to move forward. Health-policy planners rarely surrender an established program. Instead, they argue that it should be expanded with increased funding.

But the public and legislators should ask why we should fund these unsuccessful government experiments, particularly when the private market is more likely to come up with effective models for both private and government health-care programs. Private-group purchasers, such as employers and unions, have been demanding lower-cost, higher-quality health coverage for their subscribers. They have been developing lower-cost quality-improvement programs for decades. They have an important incentive to pursue value-based care: saving money.

Private-equity firms have noticed, as private investment in value-based care companies has grown more than fourfold from 2019 to 2021. Private investors and coverage providers hold an important advantage over government programs: they do not undertake and persist in coverage arrangements that lose money, and they do not double down on non-beneficial programs.

The CBO data demonstrate why we should not expand government health-reform programs launched without evidence of effectiveness. Beneficial changes to the American health-care system are far more likely to come from the private market than from unelected bureaucrats in Washington.

Photo by RHONA WISE/AFP via Getty Images


City Journal is a publication of the Manhattan Institute for Policy Research (MI), a leading free-market think tank. Are you interested in supporting the magazine? As a 501(c)(3) nonprofit, donations in support of MI and City Journal are fully tax-deductible as provided by law (EIN #13-2912529).

Further Reading

Up Next