Blog

The Promise of the Innovation Center

The Center for Medicare and Medicaid Innovation (CMMI) was established under the Affordable Care Act (ACA) in 2010 with a clear mandate: design, test, and scale alternative payment and service delivery models capable of reducing federal expenditures on Medicare, Medicaid, and Children’s Health Insurance Program, while preserving or improving the quality of care delivered. The Congressional Budget Office (CBO) estimated CMMI would save $2.8 billion in its first decade alone.

Since its first operational year in 2011, CMMI has operated 70 payment and delivery models spanning primary care, specialty care, accountable care organizations (ACOs), bundled payments, and population health management. As of January 2025, 24 models remain in active testing. Some tested models, like the Home Health Value-Based Purchasing (HHVBP) Model, have saved Medicare $949 million between 2016-2020. Likewise, the Pioneer Accountable Care Organization (ACO) Model, saved $384 million in its first two years.

Successes like these reflect the viability of CMMI’s mission, to realign incentive structures, rewarding quality and efficiency rather than volume. However, the central question raised by the Government Accountability Office’s (GAO) March 27, 2026 report is whether these successes are representative of CMMI’s potential or merely exceptions within a broader pattern of institutional failure.

Analysis of the GAO Report

Of the 70 models CMMI tested from 2011 through 2024, only four, representing a 5.7% success rate, met the statutory criteria for nationwide expansion. The CBO estimates these programs have achieved $2.6 billion in realized savings. However, CMMI spent $7.9 billion developing, implementing, and evaluating these models between FY 2011 and FY 2020, reflecting an estimated net loss of $5.4 billion.

CBO originally estimated CMMI would reach $77.5 billion in net savings between 2021 and 2030. However, current projections indicate CMMI will instead increase net federal spending by $1.3 billion over its second decade.

As Chairman of the House Budget Committee Jodey Arrington (R-TX) stated directly in response to the report: “A 5 percent success rate isn’t innovation—it’s failure. And with our national debt barreling past $39 trillion, we can’t afford programs that promise savings but consistently come up short.”

While the GAO report is not an assessment of whether CMMI models improved or degraded care quality, the fiscal evidence alone is deeply concerning for a center whose statutory purpose is to reduce federal spending. A threefold gap between program costs and savings reflects a systemic failure, not an institution operating within its mandate.

Deficiencies Identified by GAO

1. Voluntary Participation and Selection Bias: The overwhelming majority of CMMI models have been designed as voluntary programs. Of the 24 models in active testing, only two models are mandatory. The voluntary nature present in most models introduces well-documented methodological problems. Providers who opt into performance-based models are systematically more likely to be those already operating near performance thresholds. This induces selection bias that overstates the apparent efficacy of interventions. Furthermore, participation attrition, where providers exit models when financial incentives diminish, further undermines the long-term generalizability of measured outcomes. GAO identified both phenomena as factors that CMMI itself associated with lower savings potential. 
2. Insufficient Model Testing Horizon: CMMI’s typical model test duration is fewer than ten years; an inadequate time frame for demonstrating durable federal savings, particularly for models focused on preventive care and primary care management. The financial returns from preventive interventions compound over extended timescales, often nearing a decade. A patient better managing chronic disease at 50 may not show measurable savings until age 60. Testing windows shorter than this compounding horizon will, by design, fail to capture the full savings of preventive models, producing inconclusive or negative results for models that may, over appropriate timeframes, prove cost-effective.
3. Ineffective Quality Metrics: Statutory criteria require models to demonstrate the preservation or improvement of care quality to certify models for expansion. However, CMMI typically employs claims-based quality indicators; metrics that capture what services were provided but insufficiently reflects whether those services improved patient-level health outcomes. This opaque quality of care metric presents structural barriers to future expansion efforts.
4. Poor Model Design and Perverse Incentive Structures: Overly generous participant incentive payments have, in multiple models, produced apparent short-term savings that do not reflect genuine, long-term behavioral change. In these cases, incentive payments are high enough to alter provider behavior in the short-term, but the underlying care model does not restructure how services are delivered, thus, savings do not endure once the model ends or payment terms are adjusted. Additionally, poor model design has contributed to early termination and inconclusive evaluation. These outcomes deplete CMMI’s budget without generating usable data to judge model viability.

Paths Forward: Reform or Repeal

The GAO report’s findings compel action ,and policymakers must choose to reform or repeal CMMI. Each option carries meaningful implications for federal healthcare spending, the trajectory of value-based care, and democratic accountability of statutory obligation.

The Case for Reform: Chairman Arrington acknowledged that “CMMI’s mission is too important to abandon, but the status quo is unacceptable”. The four models that achieved nationwide expansion demonstrate that CMMI is not incapable of producing cost savings. Each of the expanded models have produced significant savings, which is evidence that well-designed, adequately measured models can fulfill the center’s statutory mandate. Furthermore, the report suggests some models have contributed value beyond formal expansion via integration to successor models. A reform agenda responsive to the GAO’s findings would prioritize a meaningful shift toward mandatory model participation, the adoption of durable patient-level outcome metrics, extended duration testing, and redesigned incentive structures that reward sustained behavioral and care delivery change. CMMI is already working to address some of the issues like selection bias and care quality metrics. The center plans to expand the proportion of mandatory models, identify quality goals during model design, and increase outcome-based measures across models.

The Case for Repeal: Considering the GAO report, it is plausible to conclude these findings demonstrate CMMI is structurally incapable of reform. The program has spent a decade and a half and billions of taxpayer dollars, all without ever demonstrating the net cost savings it was designed to generate. From a public financial perspective, a 5.7% success rate over 15 years and a 300% net operating loss, represents a complete and total failure to fulfill the program’s statutory mandate. If the structural deficiencies identified by GAO reflect embedded institutional norms rather than correctable procedural failures, the case for continued investment is null. Furthermore, if abolished, the center’s $8 billion in unobligated funds could be redirected toward programs which have a proven record of action and accountability.

Conclusion

The GAO conducted a neutral audit of CMMI, and its findings make the continuation of the center in its present capacity categorically indefensible. Under the current trajectory, the CBO projects the center will continue to increase, not decrease, federal spending by a factor of $1.3 billion over the next decade of operation. An institution whose mandate is to reduce federal healthcare expenditures, projecting an increase in those expenditures, is operating in direct contradiction to its purpose. While the discussion for reform versus abolition may be spirited, allowing the status quo to continue would be a disservice to the hard-working Americans who are left with the bill.