A study published in the February 2026 issue of Health Affairs has provided the most comprehensive peer-reviewed evidence to date that voluntary bundled payment models are unlikely to generate meaningful or sustained savings for the Centers for Medicare & Medicaid Services (CMS).¹ The study, authored by researchers at Brown University and Brigham and Women’s Hospital, examined the Bundled Payments for Care Improvement Advanced (BPCI-A) Model across its first four model years (2018-2021) and found that while participating hospitals reduced their episode spending, large incentive payments to those hospitals resulted in net losses of $171 million for CMS over the study period.² The findings arrive at a pivotal moment for Medicare payment policy, as CMS recently launched its new mandatory Transforming Episode Accountability Model (TEAM) and the CMS Innovation Center has signaled an increasing commitment to mandatory model designs. This Health Capital Topics article examines the study’s key findings, implications for mandatory versus voluntary payment models, and the broader landscape of alternative payment model (APM) adoption.

Background on BPCI-A

BPCI-A was launched in October 2018 as a voluntary APM operated by the CMS Innovation Center.³ Under the model, participating hospitals and physician groups received a single, retrospective payment for a 90-day clinical episode, sharing in savings if they reduced costs below a target price while meeting quality thresholds.⁴ Although both hospitals and physician groups could participate, the majority of participants were hospitals. The model’s design evolved substantially over its lifetime. In Model Year 4 (2021), CMS introduced two key structural changes: hospitals could no longer select individual conditions for participation and instead had to participate across entire service lines of clinically related conditions, and CMS applied retrospective adjustments to financial targets to more accurately reflect national spending patterns.⁵ These reforms were intended to reduce hospitals’ ability to cherry-pick favorable conditions and to mitigate overly generous bonus payments. BPCI-A ultimately concluded on December 31, 2025, after running for over seven years.⁶

Key Study Findings

The Health Affairs study examined all Medicare fee-for-service (FFS) beneficiary data from April 2014 through December 2021, evaluating spending changes across 883 participating hospitals and 1,772 nonparticipating hospitals.⁷ Across all clinical service lines and model years, BPCI-A led to an average $324 reduction in hospitals’ 90-day episode spending per episode.⁸ The largest per-episode spending reductions occurred in the orthopedics (-$797) and neurological care (-$767) service lines.⁹

Notably, the magnitude of gross spending reductions grew over the model years, from -$181 per episode in Model Years 1 and 2 to -$358 in Model Year 3 and -$462 in Model Year 4.¹⁰ When the spending reductions were stratified and compared across care settings, the greatest reductions occurred in skilled nursing facility (SNF) care (-$338 per episode), suggesting that most bundled payment savings stemmed from changes in institutional post-acute care utilization rather than improvements in acute care efficiency.¹¹ In total, CMS spending on SNFs was reduced by approximately $585 million during the study period.¹²

Despite these gross savings, CMS hospital incentive payments averaged $1,119 per episode overall, with per-episode payments of $1,443 in Model Years 1 and 2, $2,006 in Model Year 3, and -$112 in Model Year 4.¹³ When accounting for both gross savings and incentive payments, the net financial impact on CMS was a loss of $278 million in Model Years 1 and 2, a loss of $151 million in Model Year 3, and savings of $246 million in Model Year 4, resulting in a cumulative net loss of $171 million across all four model years.¹⁴ The only model year in which CMS achieved net savings was Model Year 4, following the structural reforms that required service line participation and retrospective target adjustments.¹⁵

The Selection Problem

The researchers highlighted a fundamental structural challenge inherent to voluntary bundled payment programs: favorable selection. In a voluntary framework, hospitals that anticipate earning bonuses are more likely to join, while those sustaining losses eventually exit. This dynamic was evident in BPCI-A’s participation trajectory. BPCI-A reached peak participation of over 2,000 providers in Model Year 3, but participation declined significantly in Model Year 4 after CMS tightened the model’s design.¹⁶ An earlier study by some of the same researchers, published in the Journal of the American Medical Association (JAMA), similarly found that BPCI-A generated a $279.2 million net loss to CMS during Model Years 1 and 2 alone.¹⁷

The researchers noted that while Model Year 4’s structural reforms did yield net CMS savings, hospital participation declined in Model Year 5 (the last performance year examined). They warned that savings in voluntary models “may have a short shelf life, as participants sustaining losses will exit.”¹⁸ The study concluded that “voluntary bundled payment is unlikely to generate meaningful or sustained savings for CMS” and that “CMS can likely increase savings by transitioning to mandatory bundled payment.”¹⁹

TEAM and the Shift Toward Mandatory Models

The study’s findings directly reinforce the policy rationale behind CMS’s TEAM, which launched on January 1, 2026, as a five-year, mandatory bundled payment model.²⁰ TEAM draws on lessons from BPCI-A and other Innovation Center models, holding approximately 741 hospitals in 188 selected markets accountable for the cost and quality of care across five high-volume surgical episodes through 30 days post-discharge.²¹ CMS has projected that TEAM will generate $481 million in savings over its five-year term.²²

Hospital groups have expressed significant opposition to the mandatory design. The American Hospital Association (AHA) has repeatedly urged CMS to make TEAM voluntary, arguing that “mandatory participation is inappropriate given that many of the selected organizations are neither of an adequate size nor in a financial position to support the investments necessary to transition to mandatory bundled payment models.”²³ In its June 2024 comment letter on the proposed TEAM rule, the AHA characterized the model as a “backdoor payment cut to hospitals,” citing BPCI-A’s history of cumulative CMS losses and noting that in four of the five TEAM episode categories, over 71% of costs occur during the initial hospitalization, leaving limited opportunity for downstream savings.²⁴

Notwithstanding industry opposition, CMS Innovation Center Director Abe Sutton stated in March 2026 that “mandatory models are going to have to be part of the equation,” signaling that the agency intends to continue pursuing mandatory model designs.²⁵ In addition to TEAM, CMS announced several new Innovation Center models in late 2025, including the Ambulatory Specialty Model (ASM), a mandatory episode-based payment model for outpatient specialists set to launch in January 2027; the ACCESS model, a voluntary outcomes-based chronic disease payment model; and MAHA ELEVATE, a $100 million grant-funded preventive and lifestyle medicine program.²⁶ Of these new models, only TEAM and ASM use episodic bundled payment structures; the remainder employ distinct payment mechanisms such as outcome-aligned payments, cooperative agreements, or drug pricing frameworks.

APM Adoption and the Broader Landscape

The Health Affairs study’s findings are consistent with broader data suggesting that voluntary APM adoption has plateaued. According to the 2025 APM Measurement Report from America’s Health Insurance Plans (AHIP), approximately 44.9% of healthcare payments across the U.S. were tied to APMs in 2024, a slight decrease from 45.2% in 2023.²⁷ Payments involving downside financial risk – the type of accountability central to bundled payment models – accounted for 28.7% of all payments.²⁸ While 70% of health plan respondents indicated they expected APM activity to increase over the next 24 months, and over half expected the greatest growth in shared-risk episode-based payments, the overall trajectory suggests that voluntary adoption alone may be insufficient to drive the scale of payment reform CMS envisions.²⁹

Some market analysts have questioned the efficacy of value-based payment models more broadly. In a January 2026 Health Affairs Forefront article, one of the study’s authors³⁰ argued that value-based payment and managed care approaches may not adequately address core healthcare spending drivers, positioning the BPCI-A findings within a larger intellectual framework questioning whether APMs can meaningfully contain healthcare expenditure growth.³¹ At the same time, a 2025 survey of hospital and health system executives found that 77% planned to increase value-based care participation in the next two years, even as only 20% agreed the industry had made meaningful progress in value-based payment over the preceding two years.³²

Conclusion

The Health Affairs study provides significant new evidence in the longstanding debate over voluntary versus mandatory approaches to bundled payment. While BPCI-A did succeed in reducing hospital episode spending and demonstrated that design reforms, such as requiring service-line participation and applying retrospective target adjustments, can improve CMS’s financial outcomes, the model’s overall track record of generating net losses underscores the structural limitations of voluntary participation. The pattern of favorable selection, in which hospitals exit when they face losses, appears to undermine the long-term sustainability of savings to the Medicare program.

With TEAM now operational and additional mandatory models on the horizon, CMS has clearly signaled its intent to build on these lessons. Whether mandatory bundled payment will succeed where voluntary models fell short – and whether participating hospitals can adapt to the financial and operational demands of mandatory episode accountability – remains to be seen.