Accountable Care Org Savings: Hard To Measure, Hard To Find

Accountable Care Organization (ACO) savings are central to a decade‑long experiment in paying differently for Medicare care. Policymakers hoped that encouraging groups of clinicians to coordinate care and share financial responsibility would slow spending growth while improving quality. But measuring whether ACOs actually produce net savings for the government, after bonuses, administrative costs, and shifting patient lists are counted, has proved surprisingly difficult, and the financial gains, if any, appear modest.

Why It Matters

How we pay for care shapes what care looks like. If ACOs reliably reduce spending without harming outcomes, they offer a way to move away from fee‑for‑service finance and toward value‑based payment. That would influence Medicare’s long‑term budget picture and set a model for private payers. Conversely, if ACOs mainly move money around, through bonuses, accounting changes, or participant selection, they may offer little public value while adding administrative complexity to the system.

Lingering Uncertainty

The uncertainty matters for multiple reasons. First, Medicare is a very large payer, so even small percentage improvements translate into substantial dollars. But small measured effects are also susceptible to noise: changes in which clinicians participate, how patients are attributed to a group, and how conditions are coded can all create the appearance of savings where none exist. That makes decisions about program design, expansion, and regulatory oversight high‑stakes despite ambiguous evidence.

Second, the ACO model carries downstream consequences for patients and clinicians. Incentives to reduce hospital use and prioritize outpatient management can improve care coordination and reduce avoidable procedures. Yet those same incentives can encourage practices to select healthier patients, increase use of preventive visits that change attribution, or focus attention on billing practices that raise measured patient risk scores without corresponding changes in health. These behaviors can widen disparities if providers serving sicker, more complex, or more disadvantaged populations face higher benchmarks and then exit or underperform in the program.

Finally, the administrative and operational costs of running ACOs, from data reporting and care management programs to claims reconciliation and quality measurement, are real. Those costs are borne both by the government and by clinician groups, and they can blunt or offset any apparent savings. Policymakers and program administrators, therefore, need clear evidence that the benefits outweigh the burdens before scaling models broadly or making them the backbone of payment reform.

Complexity of the Counterfactual

The difficulty in evaluating the MSSP stems from the fact that it is a voluntary, nonrandomized program. Unlike a controlled experiment, estimating savings requires identifying a “counterfactual,” defined as what spending would have been without the program, in an environment where participants have strong financial incentives to manipulate their benchmarks.

Recent analysis by Khullar and colleagues initially suggests a gross savings of $20.1 billion to $29.2 billion over 12 years, but these figures are highly sensitive to how one accounts for “selection dynamics” and “risk-coding changes”.

For instance, studies using beneficiary-level fixed effects, which better account for patients moving in and out of ACOs (churn), tend to find much smaller savings than studies using broader ACO-level effects. This discrepancy occurs because ACOs that perform poorly relative to their benchmarks often exit or reconstitute, while practices serving healthier, lower-cost patients are strategically added to improve measured performance. These subtle shifts in patient mix can create the illusion of savings that do not reflect actual changes in clinical utilization

Who It Affects

patients

Patients are at the center. Older adults and people with complex chronic conditions could benefit from better care coordination if an ACO reduces gaps in follow‑up, improves medication management, and prevents avoidable hospital stays. But the distribution of those benefits is uneven. Practices may be incentivized, intentionally or not, to attract or retain lower‑cost patients, meaning people with multiple comorbidities, dual eligibility for Medicaid and Medicare, or social needs may be underrepresented in participating panels.

Clinicians and healthsystems

Clinicians and healthsystems face practical trade‑offs. Primary care clinicians, in particular, are asked to take on population health tasks that require time and infrastructure: outreach to high‑risk patients, managing transitions from hospital to home, and tracking quality measures. For some practices, these investments pay off through shared savings; for others, the administrative burden and financial risk don’t justify participation. Convening organizations that enroll many small practices can lower barriers to entry, but they can also create incentives to optimize participant mixes rather than clinical care.

Payers

Payers and purchasers, including the Centers for Medicare & Medicaid Services (CMS), private insurers, and employers, watch ACO results closely because they inform broader payment policy. If ACOs demonstrate durable savings without compromising care, payers may expand similar contracts. But mixed or fragile evidence compels more cautious approaches: tighter risk adjustment, performance thresholds, or alternative payment models that include downside risk. Finally, taxpayers and policymakers weigh program costs, payments, and oversight needs when deciding whether to expand ACO‑style payment at scale.

The “Arms Race” of Risk Coding

For clinicians and healthsystems, the MSSP has introduced what some leaders describe as a coding “arms race”. Because benchmarks incorporate patient risk scores, ACOs can generate measured savings simply by increasing the intensity of their diagnostic coding, even if the patient’s actual health remains stable. Data shows that risk scores in ACOs have risen significantly faster than in non-ACO populations, roughly 1.5% higher by year six, a magnitude that is similar to the total estimated savings of the program.

This behavior is particularly pronounced among conveners who are third-party firms that facilitate participation but may also exploit benchmarks by strategically aggregating low-cost clinicians. Consequently, the savings reported may often be a byproduct of administrative upcoding, rather than genuine efficiency.

What Changes

• Expect continued scrutiny of how savings are measured and reported: program administrators will likely tighten rules on attribution, benchmarking, and risk adjustment to reduce gaming and better reflect true cost changes.
• Policymakers and payers will increasingly balance payments tied to spending with safeguards for equity: for example, adjustments for social risk and mechanisms to prevent disenrollment of sicker patients.
• Clinics and health systems that want to participate should invest in primary care capacity, data infrastructure, and care management, but also plan for uncertain financial returns and possible administrative overhead.
• Future payment designs will probably mix incentives: combining shared savings, downside risk, and quality pay, all while sharpening audit and oversight tools to detect coding changes that inflate risk scores without clinical change.

Where do we go from here?

Program evolution will likely focus on two dual aims: making measurements more robust and making incentives fairer. Better measurement means improving attribution rules so that beneficiaries are matched to the clinicians who actually direct their care, strengthening risk‑adjustment methods to reduce the influence of coding intensity, and designing benchmarks that reflect local cost environments without encouraging selective participation. Fairer incentives mean protecting providers that care for high‑need populations through explicit adjustments or targeted support so they are not penalized for serving sicker patients.

Operational Modifications

Operationally, this will require investment. Stronger data systems are needed to tie clinical records to claims, monitor coding patterns in near real‑time, and provide transparent feedback to participants. Auditing and compliance functions must also be scaled up to deter strategic behavior that inflates measured risk. For smaller practices, technical assistance and convener models can help, but those arrangements should be structured so they reinforce clinical improvement rather than simply optimizing financial benchmarks.

Clear Goals

Clinicians and health leaders should approach ACO participation with clear goals. If the aim is to improve care coordination and patient experience, practices should prioritize staffing and workflows that enhance follow‑up, medication reconciliation, and social supports. If the aim is to generate financial returns, organizations need realistic models of upfront costs, expected time horizons for returns, and contingencies if benchmarks change. Importantly, both goals are compatible when programs focus on outcomes that matter to patients, not just short‑term cuts in utilization.

Policymakers

For policymakers, the lesson is pragmatic: structure matters. Programs that rely on voluntary participation and retrospective benchmarking will always be vulnerable to selection effects and coding dynamics. To sustainably bend the cost curve, payment reforms need careful calibration — blending accountability with protections that prevent avoidance of high‑need patients, and pairing financial incentives with investments in primary care and social services that help clinicians deliver better outcomes.

Realistic Budgetary Expectations

From a policy perspective, the most sobering takeaway is the program’s actual impact on the federal budget. Even if one accepts the higher-end estimates of $13.4 billion in net savings over 12 years, this represents only 0.03% to 0.11% of the annual Medicare budget. To put this in context, the government is projected to spend $14 billion on quality payments alone in Medicare Advantage in 2026, far exceeding the total 12-year savings of the MSSP.

Moving forward, the best-case scenario for the MSSP is that it has resulted in only minor savings and has done little to “bend the cost curve” of national healthcare spending. Future iterations of the program must reckon with the high administrative costs borne by both CMS and participants, which further offset the modest financial gains. While the shift toward value-based payment remains a central goal of health policy, the evidence suggests that the MSSP, in its current form, relies on fragile estimates shaped more by selection and coding than by transformative changes in care.

Looking Towards the Future

In short, the accountable care idea, paying groups to manage populations rather than individual encounters, still has promise. But translating that promise into reliable, measurable savings for public payers has been harder than advocates expected. Future progress will hinge less on rhetorical commitments to “value” and more on technical fixes, stronger safeguards, and an honest accounting of costs and benefits across the health system.

Reference

1. Markovitz AA, Ryan AM. Accountable Care Organization Savings-Hard to Measure, Hard to Find. JAMA Health Forum. 2026;7(2):e256457. Published 2026 Feb 6. doi:10.1001/jamahealthforum.2025.6457