How Accountable Care Organizations Try to Pay for Health, Not Volume

An Accountable Care Organization, or ACO, is a group of doctors, hospitals, and clinicians that agrees to be judged against a single yearly spending target, called a benchmark, for a defined group of patients. Beat the benchmark while meeting quality measures, and the group keeps a share of what it saved. Under the tougher arrangements, run over the benchmark and the group owes money back. The goal is to reward keeping people well instead of billing for more visits, scans, and procedures. Whether that goal is met turns almost entirely on one unglamorous question: how the benchmark is calculated.

It takes no position on whether any particular model is good policy.

Key points#

• An ACO sits on top of ordinary fee-for-service billing and settles up against a spending benchmark at year end.
• "Upside-only" models pay a bonus for savings but carry no penalty; "two-sided" models add a repayment obligation for overspending.
• The benchmark, and specifically how it is reset over time, is the single most important design choice.
• Medicare's newest test model, LEAD, runs ten years and is built to fix problems that pushed out smaller and rural practices.

Start with the pricing problem#

Traditional Medicare pays fee-for-service, meaning every service triggers its own separate payment. That arrangement pays dependably for volume. What it does not do is put money behind the scan a patient did not need or the hospital readmission a clinician managed to prevent.

An ACO does not replace that plumbing. Clinicians keep billing the usual way through the year. What changes is the reckoning at the end: the combined spending on the group's attributed patients gets compared to the benchmark, and the gap becomes either shared savings or shared losses. In effect, the ACO is a scorecard laid over the existing payment system rather than a new payment system of its own.

The scale is no longer small. Centers for Medicare and Medicaid Services (CMS) figures put roughly 14.3 million people in Traditional Medicare under ACO care as of January 2026, up from 13.7 million a year earlier. The Medicare Shared Savings Program by itself covered 511 ACOs and about 12.6 million people, drawing on more than 700,000 participating providers and organizations.

The benchmark is the rulebook, not a footnote#

It is tempting to treat the benchmark as a technical setting. It is closer to the rules of the game.

Set it generously and an ACO collects bonuses for spending that was always going to be modest. Set it too tight and even a genuinely efficient group loses money and walks away. Two mechanics do most of the damage or repair here.

The first is rebasing. Benchmarks are periodically reset toward an ACO's own recent spending, so a group that trims costs one year can watch its future target drift downward, in effect being taxed for having succeeded. To blunt that, CMS has stretched agreement periods to at least five years so the reset happens less often. The second is regional adjustment. Blending a group's own history with what its wider region spends rewards being efficient compared with neighbors rather than only compared with your own past, which matters most for groups that started out lean.

One-sided or two-sided: how much skin in the game#

The other lever is who absorbs the swings.

In a one-sided, or upside-only, arrangement, a group shares the savings when it beats the benchmark and owes nothing when it misses. The pull toward efficiency is real but soft, because the worst outcome is simply no bonus. A two-sided, or downside, arrangement adds repayment of a slice of any overspending. That sharpens the signal and discourages a group from pocketing savings in the good years while shrugging off the bad ones.

The Shared Savings Program lays this out as a climb. Its BASIC track runs from level A to level E, starting upside-only and adding downside risk with each rung, topping out at a 50 percent savings share and a 30 percent loss share at level E. The separate ENHANCED track carries the heaviest risk and the largest reward, with a savings share up to 75 percent. CMS reports that in 2026, 82.8 percent of Shared Savings Program ACOs sat at level E or in the ENHANCED track, the highest concentration of risk-bearing participation since the program launched in 2012.

More risk is not automatically better care. Downside risk pushes groups to manage spending hard, but it can also scare off smaller, rural, or independent practices that cannot survive one bad year, which narrows who takes part at all.

CMMI's experiments: REACH winds down, LEAD begins#

The CMS Innovation Center, known as CMMI, runs models that go further than the standard program, and its choices are a useful window into how design steers behavior.

The current advanced model, ACO REACH, closes on December 31, 2026. It included a Global option under which a group could take on the full total cost of care, effectively all of the savings and all of the losses, through capitation that pays a fixed amount rather than a fee per service. To keep from overpaying, CMS applied a benchmark discount that grew over the years, reaching 4 percent under the Global option in 2026. That discount is a plain policy dial: turn it up to claw back more savings up front, at the cost of a thinner margin for participants.

Its successor, the LEAD model (Long-term Enhanced ACO Design), starts January 1, 2027, and runs a full ten years through December 31, 2036, with a request for applications due in March 2026. Two signals stand out. The ten-year window is the longest CMS has tried and is meant to spare groups the churn of frequent rebasing, giving them a steadier target to invest against. And CMS has said the model uses revised benchmarking meant to draw a wider mix of participants, including smaller, independent, and rural practices and those caring for high-need patients such as people dually eligible for Medicare and Medicaid. Taken together, these point squarely at the known weak spots of earlier designs: targets that fell after success, and risk terms only the largest organizations could shoulder.

How to read any ACO claim#

The record so far is mixed rather than blank. In performance year 2024, the most recently settled year, Shared Savings Program ACOs generated 4.1 billion dollars in shared savings and saved Medicare 2.5 billion dollars, by CMS accounting. Those numbers lean heavily on how the benchmarks were built, which is exactly why the benchmarking method, not the name on the model, deserves the scrutiny.

So when you meet a headline about an ACO, three questions cut through it. How is the benchmark calculated and rebased? How much downside risk does the group truly carry? And which patients are attributed to it? The answers separate a model that rewards genuinely better care from one that mostly rewards a favorable spreadsheet.