Last August, CMS released the 2024 results for the Medicare Shared Savings Program, its largest accountable care program. The headline was a record: according to the CMS fact sheet, 476 accountable care organizations generated $2.4 billion in savings to Medicare relative to their benchmarks. Read one line further and a second number appears. Those same ACOs earned $4.1 billion in shared savings payments.
Both numbers are true at once, and the gap between them explains most of what providers misunderstand about value-based contracts. The ACOs saved money against a benchmark the payer set. The payer paid out more than it kept. Both sides call it a success, and both are right, because they are measuring different things.
If you negotiate these contracts from the provider side, here is the thesis worth sitting with: a payer does not sign a value-based contract primarily to save money. It signs one to buy predictability. Savings are welcome. Predictability is the product.
What the payer is actually buying
A health plan’s core problem is not that care is expensive. Expensive is fine if it is priced correctly. The problem is variance. A premium is set months before the claims arrive, and a bad year for utilization can erase a year’s margin.
A capitated or two-sided risk contract moves part of that variance onto the provider. That is the real transaction. The provider group accepts a spending target, and in exchange it gets a share of any underspend and, in downside arrangements, a share of any overspend.
Seen this way, several things that frustrate providers start to make sense.
- Payers care more about the benchmark than the savings rate. A benchmark that is too generous transfers money to the provider without changing care. A benchmark that is too tight drives good groups out. Most contract fights are really benchmark fights.
- Payers tolerate thin clinical ambition. A contract that reduces avoidable admissions by a modest amount, reliably, is worth more to an actuary than a bold redesign with uncertain results.
- Payers are wary of upside-only deals. Upside-only contracts give the provider the good years and leave the payer with the bad ones. From the payer’s chair, that is not risk sharing. It is a bonus program.
A benchmark, worked through
The mechanics are easier to see with a stylised example. The numbers here are invented for illustration and do not describe any real contract.
Suppose a provider group takes responsibility for 20,000 members with a benchmark of $12,000 per member per year, so $240 million in expected spending. The contract has a 2 percent minimum savings rate, a 50 percent shared savings rate and a symmetrical downside capped at 5 percent of the benchmark.
In a good year, actual spending comes in at $234 million, 2.5 percent under. The group has cleared the minimum, so the $6 million of savings is split: $3 million to the provider, $3 million kept by the payer. In a bad year, spending runs to $246 million, and the group owes back half the overrun, $3 million, well within its cap.
Now look at what the payer has really bought. Without the contract, its spending on these members might have landed anywhere between those two outcomes and beyond. With it, every dollar of swing inside the corridor is halved. The payer’s expected savings might be modest. The narrowing of its possible outcomes is the prize, and it is worth paying for even in years when the provider collects a bonus. That is also why a payer will often accept a lower shared savings rate for itself in exchange for more downside from the provider. It is buying a tighter corridor, not a bigger discount.
The same example shows why the Medicare figures in the opening can look lopsided. If the benchmark itself sits above what spending would have been anyway, the “savings” are partly measured against a generous target, and the shared savings payments can exceed what the program actually keeps. Whether that is a flaw or the price of participation is exactly the argument payers and providers keep having.
The member you save is often someone else’s member
Commercial payers have a problem that Medicare largely does not: their members leave. People change jobs, employers change plans, and brokers move whole groups at renewal.
That changes the economics of prevention. If a provider invests in controlling a member’s diabetes this year, the avoided amputation or dialysis may arrive five years from now, when that person is enrolled with a competitor. The payer that funded the prevention does not collect the return.
This is why commercial value-based contracts tend to reward things that pay off within the contract year: fewer emergency visits, fewer readmissions, more generic prescribing, site-of-care shifts from hospital outpatient departments to cheaper settings. It is not that payers do not believe in long-term prevention. It is that their balance sheets cannot see it.
Medicare is the partial exception, because a beneficiary tends to stay in traditional Medicare for years. That is one reason the most mature accountable care arrangements in the United States sit there rather than in commercial insurance. Medicaid, with its frequent churn in and out of eligibility, has the opposite problem.
The ratchet nobody likes to name
The second uncomfortable feature is the ratchet. If a provider group beats its benchmark this year, what should next year’s benchmark be?
If the payer rebases to the new, lower spending, the group is punished for its own success and must keep finding fresh savings from an already efficient base. If the payer never rebases, it keeps paying bonuses for savings that were banked years ago.
Every program tries to split the difference, blending historical spending with regional trends, and every blend creates winners and losers. Groups that started out inefficient find it easy to beat their benchmarks. Groups that were already lean find it hard. That is a strange incentive for a program meant to reward efficiency, and payers know it. They live with it because the alternatives are worse.
Where the payer’s view is wrong
The payer’s perspective is not automatically the wise one. Three habits deserve pushback.
First, payers often underinvest in the data plumbing that makes value-based care possible. A provider cannot manage total cost of care if claims data arrive months late or in formats that need weeks of cleaning. Asking a group to take risk on spending it cannot see is not partnership.
Second, payers frequently stack quality metrics until the contract becomes a checklist. Dozens of measures dilute attention. A handful that matter clinically, and that the provider can actually move, would do more.
Third, payers underrate the administrative cost they impose. Each payer has its own attribution method, its own measures and its own reconciliation timeline. A group with contracts from several payers can spend more on compliance than it earns in shared savings. Payers that align their terms with Medicare’s, or with each other’s, make value-based care cheaper to participate in, which ultimately helps them.
The strongest case against the payer’s framing
A provider could reasonably reply that all this makes value-based care sound like reinsurance with a clinical costume, and that this is precisely the problem. If the payer only wants predictability, it will design contracts that reward coding intensity and benchmark gaming rather than better care. Groups learn quickly that documenting risk scores more thoroughly can move the benchmark more than any care management programme.
That critique partly holds. Risk adjustment is where the predictability logic is most exposed, because the same diagnoses that make a benchmark fair also make it manipulable. Payers know this, which is why they audit coding and why programmes keep adjusting how much risk score growth they will recognise. But the critique does not overturn the thesis. It sharpens it. A payer buying predictability wants a benchmark that neither side can move cheaply. Providers who show they are not gaming the benchmark, with coding patterns that look like their population rather than like an optimisation exercise, are offering the payer exactly what it wants, and should be able to trade that credibility for better terms.
A note on the policy backdrop
In the United States, federal ambition has shifted. Under the previous administration, the CMS Innovation Center set a goal of having every traditional Medicare beneficiary in an accountable care relationship by 2030. CMS reported that just over half were in one as of January 2025. The Innovation Center’s May 2025 strategy refresh did not restate that 100 percent target. For commercial payers, who often follow Medicare’s lead on contract design, a softer federal push means fewer external reasons to expand risk arrangements and more reliance on their own business case.
That makes the payer’s calculation more central, not less.
Second-order effects
Once you see the contract as a variance transfer, some downstream consequences come into focus.
Risk flows to whoever can hold it. Small independent practices cannot absorb a bad year, so they either stay in upside-only arrangements or join aggregators and management services companies that pool risk across many practices. That pooling is useful, but it also moves decision-making, and a slice of the savings, away from the clinicians delivering care.
Capital follows the same logic. Firms that can finance downside risk gain bargaining power, which helps explain why investors have poured money into companies that sit between payers and physicians. The value they add is partly clinical and partly balance sheet.
And hospitals feel the squeeze. A contract that pays for fewer admissions and cheaper sites of care is, from a hospital’s income statement, a contract that pays it to shrink. Health systems that own both the hospital and the physician group face that tension inside their own walls.
What this means for each actor
For provider executives and physician group leaders, the practical advice is to model the benchmark before the care programme. Know how it is set, how it will be rebased and how attribution works before debating which clinical interventions to fund. Ask for claims data timelines in writing.
For payers, the smartest investment may be the least glamorous: timely, clean data feeds and a small set of measures aligned with other payers. Both reduce the cost of participation and widen the pool of groups willing to take risk, which is the only route to the predictability they want.
For employers who buy coverage, it is worth asking insurers what share of spending actually sits in two-sided arrangements, as opposed to upside-only bonus programmes relabelled as value-based care.
For policymakers, the lesson is that benchmark design is the policy. Targets for the number of beneficiaries in accountable care relationships mean little if the benchmarks those relationships run on reward the wrong groups.
For founders selling into this market, remember who your customer is. Tools that help a risk-bearing group see spending sooner or predict its year-end position speak directly to the payer’s and provider’s shared problem. Tools that promise long-term prevention will struggle to find a buyer whose balance sheet can see the return.
The takeaway
Providers who treat value-based contracts as a clinical program with a financial wrapper tend to be disappointed. Providers who treat them as an insurance transaction with a clinical engine tend to negotiate better. Ask what variance the payer is trying to shed, how the benchmark will move after a good year and who keeps the member long enough to see the return. The answers will tell you more about the contract than the quality measure list ever will.
What to watch
Over the next 12 to 24 months, three things are worth tracking. First, how CMS handles benchmark and risk adjustment rules in the Shared Savings Program, since commercial payers often borrow Medicare’s approach. Second, whether the next round of program results shows the gap between gross savings and net savings to Medicare narrowing or widening. Third, whether large commercial insurers keep expanding two-sided arrangements under their own steam now that the federal push is softer. If they do, it will be because the predictability case stands up on its own.
If you have negotiated a value-based contract, what was the term that mattered most once the money started moving? Tell us in the comments.
