Why it matters
Under fee-for-service, a provider is paid for each visit, test and procedure, so more care means more revenue. Value-based care changes part of that incentive. Providers are rewarded for keeping total cost down and quality up for a defined group of patients.
It's already common. In the 2025 APM Measurement Effort (2024 payment data, released in 2026), AHIP found that 44.9% of U.S. health care payment dollars flowed through alternative payment models that tie payment to quality and cost, and 28.7% through models where providers can also lose money. Those figures cover Medicare, Medicaid and commercial plans together, and they include more than the shared-savings contracts described here.
For an employer, the question is practical. Your members may already be attributed to these arrangements through your carrier's network, and a self-funded plan may be paying the bonuses and fees.
How shared savings works
Most value-based contracts are still built on fee-for-service. The provider bills for each service during the year as usual. After the year closes, the payer compares total spending for the attributed patients with a target, often called a benchmark.
- Attribution. Patients are assigned to a provider group, usually based on where they get most of their primary care.
- Target. A spending benchmark for those patients, typically built from past cost, trended forward and adjusted for how sick the patients are.
- Quality gate. The provider has to meet quality measures to earn any savings.
- Reconciliation. Spending is compared with the target after the year ends. Medicare's Shared Savings Program, for example, releases results the following fall.
Many contracts also add a care coordination fee, a small per-member-per-month payment to support care management, paid whether or not savings appear.
Worked example: one contract, one year
An illustrative shared-savings contract. All 1,000 members are attributed to the group for the full year, the quality goals and any minimum savings threshold are met, and care fees are paid on top of claims. The 50% share and $2 PMPM fee are assumptions, not standard rates.
| Item | Amount |
|---|---|
| Spending target | $6,000,000 |
| Actual claims | $5,700,000 |
| Savings against target | $300,000 |
| Group's bonus (50%) | $150,000 |
| Care fees ($2 PMPM) | $24,000 |
| Total plan spending | $5,874,000 |
Claims came in $300,000 under the target. The group earned half of that, $150,000, and the plan paid $24,000 in care fees during the year. The plan finishes $126,000 below the target. That's savings measured against the target, not proven savings caused by the contract.
Read the transcript
Under fee-for-service, a doctor is paid for every visit, test and procedure. Value-based care ties part of that pay to cost and quality. One common version, shared savings, adds a scorecard: beat a spending target, meet quality goals, and share the savings.
In 2024, about forty-five percent of health care payment dollars flowed through models like these. Many still pay per service, then reconcile against the target after the year ends.
Say a doctor group is responsible for a thousand of your members all year, with a target of six million dollars. Claims come in at five point seven million: three hundred thousand under target. The group meets its quality goals and earns half: a hundred fifty thousand.
If your self-funded plan pays that bonus, plus care fees of, say, twenty-four thousand, you finish a hundred twenty-six thousand below target.
But beating the target doesn't prove the contract reduced spending. An overly generous target can trigger a bonus even without real savings.
So ask your carrier three things. Which doctors are our members attributed to? What have we paid in bonuses and care fees? And how is each target set? Find out more at actuariallysound dot com.
Who pays the bonus
Self-funded plans. The employer pays the claims, so it can also be charged the care coordination fees and shared-savings bonuses earned by providers in the carrier's network. Mercer documented these charges being invoiced to employers or deducted directly from their claims accounts, sometimes without the employer knowing. That article is from 2016, so treat it as evidence that this happens, not as a measure of how common it is today. Ask your carrier for a report.
Fully insured plans. The insurer pays these costs, and they're built into the premium like any other cost of care.
The catch: savings against a target
Beating the target isn't the same as saving money. The target is an estimate of what spending would have been, and a generous target can produce a bonus even when nothing changed. A tight target can do the opposite and hide real savings. The Congressional Budget Office makes the same point about Medicare's accountable care programs: a benchmark isn't an estimate of what spending would have been, so beating it doesn't measure the program's effect.
Targets are usually adjusted for how sick the patients are, so the provider isn't rewarded or penalized for the mix of patients it happens to treat. That adjustment, risk adjustment, is meant to make the comparison fair. It's also one of the places where a target can drift, which is why it's worth asking how yours is set.
Contracts also differ in how much risk the provider takes:
- Upside only. The provider shares in savings but owes nothing if spending runs over the target.
- Two-sided. The provider shares in savings and also pays back part of any overrun, subject to the contract's limits.
Most contracts also set a minimum savings rate, a threshold spending has to beat before any bonus is paid, so small swings that are just noise don't trigger payments.
Common misunderstandings
Questions to ask your carrier
- Which provider groups are our members attributed to, and how is attribution decided?
- What have we paid in care coordination fees and shared-savings bonuses, and who receives each payment?
- How is each target set and adjusted, including for how sick our members are?
- Are the reported savings net of every bonus and fee we paid?
- Are these contracts upside only or two-sided?
