The funding model underneath your plan usually matters more than the plan itself. Here is how the three compare in practice.
Most benefit conversations start with plan design — deductibles, networks, copays. But the funding model determines who carries the risk, who sees the data, and how much room you have to influence next year’s cost. It is the more consequential decision, and it gets far less attention.
The three models at a glance
| Fully insured | Level-funded | Self-funded | |
|---|---|---|---|
| Who carries claims risk | Carrier | Employer, with stop-loss | Employer, with stop-loss |
| Monthly cost | Fixed premium | Fixed payment | Variable |
| Claims data access | Limited | Good | Full |
| Surplus if claims run low | Kept by carrier | Often refunded | Retained by employer |
| Administrative load | Lowest | Moderate | Highest |
| Typical fit | Under ~50 lives, or volatile populations | ~50–250 lives | ~250+ lives with stable cash flow |
Fully insured
You pay a premium; the carrier takes the risk. It is predictable and simple, and for small or high-risk populations it is frequently the right answer.
The cost is visibility. You generally cannot see your own claims detail, which means when the renewal arrives you have limited ability to interrogate it. You are also funding the carrier’s risk margin whether or not your population uses it — a good year produces no refund.
Level-funded
A middle path that has become the default recommendation for mid-sized employers. You pay a fixed monthly amount covering expected claims, administration and stop-loss premium. If claims come in below expectation, a portion of the surplus is typically returned.
You get self-funded economics and data access with a fixed monthly payment. The trade-off is that underwriting matters — you will be medically underwritten to enter, and a population with significant known risk may not be offered attractive terms.
Self-funded
You pay claims as they are incurred, with stop-loss insurance capping exposure at both the individual and aggregate level. Maximum control, full data, and no carrier risk margin.
It requires the size to absorb month-to-month variability and the cash position to fund a bad quarter. It also requires actually using the data — self-funding without a population health strategy is just taking on risk without taking the corresponding action.
ICHRA as a fourth option
Defined contribution sits outside this spectrum entirely. Rather than choosing who carries claims risk, you exit the risk question and fix your cost, while employees select individual market coverage. Worth modeling alongside the other three, particularly for geographically distributed workforces.
How to decide
- Start with your risk profile, not your size. Headcount is a rough proxy; a scorecard showing contained risk and low variance opens doors that headcount alone would not.
- Model at least two years. A one-year comparison flatters whichever model happens to suit last year’s claims.
- Price the stop-loss honestly. Individual and aggregate attachment points drive the real economics of both self-funded and level-funded arrangements.
- Check your cash tolerance. Ask what a bad quarter looks like — if the answer is genuinely disruptive, fixed-cost structures are worth the margin you pay for them.
The most expensive funding decision is the one made by default. Staying fully insured because that is what you did last year is still a choice — just one made without the comparison.
This article is provided for informational purposes only and does not constitute legal, tax or benefits advice. Requirements vary by plan design, funding arrangement and jurisdiction. Contact BeneSkill to discuss how this applies to your plan.