Seven funding paths.
One honest comparison.

At 75 employees, fully-insured is the safe seat and usually the priciest, often 20 to 30 percent above the alternatives. Level-funded is where most groups this size start, since you pick up monthly claims data, a share of any surplus, and stop-loss protection without much added complexity. A captive gets interesting when you want self-funded economics with pooled-risk cover, roughly 10 to 20 percent under fully-insured plus laser protection. Pure self-funding only fits at 75 if you have a CFO in the room, steady claims, and room for cash-flow swings. Simple rule: if your last two years came in within 5 percent of expected, level-funded saves money at low risk. If you posted a year over 115 percent, a captive's pooling probably matters more than chasing surplus.

Taft-Hartley takes it. Rates come from the multi-employer trust and reflect the entire pooled membership, not your particular group, so one employer's bad claim does not jerk the whole trust upward. Losses spread across thousands of employers. Fully-insured ranks second on stability, but only because the carrier hides the swings from you, and your own renewal can still jump 30 to 50 percent once you are large enough to be rated. PEO comes third, since the sponsor's scale dilutes any single employer's claims. Level-funded, captive, and self-funded all sit higher on volatility by design. You are seeing your group's real experience, which is more honest and, yes, more variable.

Think of it as a ladder. From 25 to 50 employees, start fully-insured or level-funded; most carriers will not quote level-funded under 25, and it runs borderline under 50. From 50 to 100, level-funded is usually the right rung, and a self-funded captive becomes workable around 35 if you find the right one, since Blackwell, Roundstone, and Pareto draw their size lines differently. From 100 to 150, hold in level-funded or shift to a captive based on how stable your claims are, and start modeling pure self-funding. From 150 to 250, pure self-funding usually wins unless your claims are shaky, in which case stay put. Every jump adds compliance work, so move when the math changes, not just because you can.

Yes, but only in the patterns the IRS blesses. You can put one employee class, say salaried headquarters staff, on ICHRA while another class, like hourly field crews, stays on a traditional group plan. That combination is allowed as long as the classes track the IRS categories, full-time versus part-time, salaried versus hourly, and so on. What you cannot do is split people within the same class. PEO plus ICHRA generally will not fly, because the PEO contract makes its group plan the primary offering. PEO plus Taft-Hartley does work for union shops, since those trust obligations already exist. Most multi-arrangement setups need a lawyer's eyes first, because the ERISA traps here are real.

Pure self-funding wins hands down. Your TPA's portal shows every claim as it pays, broken out by member, provider, diagnosis code, and dollar amount. A self-funded captive gives you all of that plus pool-level analytics, so you can see how your group stacks up against other members. Level-funded hands you monthly claims reports, usually 30 to 45 days after the month closes, with similar detail but less immediacy. Fully-insured gives you next to nothing, maybe an annual claims-experience report with names redacted. Taft-Hartley and PEO trusts run claims at the pool level, so the data they share is thin. If transparency is what you are after, the order is self-funded, then captive, then level-funded, then everything else.

Rough monthly hours for an HR generalist run like this. Fully-insured takes 2 to 4 hours, mostly enrollment and employee questions. A PEO is the lightest at 1 to 2, since it handles administration for you. Taft-Hartley runs 3 to 6 for contribution reporting and trust coordination. Level-funded takes 4 to 8, between claims reconciliation, refund tracking, and plan-document compliance. A self-funded captive is 6 to 12, level-funded plus captive governance. Pure self-funding is the heavyweight at 10 to 20 hours: reconciliation, stop-loss coordination, ERISA compliance, plan-document upkeep, and fiduciary oversight. ICHRA lands at 4 to 8 for allowance reconciliation, tracking individual policies, and ACA reporting. Plenty of employers outsource the heavier ones to a consultant rather than carry them in-house.

Yes, in three concrete ways. First, state benefit mandates, things like expanded mental-health parity, fertility coverage, and autism care, apply to fully-insured plans but generally not to self-funded ones, which fall under federal ERISA preemption. That is one of the strongest cases for self-funding in California, New York, and Massachusetts, since you sidestep some of the costliest mandates. Second, state premium taxes, roughly 1.5 to 3 percent, hit fully-insured plans but not self-funded, which is real money on a plan over 1 million dollars. Third, state regulators police rate hikes harder on fully-insured plans, which can mean better protection from unjustified increases but slower carrier response. Net effect: heavy-regulation states tilt the field toward self-funded, and by extension level-funded and captive.

Three keep coming up. First, renewal increases are pushing fully-insured groups toward level-funded. The average 2026 fully-insured renewal is running 9 to 14 percent per KFF, while stable groups on level-funded are tracking 4 to 7 percent. Second, claims-data transparency has become a finance-team issue. Once a CFO sees that 25 to 30 percent of premium goes to carrier margin with no window into what is actually driving claims, level-funded becomes the obvious next step. Third, ICHRA is pulling employers off group plans entirely when the workforce is spread out or heavy on 1099 and W-2 mix, using the 11 employee-class rules. The common mistake is jumping the wrong way, straight from fully-insured to self-funded with no level-funded bridge in between.