Sunit Patel, Marsh's US chief actuary for health and benefits, put a two-number projection on the record on September 2: total health benefit cost per employee rising 8.2% in 2027 after employers take the action they have already planned, and 11% if they leave their plans alone (Marsh, September 2026). Closing that 2.8-point gap takes 59% of employers changing plan design, most of it through deductibles, coinsurance and premium cost share.
The 8.2% would be the steepest increase since 2003 and the fifth consecutive elevated year, built on a 6.7% rise already projected for 2026. Every outlet ran that figure. The rate that belongs in a 2027 renewal model is the other one.
Key Takeaways
- 11% gross against 8.2% net across more than 1,800 employers is the widest gross-to-net gap published this trend season. Business Group on Health's comparable pair for 2027 is 9.2% down to 8%, a wedge less than half as large.
- One full percentage point of the 2027 increase is GLP-1 weight-management use, and Marsh estimates AI-assisted physician claims software could add as much as another point, putting up to two points outside ordinary utilization drift.
- 3.0 percentage points of cumulative cost go missing over two years if a model compounds 8.2% twice rather than taking the buy-down once and applying the 11% gross rate to the rebased plan afterward.
- Three consecutive years, 2023 through 2025, saw employers' actual health costs exceed their own pre-year projections (Business Group on Health, August 2026). That is the track record of the net number, not the gross one.
- 32% of employers name catastrophic claims as their leading 2027 driver (IFEBP, August 2026), and that is the one layer, above the out-of-pocket maximum, where a cost-sharing buy-down moves nothing.
The Two Rates Marsh Published
The preliminary results come from more than 1,800 US employers, surveyed from June 10 through August 10, with the full 2026 National Survey of Employer-Sponsored Health Plans due later this year (Marsh, September 2026). Respondents answered two separate questions: what their existing plans would cost in 2027 with no intervention, and what they expect to pay after the changes they have already decided on. The first answer was 11%. The second was 8.2%.
The action behind the wedge sits almost entirely in cost-sharing. Marsh reports 59% of employers planning cost-cutting design changes for 2027, and two-thirds of employers with 500 or more employees expecting to raise the employee share of premium. Twelve percent plan variable copay designs, rising to 18% among employers with 20,000 or more workers.
Mercer's earlier read on the same buyer set pointed the same way, with 48% of large employers expecting to raise deductibles or copays for 2027, against an average cost already above $18,500 per employee (Mercer, June 2026). "Few organizations can absorb health cost increases that outpace inflation without making difficult financial decisions," said Simon Camaj, Marsh's US health and benefits leader (Marsh, September 2026).
| 2027 projection | Before plan changes | After planned changes | Wedge |
|---|---|---|---|
| Marsh, 1,800+ employers (Sept 2026) | 11.0% | 8.2% | 2.8 pts |
| Business Group on Health, 127 employers (Aug 2026) | 9.2% median | 8.0% | 1.2 pts |
| IFEBP, 112 employers (Aug 2026) | 10.0% median | not split | not published |
| PwC, 27 health plans (June 2026) | 9.0% group | not split | not published |
Marsh's 2.8-point wedge runs more than twice the gap Business Group on Health collected from 127 large employers covering 11 million lives, 9.2% median before plan changes down to 8% after (Business Group on Health, August 2026). Either Marsh's respondents face a steeper gross rate or they are cutting harder. On the payer side of the same market, PwC put 2027 group medical trend at 9.0% across 27 health plans covering more than 103 million employer-sponsored members, and restated its 2026 group figure up from 8.5% to 9.0% (PwC, June 2026).
What the 2.8 Points Actually Buy
A deductible increase does not slow medical inflation. It changes the share of a given claims distribution that the plan pays. Raise a deductible, raise coinsurance, raise the out-of-pocket maximum, and the plan's paid-to-allowed ratio steps down once, at the moment the design changes. The allowed charges underneath keep compounding at whatever the gross rate is.
That makes 8.2% a rebased figure rather than a trend selection. It is the gross rate applied to last year's plan, net of a one-time reduction in actuarial value. By 2028 the reduction is already sitting in the base, and the gross rate arrives intact against the new, lower plan-paid level.
The arithmetic on Marsh's own two numbers shows the size of the error. Compound 8.2% twice and a plan's cost index reaches 1.171. Take the buy-down once and then the gross 11%, and it reaches 1.201. That 3.0-point divergence over two years is roughly $560 per employee against Mercer's $18,500 base. A plan sponsor or a stop-loss underwriter carrying the net rate forward as trend has booked a favorable variance nobody earned.
Business Group on Health supplies the track record. Employers' actual costs exceeded their own pre-year projections in each of 2023, 2024 and 2025 (Business Group on Health, August 2026), and the projection that missed was the net one. The same survey puts cumulative growth before plan changes at 76% over the ten years from 2018 to 2027.
Composition makes the gross rate harder to buy down than its size suggests. Marsh attributes a full percentage point of the 2027 increase to GLP-1 weight-management use, and estimates that AI-enabled software helping physicians submit claims could add as much as another point by producing more claims and higher-level claims for payment (Marsh, September 2026).
A coding-intensity shift is neither a price change nor a utilization change. It raises the allowed charge on a service that did not change, so it is invisible to a per-service unit cost decomposition and lands on the trend line anyway. It is the employer-side counterpart to the revenue-optimization pressure PwC's payer panel flagged in June. Cost-sharing does not reach it either: once a member clears the deductible, the plan absorbs the upcoded difference.
Where the Buy-Down Stops Working
Cost-sharing is bounded by the out-of-pocket maximum. Below that ceiling the plan and the member split claims on whatever schedule the design sets. Above it the plan pays 100%. The wedge therefore lives in the frequency layer, and it does nothing to the layer employers themselves name first.
The International Foundation of Employee Benefit Plans surveyed 112 US employers in July and August and found 32% naming catastrophic claims as the primary driver of their 2027 increase, ahead of specialty drugs at 21% (IFEBP, August 2026); its respondents put median 2027 growth at 10%. Patel drew the same conclusion from the Marsh data: "As very expensive new therapies for cancer and rare diseases reach the market, extremely high-cost claims have become more common" (Marsh, September 2026). That driver mix is the one the site examined when the IFEBP figures landed.
Raising cost-sharing also moves the specific stop-loss corridor in a direction most renewal conversations leave unpriced. The specific deductible attaches on plan-paid dollars. When the member absorbs more of the early spend, the plan accumulates paid dollars more slowly on the same claimant, so a catastrophic case crosses attachment later in the policy year and the carrier's expected loss falls on a claim that has not changed at all. IFEBP reports medical stop-loss premiums still rising above 10% a year, which is the rate an employer pays for a corridor its own plan design just narrowed.
The cushion is finite, and part of it has already been spent in a form that cannot be spent twice. Mercer finds 5% of large employers dropping or actively considering dropping GLP-1 coverage for 2027 and 27% tightening utilization controls (Mercer, June 2026), so some of the 2.8 points is an exclusion rather than a cost-sharing change. An exclusion pays once. So does a deductible reset. Each year the buy-down is taken, more of what remains sits above the out-of-pocket maximum, where the next one has nothing left to move.
Further Reading
- Employer GLP-1 Coverage Falls to 60% as the 2027 Trend Print Lands at 9.2%: the earlier read on the same trend season, and the competing large-employer survey whose gross-to-net wedge is less than half of Marsh's.
- Catastrophic Claims Lead IFEBP's 2027 Health Cost Trend Drivers at 32%: the driver mix sitting above the out-of-pocket maximum, where a plan design buy-down has no purchase.
- Leveraged Trend on a Flat Specific Deductible: how attachment-point mechanics convert a modest gross trend into a much larger stop-loss premium move.
- Million-Dollar Claim Frequency and the Attachment Point: the severity layer that keeps compounding whatever employers do to their deductibles.
- Segal's 2027 Survey Pegs Medical Trend at 9.9%, a 15-Year High, With Rx Still Ahead at 11.5%: the supply-side manual loaded into the renewals these employers are buying down.
Sources
- Marsh: Employers Expect Health Benefit Costs to Jump 8.2% in 2027 (September 2, 2026)
- Mercer: Employers Are Shifting Health Care Costs to Employees, Survey on Health and Benefit Strategies for 2027 (June 11, 2026)
- Business Group on Health: 2027 Employer Health Care Strategy Survey, Executive Summary (August 2026)
- PwC Health Research Institute: Medical Cost Trend, Behind the Numbers 2027 (June 2026)
- International Foundation of Employee Benefit Plans: Health Care Costs Pulse Survey, 2027 Cost Trend (August 2026)
- International Foundation of Employee Benefit Plans: Stop-Loss Premiums Increase to Over 10% Annually (2026)
- Fierce Healthcare: Employers' Health Benefits Costs Could Rise 8.2% in 2027, Marsh Survey (September 2026)