Stop-Loss Pricing Under Pressure as $1M Claims Double in a Year

From reviewing stop-loss pricing models across multiple renewal cycles, the process for calibrating specific deductible rates has relied on two assumptions that held reasonably well through the post-ACA era: that large-claim frequency above typical attachment points follows a roughly stable Poisson process, and that severity trend can be applied uniformly across the claim size distribution. The 2025 IFEBP survey data breaks both assumptions simultaneously. When million-dollar claim prevalence doubles in a single measurement period, the pricing actuary faces a distributional problem that trend factors alone cannot solve.

The Excess Loss Factor Framework

Specific stop-loss pricing starts with the excess loss factor (ELF): the ratio of expected losses above a given specific deductible d to total expected losses. If the claim size distribution is F(x) with density f(x), the ELF at attachment point d equals:

ELF(d) = ∫_d^∞ (x - d) · f(x) dx  /  E[X]

The specific stop-loss premium per employee per month then equals the ELF multiplied by the total expected claim cost PEPM, loaded for expenses, profit, and risk margin. For a group with expected total claims of $800 PEPM and a $250K specific deductible, an ELF of 0.020 produces an expected excess loss of $16 PEPM before loads.

The critical property of the ELF is its sensitivity to the tail of the distribution. A rightward shift in the severity distribution increases the ELF at every attachment point, but the percentage increase grows with the deductible level. This is the leverage effect: the same underlying distribution change produces a 25% ELF increase at $100K but might produce an 80% increase at $500K and a 120% increase at $1M. The IFEBP survey PEPM data confirms the pattern. At the $100K specific deductible, the average PEPM rose from $210.80 to $229.40 (an 8.8% increase). At $500K, the increase was 10.1%. At $1M, the increase was steeper still relative to the prior year's $13.84 base, reaching $17.69 (a 27.8% increase).

Fitting the Claim Size Distribution: Why Uniform Trend Fails

The standard approach to stop-loss pricing updates the ELF table by applying a uniform severity trend to the entire claim size distribution. If medical trend is 8% and the prior-year ELF table was derived from a lognormal with parameters μ = 10.2 and σ = 2.1, the actuary shifts μ upward by ln(1.08) = 0.077 and recomputes the excess integrals. This produces a smooth, proportional increase in the ELF at every deductible level.

That approach fails when the distribution's shape changes, not just its location. The doubling of $1M+ claim prevalence is not consistent with a location shift alone. It implies that the tail of the distribution has thickened: the shape parameter (the Pareto α or the lognormal σ) has changed, not merely the scale. Applying a uniform trend to an outdated shape parameter systematically underestimates the ELF at higher attachment points while potentially overestimating it at lower ones.

Three distributional forms are standard in stop-loss actuarial practice:

Lognormal. Tractable, closed-form excess loss integrals, and familiar to most pricing actuaries. The lognormal works well for the body of the distribution (claims from $50K to $500K) but underestimates the probability of extreme claims above $2M. Its light tail relative to observed catastrophic claim data is a known limitation, and the current environment with gene therapies priced at $2M to $4.25M per treatment makes this limitation material.

Single Pareto. Heavy-tailed by construction, the Pareto distribution is parameterized by a shape parameter α and a threshold. Lower values of α produce heavier tails. The Pareto captures extreme claim behavior well but overestimates density in the moderate-severity range ($100K to $300K), producing ELFs that are too high at lower deductible levels.

Spliced distribution. A lognormal body spliced to a generalized Pareto tail at a threshold (typically $250K to $500K) combines the tractability of the lognormal in the moderate range with the Pareto's heavy-tail behavior. The splice point is selected where the empirical hazard rate stabilizes. Maximum likelihood estimation fits the lognormal parameters below the splice and the GPD parameters above it. This approach currently offers the best fit to the observed data, particularly the combination of stable moderate-claim frequencies with rapidly accelerating extreme-claim frequencies.

The practical test: does the fitted distribution replicate the IFEBP's observed claim exceedance probabilities? If 49% of 1,268 plans (each with an average of roughly 950 covered employees) report at least one $1M+ claimant over two policy years, the implied per-member-per-year probability of a $1M+ claim can be backed out. For a spliced model, the GPD tail parameters should be re-estimated from the most recent two years of data rather than trended forward from an older vintage. The Sun Life dataset (65,000+ claims across 3,000 employers) provides an independent calibration source showing $1M+ claims per million covered lives up 61% over four years.

What Is Driving the Tail Shift

The claim drivers producing the distributional change are not uniform in their severity profiles, which is precisely why a single trend factor fails to capture them:

• Cancer (92% of catastrophic claims): CAR-T cell therapies at $373K to $475K wholesale acquisition cost, with total episode costs exceeding $1M. Checkpoint immunotherapy (Keytruda at roughly $200K/year at commercial rates) produces multi-year accumulations that breach specific deductibles at $500K and above. Cancer's dominance has increased from 83% to 92% of catastrophic claims in one survey cycle.
• Specialty pharmacy (47% of catastrophic claims): Biologic infusions, orphan drugs, and specialty injectables with annual costs from $100K to $500K. These drive claim counts above lower deductible levels but less frequently produce the $1M+ claims reshaping the far tail.
• Gene and cell therapies: With over 60 CGTs expected to receive FDA approval by 2030, individual treatment costs from $2.2M (Casgevy) to $4.25M (Lenmeldy) create single-claim severity that did not exist in the pricing data five years ago. The IFEBP survey found gene therapy claims remain rare in carrier portfolios (only three claims reported across nearly 2 million covered lives between 2022 and 2024), but each claim is large enough to dominate an entire plan's stop-loss experience.
• GLP-1 medications: At $15K to $25K annually per patient at gross cost, GLP-1s fall below the specific deductible threshold. Their impact flows through aggregate stop-loss and total plan cost, not specific deductible pricing. The CMS Medicare GLP-1 Bridge launching July 2026 (with a $50 copay and $245 net manufacturer price) may expand utilization norms that eventually spill into the commercial self-funded market.

Each category has a different development pattern and trend credibility. Cancer severity trends are supported by 15+ years of escalating treatment cost data. Gene therapy frequency is too sparse for credible trend estimation and must be modeled as a shock load. GLP-1 is a volume play on aggregate exposure. A pricing model that applies a single trend factor to all four categories will misprice two or three of them.

Aggregate Stop-Loss: How Specific Recoveries Reshape the Corridor

The specific-aggregate interplay is the second pricing problem exposed by the frequency shift. Aggregate stop-loss covers total plan claims exceeding a corridor above expected claims (typically 120% to 130% of expected). The aggregate attachment factor equals the corridor percentage multiplied by the expected claim cost net of specific stop-loss recoveries.

When a plan purchases specific coverage at a $250K deductible, any individual claim above $250K is partially or fully reimbursed by the specific policy. Those recoveries reduce the variance of the plan's retained claims. The aggregate corridor is then set based on the distribution of net retained claims: total claims minus specific reimbursements.

Higher specific claim volatility changes this calculation in two ways. First, more frequent specific deductible breaches increase the expected specific recoveries, which reduces the variance of retained claims and, all else equal, should narrow the aggregate corridor needed for a given loss ratio target. Second, the same frequency shift that increases specific claims also increases the plan's total expected claims. If the aggregate corridor is expressed as a fixed percentage of expected, the absolute dollar attachment rises with expected claims, but the underlying distribution of retained claims has shifted in ways the fixed percentage may not capture.

The actuarial methodology for deriving the aggregate corridor starts with the expected claim distribution. After netting specific recoveries:

Net claims = Total claims - Σ max(claim_i - d, 0)

The distribution of net claims has lower variance than total claims, with the reduction depending on the specific deductible level. A lower specific deductible removes more variance from the aggregate layer, producing a tighter net claim distribution and allowing a narrower aggregate corridor for equivalent aggregate premium. A higher specific deductible leaves more variance in the aggregate layer, requiring a wider corridor or higher aggregate premium.

The Specific-Aggregate Optimization

This creates a practical optimization problem for plan sponsors and their actuaries. Lowering the specific deductible increases specific premium but reduces aggregate volatility and aggregate premium. Raising the specific deductible reduces specific premium but increases aggregate exposure. The total cost of stop-loss protection is the sum of specific and aggregate premiums, and the minimum-cost combination depends on the plan's size, claim distribution, and risk tolerance.

Specific Deductible	Specific PEPM	Aggregate Corridor	Aggregate PEPM (est.)	Total Stop-Loss PEPM
$100,000	$229.40	125%	$18 - $25	$247 - $254
$250,000	$68.90	125%	$30 - $42	$99 - $111
$500,000	$50.96	125%	$38 - $55	$89 - $106
$1,000,000	$17.69	125%	$50 - $70	$68 - $88

The table illustrates the tradeoff using IFEBP 2025 specific PEPM data with estimated aggregate costs. At a $100K specific deductible, the plan pays $229 PEPM in specific premium but faces minimal aggregate exposure because most large-claim variance is already covered. At a $1M specific deductible, the specific premium drops to $18 PEPM, but aggregate exposure rises substantially because claims between $100K and $1M remain in the retained layer and contribute to aggregate volatility.

For a 1,000-employee group, the annual cost difference between the lowest specific deductible ($247 to $254 PEPM, roughly $3.0M annually) and the highest ($68 to $88 PEPM, roughly $0.9M annually) is significant. But the $1M specific deductible leaves the plan exposed to multiple claims in the $200K to $900K range that could individually be manageable but collectively breach the aggregate attachment. In the current environment, where claim frequency above $250K is elevated, the aggregate risk at high specific deductibles is larger than historical models suggest.

Market Pricing Benchmarks: Rate Adequacy in 2026

The stop-loss market has grown to $35.4 billion in annual premium, a 12% increase from the prior year. Segal's 2025 national dataset across 221 health plans found an average premium increase of 9.7% for plans maintaining comparable coverage. The IFEBP survey confirmed increases of 8.8% at $100K to 10.1% at $500K. Meanwhile, 67% of covered workers are now enrolled in self-funded plans (80% at large firms), expanding the addressable market for stop-loss coverage.

Carrier-level pricing dispersion is wide. Voya pushed average increases of 21% or higher for 2025 renewals after experiencing elevated claims. Sun Life raised rates 17% while simultaneously growing sales 56% as competitors exited. Cigna's CFO cited specialty drugs and cancer surgery claims as drivers of stop-loss renewal increases. Several of the largest stop-loss writers, including Cigna, Voya, and Sun Life, experienced poor Q4 2024 claims driven by advanced cancer treatments, premature births, and health system revenue optimization.

The spread between the modal market increase (8% to 10%) and the ELF-indicated increase (25% to 120% depending on deductible level) reflects multiple factors: expense ratio dampening (if expenses are 25% of premium, a 29% pure loss increase produces a 22% indicated premium increase), multi-year rate smoothing, and the credibility lag in pooled manual rates that pre-date the frequency shift. Carriers pricing at the low end of the range are likely running on pooled data from 2022 and 2023 policy years that do not yet contain the structural acceleration. The risk is concentrated in the tail: carriers that have not re-estimated their distributional shape parameters may find themselves underreserved at higher attachment points precisely where the leverage effect is largest.

Why This Matters for Pricing Actuaries

The stop-loss pricing challenge in 2026 reduces to three technical imperatives:

Re-estimate the distribution, not just the trend. Fitting a spliced lognormal-GPD model to the most recent two years of claim data, rather than trending a stale lognormal forward, is the minimum response to a distributional regime change. The Sun Life dataset (65,000+ claims across 3,000 employers) and the IFEBP survey (1,268 plan sponsors covering 1.2 million employees) provide the sample sizes needed for credible parameter estimation in the excess layer.

Re-derive deductible relativities from the updated distribution. Carrying forward prior-year ELF relativities when the shape parameter has changed produces systematic mispricing. The percentage increase in the ELF should be larger at higher deductible levels; if your renewal rate increases are roughly flat across deductible levels, your relativities are stale.

Model specific and aggregate jointly. The frequency shift in specific claims changes the net retained claim distribution that drives aggregate pricing. Actuaries optimizing total stop-loss cost for plan sponsors should model the joint distribution of specific and aggregate exposure, not price each layer independently. Plans raising their specific deductible to save on specific premium may be taking on more aggregate risk than the aggregate corridor reflects.