Stop-Loss Actuaries Are Working With a Broken Frequency Baseline

From tracking stop-loss renewal data across multiple survey cycles, the frequency of claims above any given specific deductible level moved in a narrow band for most of the post-ACA era. Annual variation existed, but structural shifts were rare. The 2025 Aegis Risk survey breaks that pattern. A doubling of million-dollar claim prevalence in a single measurement period is not noise; it is a regime change in the severity distribution that flows directly through the excess loss factor calculation and into the specific deductible charge at every attachment point.

The Frequency Data: What the Aegis Risk Survey Shows

The 2025 Aegis Risk Medical Stop-Loss Premium Survey, published by the International Foundation of Employee Benefit Plans (IFEBP) and covering 1,268 plan sponsors representing over $1.2 billion in annual stop-loss premium, provides the most comprehensive year-over-year frequency benchmark available for high-cost medical claims in the self-insured market. The survey spans group sizes from 17 to over 21,000 employees, providing broad market representation.

The headline finding: 49% of responding employers reported at least one claimant exceeding $1 million in paid claims over the most recent two policy years. That figure was 23% in the 2024 survey. Additionally, 16% of respondents reported at least one claimant exceeding $2 million, and survey presenters characterized the current environment with the phrase "$3 million is the new $1 million."

Sun Life's 2025 High-Cost Claims Report, drawing from $15.2 billion in medical costs across 3,000 employers and over 65,000 high-cost claims, confirms the acceleration. Million-dollar stop-loss claims increased 29% year-over-year on a per-million-covered-employees basis and are up 61% over the past four years. These are two independent large-sample datasets pointing to the same structural shift.

The Aegis Risk survey also captured stop-loss premium rates by specific deductible level. The average PEPM was $229.40 at the $100K specific deductible, $68.90 at $250K, and $17.69 at the $1M level. Premium increases for the most recent policy year: 8.8% at the $100K deductible, 10.1% at $500K. These rate changes matter because the relationship between them reveals whether the market has correctly priced the leverage effect of the underlying severity shift.

The Excess Loss Factor Methodology

Stop-loss specific deductible pricing is fundamentally an excess loss calculation. The specific stop-loss charge equals the expected cost of claims in excess of the attachment point, loaded for expenses, profit, and risk. The core actuarial quantity is the excess loss factor (ELF): the ratio of expected losses above the specific deductible to total expected losses (or to some reference base like standard claim cost).

For a given claim severity distribution f(x), the expected excess loss above deductible d is:

E[excess | d] = ∫_d^∞ (x - d) · f(x) dx

The ELF at deductible d equals this expected excess divided by the total expected claim cost E[X]. When the severity distribution shifts rightward, both the probability of exceeding d and the expected excess given exceedance increase. The ELF rises at every deductible level, but it rises by a larger percentage at higher deductibles. This is the leverage effect.

Consider a simplified example. Suppose the baseline severity distribution produces these excess loss factors:

Specific Deductible	Baseline ELF	Shifted ELF	% Change in ELF
$100,000	0.045	0.058	+29%
$250,000	0.018	0.027	+50%
$500,000	0.007	0.013	+86%
$1,000,000	0.0025	0.0055	+120%

The "shifted" column reflects a severity distribution where the probability of claims exceeding $1M has approximately doubled, consistent with the Aegis Risk survey data. The leverage effect is visible in the rightmost column: the percentage increase in the ELF grows with the deductible level. At $100K, the ELF increases roughly 29%. At $500K, it increases roughly 86%. At $1M, it more than doubles.

The Leverage Gap: Market Rates vs. the ELF Signal

Now compare the illustrative ELF shifts to the observed market premium increases: 8.8% at the $100K deductible and 10.1% at the $500K deductible. If the ELF has increased by approximately 29% at $100K and 86% at $500K, the market rate increases are capturing roughly 30% of the indicated change at $100K and only 12% at $500K.

The gap between the rate increase and the ELF shift is not entirely rate inadequacy. Expense loads dilute the pure loss rate increase, multi-year rate smoothing absorbs some of the volatility, and not all carriers price off the same pooled data vintage. But even accounting for expense ratio dampening (if expenses are 25% of premium, a 29% loss rate increase translates to roughly a 22% indicated premium increase, not 8.8%), the market appears to be underloading the tail shift at higher deductible levels.

Segal's 2025 national dataset across 221 health plans found an average stop-loss premium increase of 9.7%, broadly consistent with the Aegis Risk figures. Voya, by contrast, pushed rate increases averaging 21% or higher for 2025, and Sun Life raised stop-loss rates 17% for 2026 while simultaneously growing sales 56% as competitors exited. The carrier-level dispersion is wide, but the modal increase in the market (8% to 10%) sits below what the ELF mathematics indicate.

Why Market Rates Lag: ASOP No. 25 and the Pooled Data Problem

The structural reason market rates underreact to a sudden frequency shift is credibility weighting. ASOP No. 25 (Credibility Procedures) governs how stop-loss actuaries blend group-specific experience with manual rates. The standard credibility formula assigns weight Z to subject experience and (1-Z) to the manual rate: Estimate = Z * X + (1-Z) * M.

For most self-insured employers purchasing stop-loss coverage, group-specific experience above a $250K or $500K specific deductible produces very few claims per year. A 500-life employer with a $250K attachment might generate 2 to 5 claims above the deductible annually, yielding a credibility weight near 0.10 to 0.20. That means 80% to 90% of the renewal rate derives from the manual rate, which is constructed from pooled industry data.

The manual rate is the problem. Pooled datasets used to build stop-loss manual rates typically reflect experience from policy years that ended 12 to 24 months before the current pricing date. If the structural shift in million-dollar claim frequency occurred primarily during the 2024 and 2025 policy years, pooled data from 2022 and 2023 policy years will not contain it. The manual rate is stale by construction, and ASOP No. 25 credibility mechanics concentrate nearly all pricing weight on that stale manual rate for groups too small to produce credible excess experience on their own.

This creates a market-wide underreaction. No individual carrier is ignoring the data; each is applying credibility procedures correctly. But the credibility complement itself lags the true current severity distribution, and the lag is longer for higher deductible levels where individual group credibility is lowest. The carriers with the most aggressive rate increases (Voya at 21%, Sun Life at 17%) are likely applying prospective judgment loads on top of the credibility-weighted manual rate, while carriers pricing strictly off the blend are systematically underloaded.

What Is Driving the Frequency Shift

Cancer treatment costs are the dominant driver. The Aegis Risk survey found cancer at 92% of catastrophic claims, up from 83% in prior surveys. The shift reflects three specific cost escalators in oncology treatment that push claims above the $1M threshold:

• CAR-T cell therapy: Wholesale acquisition cost of $373,000 to $475,000 per treatment, with total cost of care (including hospitalization and adverse event management) averaging over $700,000 and frequently exceeding $1 million per patient.
• Checkpoint inhibitors (immunotherapy): Keytruda runs approximately $200,000 per year at commercial rates, with patients continuing treatment until disease progression. Multi-year treatment courses routinely breach $500K and reach $1M+.
• Gene therapies: Zolgensma at $2.3 million, Hemgenix at $3.5 million, Casgevy at $2.2 million, and Lenmeldy at $4.25 million represent single-claim severity that was essentially nonexistent in the stop-loss pricing data five years ago.

Critically, this frequency shift is separate from GLP-1 drug cost pressures. GLP-1 medications (semaglutide, tirzepatide) run $15,000 to $25,000 annually at gross cost, well below the $100K specific deductible floor. GLP-1 affects aggregate stop-loss and total plan cost, but it does not drive the million-dollar claim frequency acceleration. The two risks require different pricing treatments: GLP-1 is an aggregate trend issue, while the catastrophic cancer and gene therapy shift is a specific deductible tail issue.

Why This Matters for Stop-Loss Pricing Actuaries

The pricing adequacy question has immediate practical implications for actuaries filing and reviewing stop-loss rates in 2026:

Manual rate vintaging. Any manual rate built from pooled data ending before policy year 2024 is structurally underloaded for the new catastrophic claim frequency. Actuaries should document the data vintage of their manual rate base and apply an explicit prospective load if the base does not yet incorporate the frequency shift visible in the Aegis Risk and Sun Life data.

Deductible relativities. If the manual rate at $100K is updated but the relativities between deductible levels are held constant from prior years, higher deductible levels will be disproportionately underpriced. The leverage effect requires that deductible relativities be re-derived from the shifted severity distribution, not carried forward from the prior year's ELF table. A 10.1% increase at $500K versus 8.8% at $100K implies the market has barely adjusted relativities, when the ELF mathematics suggest the $500K rate should be increasing at two to three times the rate of the $100K rate.

Renewal adequacy at the group level. For individual group renewals, the credibility-weighted rate will understate the true expected cost at higher deductible levels for the reasons described above. Actuaries and underwriters pricing renewals at $500K+ specific deductibles should consider whether a prospective frequency adjustment (outside the standard credibility blend) is warranted, documenting the justification under ASOP No. 25's provisions for professional judgment when the complement of credibility is itself in transition.

Reinsurance implications. Stop-loss carriers purchasing their own specific excess reinsurance face the same leverage problem one layer up. Summit Re has noted secondary reinsurers beginning to exit the market due to increasing severity and frequency of gene therapy claims. The Aegis Risk data suggests that stop-loss carriers themselves may be underreserved for the current policy year at higher attachment points, particularly carriers using retained manual rates that predate the frequency shift.