Umbrella Pricing Holds at 9.4% as Other Commercial Lines Soften: Why ILFs Need Re-Derivation

From tracking commercial lines rate filings over the past several renewal cycles, one pattern keeps reasserting itself: umbrella and excess liability pricing refuses to follow the rest of the market downward. The Ivans Index for Q1 2026, reported by Insurance Journal on April 27, 2026, shows umbrella renewal rates at +9.36%, down a mere 13 basis points from Q4 2025’s +9.49%. Every other major commercial line softened more aggressively over the same period: general liability fell to +6.85% from +7.23%, commercial property dropped to +6.83% from +8.01%, commercial auto declined to +5.28% from +6.97%, and business owners policies eased to +6.74% from +7.52%.

This is not a statistical fluke. Umbrella and excess liability sit at the top of the insurance tower, where they absorb the tail of the claim severity distribution. When nuclear verdicts exceeding $10 million rose 27% in 2023 and thermonuclear verdicts above $100 million jumped 35% (Allianz Commercial, Liability Claims Report 2024), the tail got fatter. Traditional Increased Limits Factors derived from historical severity distributions no longer reflect the cost of excess exposure. The result is a pricing divergence that will persist until the underlying severity parameters stabilize, and there is no evidence of that happening.

The Rate Divergence in Context

The gap between umbrella and primary lines is widening each quarter. In Q4 2024, umbrella rates stood at roughly +10.5% while GL was closer to +8%. By Q1 2026, the spread has expanded to 251 basis points (umbrella at +9.36% versus GL at +6.85%). Workers’ compensation, the one line with truly declining frequency, fell further to -1.73% in Q1 2026.

This divergence makes structural sense. Primary-layer pricing responds primarily to frequency trends, which remain manageable across most casualty lines. Excess layers respond to severity trends, particularly in the right tail of the loss distribution. When a single verdict can exceed $1 billion (five such verdicts occurred in 2024 alone, per Marathon Strategies), the severity tail drives all the pricing action, and that action concentrates in the excess stack.

Characterizing the commercial lines market with a single cycle descriptor misrepresents what is actually happening. The primary market is softening. The excess market is firming, or at minimum holding. These are two distinct pricing environments governed by different loss drivers, and treating umbrella as just another casualty line understates the structural repricing underway.

Nuclear Verdicts Are Reshaping the Severity Distribution

The data from Marathon Strategies’ “Corporate Verdicts Go Thermonuclear” 2025 edition quantifies what pricing actuaries have felt in their loss picks for several years. In 2024, 135 nuclear verdicts exceeded $10 million, a 52% increase from 2023. Total nuclear verdict value reached $31.3 billion, a 116% jump year over year. Of those, 49 exceeded $100 million, and five surpassed $1 billion.

Texas led with 23 nuclear verdicts, followed by California at 17 and Pennsylvania at 12. Product liability accounted for 32 nuclear verdicts totaling $13.9 billion, making it the dominant line by aggregate severity. The Bayer Roundup litigation alone produced a $2.2 billion verdict in Philadelphia, with separate awards of $175 million and $78 million in other jurisdictions.

Allianz Commercial’s Liability Claims Report provides the longer-term context. Product liability severity grew 20.4% over the decade from 2013 to 2023, while economic inflation averaged just 2.7% annually over the same period. That 7.6x multiplier between actual severity growth and economic inflation represents pure social inflation, and it concentrates disproportionately in the excess layers. Total U.S. commercial casualty losses reached $143 billion in 2023, exceeding the $108 billion in global natural catastrophe losses for the same year.

The critical point for pricing actuaries: this is not just a rightward shift of the severity distribution. The tail is getting heavier. The shape of the distribution itself is changing, which means the parameters governing Increased Limits Factors need re-estimation, not just trend adjustments to existing factors.

ILF Derivation: How a Falling Alpha Reprices Excess Layers

Increased Limits Factors quantify the relative cost of providing coverage at limits above the basic limit (typically $100,000 or $300,000 for GL). The standard actuarial approach fits a severity distribution to closed-claim data, then calculates ILFs as the ratio of limited average severity at the higher limit to limited average severity at the basic limit:

ILF(L) = E[min(X, L)] / E[min(X, B)]

where X is the claim severity random variable, L is the coverage limit, and B is the basic limit. The most commonly used severity model for liability lines is the single-parameter Pareto distribution, where the survival function is S(x) = (B/x)^α for x ≥ B. The shape parameter α controls the tail weight: lower α means a heavier tail and higher ILFs at elevated limits.

Consider what happens when nuclear verdict frequency inflates the tail. Using a simplified Pareto model with basic limit B = $100,000:

• At α = 1.5 (a historical estimate consistent with pre-2015 GL severity data), the ILF at $1 million is approximately 3.16 and at $5 million is approximately 5.85.
• At α = 1.2 (reflecting the heavier tail from recent nuclear verdict experience), the ILF at $1 million rises to approximately 3.98 and at $5 million climbs to approximately 10.0.

That shift from α = 1.5 to α = 1.2 increases the $5 million ILF by roughly 71%. At the $10 million attachment point, the effect is even more dramatic, with the ILF approximately doubling. The mathematical leverage of excess layers means that small changes in the severity shape parameter produce large changes in excess loss costs. A 0.3-point decline in the Pareto α can translate into a 30-40% increase in the layer cost at the $5 million limit, compounding further at higher attachment points.

The practical challenge is that ISO advisory loss costs for GL, which form the starting point for primary-layer pricing at most carriers, embed ILFs that are updated less frequently than the underlying verdict severity trends change. If advisory ILFs reflect an α estimated from data that predates the current nuclear verdict acceleration, the basic-limits loss cost may appear adequate while the excess exposure is systematically underpriced.

Loss Development Compounds the Problem

Excess layer pricing is not just an ILF problem. It is also a development problem. High-layer losses develop over longer reporting periods than primary losses. A $50 million product liability verdict may not appear in the triangle until five or more years after the accident date, and social inflation may be accelerating development factors in later maturity years.

Patterns we have observed in recent Schedule P filings suggest that casualty loss development factors at 60-to-ultimate and 72-to-ultimate months are increasing, particularly for GL and commercial auto. The industry posted $15.8 billion in adverse prior-year casualty development in 2024 (Milliman), and CNA alone booked $106 million in unfavorable development in Q1 2026, split between excess casualty ($56 million) and professional E&O ($50 million). These development pattern changes interact with the ILF shift: if the tail is both heavier and slower to develop, IBNR for excess layers must increase on both dimensions simultaneously.

For the pricing actuary, this means a trend load applied at the basic limit understates the trend needed at excess attachment points. A 5% annual severity trend at the $100,000 basic limit can produce 8-12% trend at the $5 million excess layer, depending on the Pareto α. This leverage effect arises because trend applied to the entire severity distribution shifts more probability mass above higher attachment points when the tail is heavy. The lower the α, the greater the leverage.

Credibility and the Sparse-Data Problem Above $10 Million

Even for large national carriers, claim counts above $10 million are sparse enough to fail traditional actuarial credibility thresholds. A carrier writing $500 million in GL premium might see 200-300 claims per year exceeding $100,000 but only 3-5 exceeding $10 million, producing a complement of credibility that heavily weights industry data or fitted distribution assumptions.

This is where the choice of severity model becomes load-bearing. A lognormal distribution, which has a lighter tail than the Pareto, will produce lower ILFs at high limits and may systematically understate excess layer costs in the current environment. A mixed exponential-Pareto model, which allows the body and tail of the distribution to follow different functional forms, provides more flexibility but requires careful selection of the splice point where the Pareto tail begins.

Bayesian and hierarchical modeling approaches offer a path forward. By pooling individual-carrier experience with industry-wide verdict data (such as the Marathon Strategies nuclear verdict database or Advisen large-loss data), pricing actuaries can stabilize excess layer estimates without relying solely on an individual carrier’s thin large-loss experience. The CAS Variance journal’s treatment of NCCI excess loss factor methodology provides a framework for this kind of credibility-weighted severity modeling, though adapting it to the GL and umbrella context requires adjustments for the different loss development patterns and the absence of a single advisory body comparable to NCCI in workers’ compensation.

Capacity Constraints Fragment the Pricing Framework

The severity shift has real consequences for how umbrella programs are structured. Capacity above $25 million has become constrained, forcing layered towers that involve multiple carriers, including E&S writers, each applying different rating methodologies to their slice of the program. Amwins’ 2026 State of the Market report notes that construction casualty excess underwriters are seeking rate increases of 7-15%, with spikes exceeding 20% for distressed risks. High excess layers are seeing flat-to-single-digit rate movement, but only because capacity competition at those heights is limited enough that carriers can maintain pricing discipline.

The absence of a single advisory framework for excess layers creates pricing inconsistency across the tower. Primary GL pricing starts from ISO advisory loss costs, which provide a common benchmark. Above the primary layer, each carrier develops its own loss costs, its own ILFs or layer-specific factors, and its own severity trend assumptions. Two carriers writing the $10 million xs $15 million layer on the same account may be using materially different severity parameters, producing a wide range of indicated rates that the broker arbitrages.

This fragmentation is self-reinforcing. As carriers pull back capacity at elevated attachment points, the remaining markets gain pricing power but lose volume credibility. The E&S carriers that fill the gap bring different data sources and different regulatory constraints (no rate filing requirements in most states), further widening the methodology spread across the tower.

Why This Matters for Pricing Actuaries

Sources

1. Insurance Journal, “Commercial Lines Rates Continue to Soften, Says Ivans Index,” April 27, 2026 - insurancejournal.com
2. Allianz Commercial, “Liability Claims Trends Report,” 2024 - commercial.allianz.com
3. Marathon Strategies, “Corporate Verdicts Go Thermonuclear, 2025 Edition,” via Insurance Journal, May 2025 - insurancejournal.com
4. Amwins, “State of the Market: 2026 Outlook,” 2026 - amwins.com
5. Risk & Insurance, “Nuclear Verdicts Drive Rising US Liability Claims,” November 2024 - riskandinsurance.com
6. CAS Variance Journal, “NCCI’s 2014 Excess Loss Factors,” referenced for ILF methodology
7. Milliman, “U.S. Casualty Insurance 2024 Financial Results,” 2025 - milliman.com
8. AM Best, “Market Segment Outlook: US Commercial Lines,” 2026

Further Reading

• Social Inflation and Litigation Trends 2026: The $529B Casualty Challenge
• How Social Inflation Distorts Casualty Loss Development Factors
• Casualty Reserves Show Cracks Across 2021-2024 Accident Years
• Reserve Adequacy in a Softening Market: The 2026 Playbook
• CNA Q1 2026 Casualty Reserve Charge Flags Soft-Cycle Risk