US Casualty Runs 11% Ex-Comp While WC Softens: Reading Marsh's Q2 Split

US casualty rose 7% in Marsh's second-quarter 2026 rate index, but excluding workers' compensation the same book ran 11%, a four-point spread that means the headline number a filer plugs into a trend memo is understating the commercial auto, general liability, and umbrella loss-cost signal by roughly a third (Marsh, July 2026). Workers' comp is doing the dragging, not the lifting, and NCCI's own reserve data shows why that should worry a pricing desk more than it comforts one.

Marsh's Global Insurance Market Index recorded a 6% global composite decline for the quarter, the eighth straight decrease and the deepest yet, with property down 12% globally and 13% in the US, cyber down 4% for a twelfth consecutive quarter, and financial and professional lines down 3% globally (Insurance Journal, July 23, 2026). Against that backdrop, US casualty is the outlier that composite reporting keeps flattening into a single misleading digit. Marsh and Gallagher Re named the drivers as claims severity, third-party litigation funding, nuclear verdicts, and climbing defense costs (Insurance Business, July 24, 2026), the same forces that have pushed third-party litigation funding into a roughly $400 billion global industry with the deepest pool of capital concentrated in US tort claims.

The Four-Point Wedge and Where It Actually Lands

A blended +7% US casualty print sitting inside a filing memo obscures more than it discloses once you know the composition. Workers' comp is, in Marsh's own characterization, the most competitive segment of the casualty book right now, and NCCI's approved 2026 loss cost filings average roughly a 5.0% reduction across jurisdictions, with individual state filings ranging from a 15.6% decrease to a 21.6% increase depending on experience (ResourcePro, 2026). That segment pulls the blended casualty average down by roughly four points relative to the rest of the book. Strip WC out and commercial auto, general liability, and umbrella/excess are running rate at +11%, a figure that reads much closer to what the severity trend Marsh and Gallagher Re describe would actually require.

The practical filing consequence is specific: a pricing actuary who treats the +7% print as the trend signal for a commercial auto or umbrella indication is importing four points of dilution from a line whose loss-cost trajectory has nothing to do with the excess-tower severity driving the rest of the casualty book. The two segments should never have been blended into one filed indication in the first place, and the four-point gap is the clearest evidence yet that they are moving on independent claims processes.

Workers' Comp Looks Adequate Until You Read the Severity Line

NCCI's 2026 State of the Line put the workers' comp calendar-year combined ratio at 91% for 2025, well inside the industry's overall 93% mark and among the strongest results of any major line, with net written premium of $41.6 billion, down 0.2% for the year (NCCI, via Healthesystems, 2026). Lost-time claim frequency fell 2%, a more moderate decline than the long-run average, while both medical and indemnity claim severity grew 4% in the same year. The reserve position is where the trend actually shows up: NCCI now estimates the industry's redundant WC reserve position at $14 billion, down from $16 billion in 2024. A line does not lose $2 billion of reserve cushion in a single year for no reason, and the reason on the table is the same medical and indemnity severity growth that has been running the excess casualty tower.

That combination, frequency falling while severity climbs and the reserve buffer shrinks, is the profile of a line pricing off a favorable trailing average rather than a forward severity curve. WC being called Marsh's "most competitive" segment is a statement about current filed loss costs and market appetite, not a statement about whether those loss costs still hold margin against the medical and indemnity trend NCCI is reporting in the same document. A WC pricing team reading the "most competitive" label as license to cut further should instead read the $16B-to-$14B reserve slide as the adequacy flag it is: the same attorney involvement and litigated-settlement dynamics that are repricing GL and auto do not stop cleanly at the WC boundary, and a 5% average loss cost reduction moving into 2027 filings against 4% severity growth is compounding a gap, not closing one.

Global Casualty at +2% Is a US-Only Story, and That Is a Treaty Lever

Global casualty rose only 2% in the same quarter, decelerating from 3% in the first quarter, with every non-US region posting a decline (Insurance Business, July 24, 2026). The entire casualty hardening story is domestic. Gallagher Re's facultative placement data found international and UK casualty rates falling 5% to 25% in the first half of 2026, and characterized US casualty as "the clearest holdout in an otherwise buyer-driven market," noting that only four or five traditional reinsurance markets remain actively writing US casualty facultative business (Gallagher Re, 2026). That is a capacity statement as much as a pricing one: when the pool of willing facultative markets narrows to single digits, the remaining underwriters hold real leverage on attachment point and price, and casualty reinsurance buyers negotiating 1/1 treaty terms should expect the domestic-versus-international spread to be used against them directly.

A cedent with a US-heavy casualty book cannot point to benign international loss experience to argue down ceded rate at renewal, because the reinsurer's own facultative desk is watching the same regional split. The four or five markets still active in US casualty facultative business are pricing off the severity story, not the composite, and a treaty submission built around the blended +7% figure hands the reinsurer's actuaries an easy opening to reprice off the +11% ex-comp number instead. The wedge that matters at the treaty table is not property-versus-casualty; it is US-casualty-versus-everywhere-else, and it is widening.

Capacity Is Being Engineered Around the Gap, Not Repriced to Close It

The broker and MGA response to the excess/umbrella severity problem has been structural as much as it has been about price. Marsh launched MLOne on July 13, a lead umbrella casualty facility that consolidates up to $30 million of lead capacity into a single quota-share block with Allianz Commercial as sole lead insurer and sole claims handler, the third in a series of follow-form excess facilities that can be stacked with Marsh's existing BX1 and MX1 programs (Insurance Business, July 2026). A single claims handler across a consolidated capacity block is a direct response to a specific loss driver: when multiple lead insurers share umbrella layers without coordinated claims positions, conflicting reserve and settlement decisions slow resolution and, in a nuclear-verdict environment, that delay itself becomes a severity multiplier. MLOne is capacity manufactured to control the claims-handling variable, not capacity priced to absorb more severity risk.

Rate on the excess/umbrella layer itself reflects the same tension rather than a clean hardening trend. Q2 2026 middle-market placements ran flat to low double digits, with materially higher increases concentrated in poor-loss and heavy-auto accounts, while primary GL and products pricing sat flat to up 5%-10% for clean risks even as loss-cost trend in the most challenged segments ran 12% to 15% (Risk Placement Services, 2026). RPS characterized the market as bifurcated rather than uniformly hard, with new capacity from London, Bermuda, and MGA markets competing hardest on the cleaner accounts and leaving the loss-affected segment to absorb most of the rate. Some carriers have gone further and simply capped individual risk capacity around $10 million for US casualty exposures given the litigation environment, trading line size for control rather than repricing the exposure at the same limit (Insurance Business, July 2026). Every one of these responses caps how far excess and umbrella rate can run in practice: capacity that would otherwise chase the severity trend upward is instead being restructured, split into smaller lines, or routed through single-claims-handler facilities designed to blunt the litigation dynamic rather than price fully for it.

What Changes in the Next Indication

A commercial auto or GL pricing actuary building the next filed indication off Marsh's print should decompose the blend before using it: hold workers' comp trend separate from the rest of the casualty book, and anchor the excess/umbrella layer to the ex-comp 11% and the RPS-reported 12%-15% loss-cost trend in challenged segments rather than the 7% composite. A WC pricing team should treat the shrinking NCCI reserve redundancy, not the "most competitive segment" label, as the number worth watching into the next loss cost filing cycle; a fourth consecutive year of redundancy decline would be the signal that WC's own severity trend has caught up with its written rate. Casualty reinsurance buyers heading into any mid-year or 1/1 treaty conversation should expect facultative markets to lean on the US-versus-global split as a negotiating point and should be prepared to show their own book's ex-WC severity development rather than defend the blended composite. And every desk touching excess or umbrella capacity should watch whether facilities like MLOne actually hold claims-cost outcomes down over the next several accident quarters, or whether single-claims-handler structuring turns out to manage optics on delay without changing the underlying verdict exposure. NCCI's next loss cost filing cycle and Marsh's Q3 2026 index, due in October, are the next two data points that will show whether the four-point wedge is stable or still widening.