Global commercial insurance rates fell 6% in the second quarter of 2026, the eighth straight quarterly decline, and every major line joined the retreat except one: casualty, up 2% globally and 7% in the US, the sole line Marsh's Global Insurance Market Index still classifies as hardening (Marsh, July 2026).

The site's earlier read of this release chained the property and casualty prints into two-year cohort trends and a follow-up isolated the 11% ex-comp casualty figure against NCCI's workers compensation reserve position. Both pieces treated casualty and property as the story. That framing misses two lines that moved for reasons neither piece covered: US directors and officers rates turned positive for the first time since 2025, and cyber posted its twelfth consecutive quarterly decline against a loss ratio that AM Best now puts at its highest level since the pandemic-era ransomware surge. Read line by line rather than composite by composite, Marsh's Q2 print is less a single soft-market story than four separate rate-adequacy trajectories running at once, and only one of them, casualty, is still being priced to keep pace with its own severity trend.

Four Lines, Four Trajectories in One Quarter

The Global Insurance Market Index nets renewal-level, year-over-year rate changes across regions and lines into a single composite, and the composite for Q2 2026 fell 6% after a 5% decline in Q1, the steepest print of an eight-quarter run (Marsh, July 2026). Underneath it, the four major commercial lines split into two pairs moving in opposite directions and at markedly different speeds.

LineGlobal Q2 2026US Q2 2026Prior quarter (US)
Property-12%-13%-9% (Q1)
Casualty+2%+7% (+11% ex-comp)+9% (Q1)
Financial & professional lines-3%+1% (D&O)-3% D&O (Q1)
Cyber-4%-2%-2% (Q1)

Property is still the steepest line and still accelerating, from 9% down in the first quarter to 12% down in the second (Insurance Journal, July 2026). Cyber's 4% global decline extended a run that now stretches back three years, to a market that was still hardening in mid-2023. Casualty decelerated slightly, from a 3% global increase in the first quarter to 2% in the second, but it remains the only line with a positive sign attached to it, and the US book, where casualty rose 7% overall and 11% excluding workers compensation, is doing nearly all of that lifting (Insurance Journal, July 2026). Financial and professional lines fell 3% globally even as the US segment moved the other way: D&O rates there rose 1% after a 3% decline in the first quarter, the first positive US D&O print since the market began softening (Marsh, July 2026). John Donnelly, Marsh's president of global placement, framed the pattern as underwriters competing on more than headline price: insurers are increasingly "seeking to differentiate themselves" through broadened coverage and improved terms as capacity chases the softening lines (Marsh, July 2026).

A single 6% composite cannot carry that spread. It nets a 12-point property decline, a 4-point cyber decline, and a 2-point casualty increase into one number, and by the time D&O's 1-point US turn is folded in, the composite has stopped describing any actual line of business. The adequacy question, whether current rate is keeping pace with the loss cost it is meant to fund, has to be asked line by line, because the four lines are answering it in opposite directions.

The D&O Turn: A Small Print With a Specific Trigger

US D&O's move from a 3% decline in the first quarter to a 1% increase in the second is a four-point swing, small in isolation but the first positive print the line has recorded since the broader financial-lines market began softening. Marsh attributed the shift to underwriters becoming more selective as market conditions stabilized following years of steady reductions (Marsh, July 2026), language that describes the tail end of a soft cycle rather than a hard-market repricing. The securities litigation data explains why underwriters had reason to stop cutting.

Federal securities class action filings ran to 118 in the first half of 2026, a pace that projects to roughly 236 for the full year, 15% above the 205 filings recorded in 2025 and the highest annual total since 2020 (The D&O Diary, July 2026). Settlement values moved with the filings: the median settlement in the first half of 2026 ran $18 million to $20 million, above the full-year 2025 median of $17.6 million and well above the 2017-2025 median of $13 million, while four mega-settlements of $100 million or more accounted for roughly a tenth of the half's settlement value (The D&O Diary, July 2026). Artificial intelligence has become its own filing category rather than a niche one: 18 AI-related securities suits were filed in the first six months of 2026, already ahead of the 17 filed in all of 2025.

Set against that docket, a 1% US rate increase reads as underwriters catching up to severity rather than getting ahead of it. Financial-lines actuaries carry more exposure to this gap than the headline suggests, because D&O and its neighboring financial lines were priced down through 2023, 2024, and most of 2025 while filing volume and settlement values were already climbing. The accident years written during that trough are the ones now developing against a severity trend the Q1 2026 rate level had not yet acknowledged. A single positive quarter does not retroactively fix the pricing on business already on the books; it flags that the trough is likely behind the line and that reserve reviews on 2023-2025 D&O and side-A/side-B vintages should weight the litigation trend more heavily than the written rate history from those years would suggest on its own.

Cyber's Twelfth Straight Decline Against a Climbing Loss Ratio

Cyber rates fell 4% globally in the second quarter, the twelfth consecutive quarterly decline, a run that started when the line was still hardening in 2023 (Insurance Journal, July 2026). US cyber rates fell 2%, flat with the first quarter's pace (Marsh, July 2026). Unlike casualty and D&O, nothing in the Q2 index suggests the cyber decline is decelerating, and that persistence is what makes the loss-ratio data worth reading against it rather than in isolation.

The industry-wide cyber insurance loss ratio reached 53.0 in 2025, its highest level since the pandemic-era ransomware surge, with surplus lines carriers running a 55.9 incurred loss ratio against 50.2 for admitted carriers (Risk & Insurance, July 2026, citing AM Best). Total direct cyber premium reached $7.5 billion in 2025, up from $7.1 billion in 2024, but the growth traced largely to one carrier's book transfer rather than organic rate or exposure expansion. "The gap is actually narrower than last year," Christopher Graham, AM Best's senior industry analyst, said of the surplus-versus-admitted split, "but consecutive years of a higher incurred loss ratio suggest surplus lines carriers may be writing different business than admitted carriers, particularly business prone to a longer tail for which losses take longer to settle" (Risk & Insurance, July 2026).

The arithmetic behind a 53 loss ratio is not yet alarming on its own. Cyber's distribution-heavy, MGA-fronted structure typically carries an expense ratio in the mid-to-high 30s once commissions, policy acquisition costs, and reinsurance cede are counted, which puts an all-in combined ratio somewhere in the high 80s to low 90s, still comfortably underwriting profit. What twelve consecutive quarters of rate decline changes is the trajectory, not the current snapshot. Each further quarter of falling premium against a claims environment that AM Best already flags as lengthening in tail pushes the loss ratio higher through the denominator alone, before any change in claim frequency or severity. A line can be profitable on this year's earned premium and still be pricing itself toward an unprofitable one two or three years out if the rate decline keeps outrunning the improvement in underlying cyber risk quality that cheaper premium is supposed to reflect. Reserving actuaries on cyber books should be watching the surplus-lines-versus-admitted split specifically, since that gap is where the longer-tail, harder-to-reserve business Graham described is concentrated.

Casualty's Own Number: Adequate, or Just Less Inadequate

Casualty is the line already covered in depth elsewhere on the site, so the point worth adding here is narrow. On Marsh's own Q2 2026 earnings call, CEO John Doyle described excess casualty pricing as "still up in the mid-teens, which is really a reflection of the very challenging litigation environment" in the US (Marsh, Q2 2026 earnings call, July 2026). That figure sits well above both the blended US casualty print of 7% and the ex-workers-comp print of 11%, and the gap matters for how those two more commonly cited numbers should be read. If excess layers, the part of the tower most exposed to nuclear verdicts and litigation funding, are being priced mid-teens while the primary and blended layers move single digits, then the blended 7% is not a market-wide adequacy signal so much as an average of a layer keeping pace with severity and layers that are not. An actuary using the headline 7% as evidence that US casualty pricing has caught up to its own trend is averaging across that split without seeing it.

The Regional Spread Inside a Single Global Number

The 6% global composite also compresses a wide regional range that rate-adequacy analysis by line tends to skip past. India, the Middle East and Africa led all regions with a 16% composite decline, followed by the Pacific at 13%, Latin America and the Caribbean at 9%, the UK at 8%, Canada at 7%, Europe at 6%, and Asia at 5%; the US, down 2%, was the shallowest market in the index (Marsh, July 2026). Every region posted a decline this quarter, so the regional spread is a matter of degree rather than direction, but a 14-point range between the steepest and shallowest region is wide enough to change how a multi-territory program or a global quota share treaty should be read.

A cedant or reinsurer pricing a global casualty or financial-lines quota share on the blended composite is implicitly assuming the book's territorial mix looks like the index's, which it rarely does. A treaty weighted toward IMEA or Pacific-region property is riding a market softening twice as fast as the US, while a book concentrated in US casualty and financial lines is exposed to the two segments where rate is still catching up to loss cost rather than running ahead of it. The dispersion is the reason a single global rate-on-line adjustment applied across a multi-territory treaty will misstate adequacy in both directions at once, understating margin erosion where the local market is softening fastest and overstating it where, as in the US, price is still being pushed up by claims experience the composite does not show.

Reading the Split Rather Than the Composite

Put the four lines side by side and the pattern is not "rates are falling" or "rates are rising." It is that price is tracking loss cost closely only where loss cost is visibly, immediately painful, US casualty and, to a smaller and more recent degree, US D&O, and is decoupling from loss cost everywhere capacity has room to compete, property globally and cyber for three straight years. That is a familiar cycle mechanic, but the D&O data gives it a specific, dated marker this quarter: a line does not need years of adverse development to start repricing, it needs a filing count and a settlement median that underwriters can see moving in real time, and 2026's securities litigation numbers gave D&O underwriters exactly that. Cyber has not yet had its equivalent trigger; a 53 loss ratio is a warning line on a chart, not a filing count an underwriter has to respond to at the next renewal. Property's version of that trigger, per Marsh's own placement leadership, is contingent on a storm season that has not yet delivered one this cycle.

The practical read for pricing and reserving actuaries is to treat this index as four separate adequacy signals rather than one. Casualty pricing is closing a gap that excess layers show is still mid-teens wide. D&O pricing has just started closing a gap the litigation data suggests has been open since at least 2023. Cyber and property pricing are both moving away from their own loss cost, with cyber's distance measured in a loss ratio already at a multi-year high and property's measured in a reinsurance-fed retreat that has outpaced any comparable improvement in underlying catastrophe risk. None of those four gaps closes or widens on the composite's schedule.

Further Reading

Sources