Global commercial insurance rates fell 6% in the second quarter of 2026, the eighth consecutive quarterly decline and the steepest of the run, after 5% in the first quarter (Marsh, July 2026). Property fell 12% globally and 13% in the US. Casualty moved the other way, up 2% globally and 7% in the US. One index, two cycles.

Inside the Q2 Print: Every Region Down, Two Lines Apart

Marsh's Global Insurance Market Index has now declined for eight consecutive quarters, and the sequence is steepening rather than bottoming: 1% in Q3 2024, the first drop since 2017, then 2%, then 3%, then three straight quarters at 4%, then 5% in Q1 2026, and now 6% (Marsh, July 2026). Every region posted a lower composite. India, Middle East and Africa led at 16%, 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, at 2% after a 1% decline in Q1, was the shallowest market in the index (Marsh, July 2026). Marsh tied the pressure to strong insurer profitability, surplus capital, lower reinsurance costs, and higher investment returns, the conditions feeding competition in every line that is falling.

The line detail is where the composite stops being useful. Global property fell 12% after 9% in the first quarter. Financial and professional lines fell 3% globally even as the US booked a 1% increase, and cyber fell 4% for a twelfth consecutive quarter (Insurance Journal, July 2026). Casualty rose 2% globally against all of that, decelerating from 3% in Q1, and the US book does the lifting: US casualty rates rose 7% in the quarter, and 11% with workers compensation excluded (Insurance Journal, July 2026). A composite that nets a 12-point property decline against a 2-point casualty increase into a single 6 answers a portfolio question, not a pricing one.

None of this will surprise anyone who sat through the Q2 calls. Chubb disclosed major-account E&S property pricing down 12% in the same window, and the broader earnings season showed commercial pricing decelerating while casualty holds. What the Marsh print adds is breadth and a clean time series: this is renewal-level data across regions and lines from the largest broker in the world, which makes it the closest thing the market has to a published rate index, and the divergence it shows is now wide enough to require separate treatment in every pricing and reserving assumption that touches it.

What Eight Year-over-Year Prints Compound To

Each Marsh print is a year-over-year rate change on programs renewing in that quarter, not a sequential move from the prior quarter. That distinction controls the cumulative arithmetic. Multiplying all eight prints together, 0.99 down through 0.94, produces a 26% cumulative decline, and that figure is wrong: consecutive quarterly prints measure overlapping twelve-month windows, so chaining them double counts nearly the entire decline.

The chain that describes a real account runs down the same renewal quarter in successive years. An account renewing every second quarter renewed 4% lower in Q2 2025 and another 6% lower in Q2 2026, which compounds to roughly 10% below its Q2 2024 premium. Running the same two-year cohort chain by line: global property renews about 18% below its 2024 level (7% then 12% declines, per Marsh's Q2 2025 and Q2 2026 releases), US property about 21% below (9% then 13%), global casualty about 6% above (4% then 2% increases), and US casualty roughly 17% above (9% then 7%). Two simplifications ride along and should be stated rather than buried: Marsh's figures are averages across a shifting book of renewing programs rather than a matched sample, and the published prints are rounded to whole points, so the chained values carry a point or so of slack in either direction.

QuarterGlobal compositeLine detail where published
Q3 2024-1%First composite decline since 2017
Q4 2024-2%
Q1 2025-3%
Q2 2025-4%Property -7%, casualty +4%, US casualty +9%
Q3 2025-4%
Q4 2025-4%US composite flat
Q1 2026-5%Property -9%, casualty +3%
Q2 2026-6%Property -12%, casualty +2%, US casualty +7%

Those cohort chains are the written-rate half of an adequacy statement. A US property account renewing 21% below its 2024 premium needs its expected loss cost to have fallen by about that much for margin to hold, and expected loss cost has done nothing of the sort. On casualty, a cumulative 17% sounds like margin being built until it stands next to the severity trend. That comparison is the next piece of arithmetic.

Casualty at +2% Against an 11% Loss Growth Trend

The severity backdrop comes from the Swiss Re Institute's sigma study on social inflation, which found litigation costs drove US liability claims up 57% over the decade through 2023, with the social inflation component alone peaking at 7% a year in 2023, its highest reading in two decades (Swiss Re Institute, September 2024). Over the five years to 2023, US commercial casualty insurance losses grew at an average 11% a year to reach $143 billion, liability lines exposed to bodily injury accumulated $43 billion of underwriting losses, and 2023 produced 27 courtroom awards above $100 million each (Swiss Re Institute, September 2024). Nothing in the 2024-2026 docket data suggests that process has stopped; the site's coverage of the litigation-abuse reframe tracks the same trend through this year's broker studies.

Set Marsh's casualty prints against that trend with the assumptions in the open. Swiss Re's 11% measures growth in losses, which bundles exposure growth with per-unit severity; grant two to three points for exposure and the residual per-claim trend runs high single digits, consistent with a 7% social inflation reading stacked on ordinary economic inflation. Against a trend in that range, the global +2% concedes five or six points of adequacy a year, the US blended +7% concedes a point or two, and only the ex-comp +11% holds level. The four-point wedge between the US blend and the ex-comp figure is workers compensation dragging the average down, so an actuary reading the blended +7% as a margin signal is averaging a book that is keeping pace with severity against one that is being repriced downward. Same word on the tab, different lines underneath.

The two-year version is less forgiving. Chained, US casualty written rate stands roughly 17% above Q2 2024, while a persistent 11% growth trend compounds to 23% over the same two years, and even the friendliest exposure assumptions leave the blended book no better than level. Written rate that treads water against severity during the accident years now earning through is precisely the mechanism that built the 2016-2019 casualty reserve problem the industry is still developing through. The reserving translation runs through the development factors: severity that outruns written rate shows up first as LDF selections that keep proving light, then as the calendar-year reserve strengthening that Q2 statutory data is already recording in Schedule P. The +2% global casualty print records a line conceding ground to its own claims trend, just more slowly than property is conceding ground to competition.

Property at -12% While the Reinsurance Leg Falls Faster

Property is the steepest line in the index and it is accelerating, from a 9% global decline in Q1 to 12% in Q2. The declines land everywhere: 19% in IMEA, 15% in the Pacific, 14% in Latin America and the Caribbean, 13% in the US, 11% in the UK, 9% in Europe, 8% in Canada, and 5% in Asia (Insurance Journal, July 2026). Inside the US book the decline splits sharply by segment: catastrophe-exposed programs above $1 million in premium renewed down 20%, twice the 10% decline on non-catastrophe programs below $1 million (Insurance Journal, July 2026). The soft market is concentrating exactly where the hard market concentrated, in large cat-exposed accounts, which is what a capacity-driven cycle looks like when the capacity returns.

The ceded leg is falling faster than the gross leg. Guy Carpenter's July 1 renewal data put global and US property catastrophe rate on line down 16% for 2026 and 23% below the 2024 hard-market peak, though still about 32% above the 2017 soft-market low (Guy Carpenter, via Artemis, July 2026). The loss side is cooperating: Gallagher Re counted $46 billion of global insured natural catastrophe losses in the first half of 2026, 28% below the 10-year average of $64 billion and down from $84 billion a year earlier (Gallagher Re, July 2026). Benign experience plus cheapening reinsurance is the standard fuel mixture for a property soft market, and both tanks are currently full.

Work the net economics of a cat-exposed account before concluding the reinsurance relief pays for the rate cuts. Take a program at 100 of premium in Q2 2025 that spends 30 of it on a catastrophe tower. The gross renewal at 13% down brings premium to 87; the tower repriced at Guy Carpenter's 16% decline costs 25.2; net premium falls from 70 to 61.8, a decline of nearly 12%. Cheaper reinsurance claws back barely more than a point of the 13-point gross decline, because the ceded spend is a minority of the premium base. Meanwhile the expected loss against that net premium has not repriced at all: one benign half-year moves the historical average, not the modeled annual aggregate loss. The margin compression that pushed reinsurer ROE projections from roughly 19% toward 14-15% is arriving one layer down, on primary cat books whose rate level is now a fifth below 2024 while their cat load is not.

The 2013-2017 Precedent for How Soft Cycles End

The last soft cycle ran four and a half years, and its ending is the relevant part. Global pricing fell from Q2 2013 through Q3 2017, with the final print a 1.9% decline, a sequence that had shallowed toward zero; then hurricanes Harvey, Irma, and Maria plus the California wildfires flipped the index to a 0.8% increase in Q4 2017, the first rise since Q1 2013, inside a single quarter (Insurance Journal, February 2018). The hard run that followed lasted far longer than the decline that preceded it: Marsh was still counting consecutive increases at Q4 2023, the 25th in a row (Reinsurance News, February 2024), and the streak only broke with Q3 2024's 1% decline.

Set this cycle against that template and the differences instruct more than the similarity. Eight quarters into the 2013-2017 decline, prints sat in shallow low single digits; eight quarters into this one they are accelerating, three 4s into a 5 into a 6, which says capacity is still arriving faster than discipline. The exit last time came from a catastrophe landing on a market priced to thin margins rather than from any gradual recognition that rates had fallen too far. Marsh's placement leadership is naming the same contingency in advance this time. "Current market conditions are likely to persist absent a severe northern hemisphere storm season or string of major natural catastrophes" (Marsh, July 2026), per John Donnelly, Marsh's president of global placement. An actuary pricing property today should read that sentence as a conditional probability statement: the soft market's duration is a function of the wind, and the wind season that answers it is the one now underway.

Reading a Broker Index Against Your Own Book

A broker index is renewal-level, year-over-year, and mix-weighted, which makes it a directional prior rather than an on-level input. The rate change that belongs in an indication or an a priori loss ratio is the one measured on your own renewing book, by line, layer, and account size; Marsh's series is the cross-check that tells you whether your book's movement is idiosyncratic or the market. Used that way, this quarter's release carries three specific instructions. First, never carry the composite anywhere: the 6 nets a 12-point property decline against a 2-point casualty increase, and even within US property the large cat-exposed segment moved twice as fast as the small non-cat segment. Second, chain cohorts, not prints, when building rate-level history; the eight-quarter streak compounds to roughly 10% on composite and 21% on US property, not 26%. Third, mind the written-to-earned lag: a Q2 2026 written decline earns through mid-2027, so earned rate level keeps falling into next year even if written prints stabilize this fall, and calendar 2027 loss ratio projections need that transform rather than the headline.

The divergence itself sets the watch list. The wedge between Marsh's global property and casualty prints has widened from 11 points in Q2 2025 to 12 in Q1 2026 to 14 now, and that spread, more than the composite, is the cycle signal: it measures how hard capacity is chasing the line with benign losses while severity keeps repricing the line without them. The site's read of the Q1 index against combined-ratio quality made the durability argument; a quarter later, the property side has moved 3 points further and casualty has given up another point of rate against an unchanged severity trend. Pricing actuaries should be filing 2027 property indications that assume the current decline persists and documenting the adequacy drawdown explicitly, and casualty reserving actuaries should treat +7% written rate as roughly breakeven against trend, not as margin. The index will print its ninth quarter in October. The wind will decide whether it prints a tenth.

Further Reading

Sources