In the same quarter, US property rates fell 13% and US casualty rates rose 7%. Calling that a soft market is costing people money. Here is what the two halves are actually doing, and why the winning response is the same in both.
Picture a mid-sized US manufacturer renewing its programme this summer. The property quote comes back 13% cheaper than last year. The general liability quote comes back 7% dearer. Same broker, same week, same market.
Both numbers are right. They come from the same page of the same report: Marsh’s Global Insurance Market Index for the second quarter of 2026.
The industry has settled on a single phrase for this: soft market. That phrase is now doing real damage, because half the market is not soft at all, and the people planning around the average are getting both halves wrong.
Global commercial rates fell 6% in the second quarter, after 5% in the first. That is the eighth consecutive quarter of decline. Property led it at 12% down worldwide, and the regional spread is wide: down 19% across India, the Middle East and Africa, 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, 5% in Asia.
Cyber fell 4%. Financial and professional lines fell 3%.
The pressure is coming from upstream. At the January 2026 reinsurance renewals, Howden measured risk-adjusted property catastrophe rates on line down 14.7%, with retrocession down 16.5%. The year before, that figure was 8%. It is the sharpest fall since 2014.
Two things caused it. Dedicated reinsurance capital reached roughly $501bn by the end of 2025, up 8% in a year, pushing the sector’s solvency margin ratio to 113% and its highest level since 2021. And the losses did not arrive: Swiss Re put insured natural catastrophe losses at $42bn for the first half of 2026, the lowest first half since 2020 and 16% below the ten-year average.
Record capital, quiet weather. Prices fall.
David Howden, the founder and chief executive of the broker that bears his name, read it optimistically in January. “The message coming from our analysis of (re)insurance markets is clear: this is a rare moment where everyone stands to benefit. We’re in the midst of a softening market where prices are falling despite elevated political and economic volatility.”
About the half he is describing, he is right.
Marsh recorded casualty up 2% globally. That average hides the important part, which is that casualty fell in every region except one. In the US it rose 7%.
The line-level forecasts are steeper. General liability rose 5.6% in the fourth quarter of 2025 with broker forecasts pointing to as much as 9% in the first quarter of 2026. Commercial auto liability rose 9.2% and was forecast to run between 7% and 15%. Underlying loss trends are holding at 12% to 15%.
The reason is not mysterious. It is showing up in courtrooms.
Marathon Strategies counted 135 US verdicts above $10m in 2024. That is 52% more cases than the year before, and the total value more than doubled, up 116% to $31.3bn. Forty-nine of those verdicts passed $100m and five passed $1bn. They landed in 34 states, across 55 industries, with product liability leading. It was the heaviest year since the firm started counting in 2009.
Then there are the reserves, and this is where the number most people quote is misleading.
Milliman’s review of 2024 statutory results found $15.8bn of adverse prior-year development across four casualty segments, the worst on record for those lines. But the figure the industry repeated was $7.8bn, the net across all liability lines. The gap between the two is one line running the other way.
| Line | 2024 prior-year development |
|---|---|
| Other liability, occurrence | $10.0bn adverse |
| Commercial auto | $3.8bn adverse |
| Non-proportional reinsurance liability | $1.7bn adverse |
| Product liability, occurrence | $0.4bn adverse |
| Casualty subtotal | $15.8bn adverse |
| Workers’ compensation | $6.4bn favourable |
| All liability lines, net | $7.8bn adverse |
Workers’ compensation absorbed $6.4bn of the damage. Take that one line out and the casualty problem is roughly twice the size the headline suggests. The net figure was still more than double 2023’s $3.7bn.
So: property is being repriced by abundant capital, and casualty is being repriced by juries and by reserve reviews. Neither of those forces cares what the other is doing.
There is a second number from this year that gets less attention than the rate indices and matters more to anyone who has to run the business day to day.
Through the first half of 2026, the Wholesale and Specialty Insurance Association recorded $47.6bn of surplus lines premium across the fifteen stamping office states, up 2.8%. Item filings rose 16.9%, to 4.3 million.
Inside that, E&S property premium fell 13.7% year on year while property transaction volume rose 15.2%.
Read those two together. The same books are handling meaningfully more risks and collecting less money for each one. An underwriter who quoted a hundred property submissions last year is quoting a hundred and fifteen this year, for less premium.
That is a quiet but total change in what determines margin. When premium per policy was rising, a manual step in the middle of the process was an irritation. When policy count rises and premium per policy falls, the same manual step is the reason the account is unprofitable. Cost to serve stops being an operations metric and becomes the underwriting result.
It is also the part of this that technology genuinely answers. The work is in handling the policy lifecycle without a person retyping anything, and in automating the steps between submission and issue so that the fifteen extra quotes do not need fifteen extra hours.
John Donnelly, president of global placement at Marsh Risk, described what insurers are doing about all this competition. “In many markets, in addition to competing based on price, insurers are seeking to differentiate themselves through broader coverage, expanded policy terms, and lower deductibles.”
Sit with that for a second. Broader coverage, longer terms, lower deductibles. Those are not product innovations. They are the price, paid in a currency that does not show up in a rate index.
It is also the opposite of what the market told itself in January. Howden’s renewal report was careful to say this cycle was different: “not a return to the underwriting practices of the last soft market. Attachments remain elevated by historical standards, terms and conditions are tighter; capital is being deployed selectively.”
Six months separate those two statements. One says terms are tighter. The other says insurers are widening them to win business.
Both can be true at once, because reinsurance and primary insurance are different markets with different disciplines, and the discipline erodes first where it is cheapest to erode. But it should worry anyone planning a 2027 book. A rate cut can be undone at the next renewal. A deductible you gave away and a term you extended follow the account for years, and they surface as loss development long after the person who agreed them has moved on. The last soft market ended that way.
Which leaves the harder route: sell something different rather than the same thing cheaper.
The growth in this market is not in the lines everyone already writes. It is in the risks the admitted market is walking away from. Severe weather, cyber exposure and now AI are pushing standard carriers to pull back, and surplus lines are picking up what they drop.
Delegated authority is where much of that lands. Gallagher Re puts total MGA premium above $125bn, around 12.5% of US property and casualty premium, with roughly 10% growth expected again this year. Those are not businesses winning on capital. They win on being able to open a niche before anyone else does.
Some of the newest examples are lines that did not exist two years ago. AI liability and AI performance cover went from a conference topic to Lloyd’s policies with $25m limits. Parametric triggers left catastrophe and turned up in cloud outage endorsements. Neither existed because rates were attractive. They existed because someone built the product.
That is the capability worth having when price is not available as a lever, and it is mostly a question of how long a product takes to ship. Writing a new specialty product should be a configuration job, not a release: describing the cover and having the coverages, rules and rating come back as a reviewable change with the product builder, then moving rates per line and per state in a rating engine you can edit. In a market pulling in two directions you need rate to move both ways at once, which is difficult if every change is a ticket. For MGAs and coverholders especially, the time from idea to binder is the actual product.
A few things worth keeping straight while all this is happening.
The cycle is not something any single carrier or MGA controls. Capital will keep arriving until something makes it stop, juries will keep doing what juries do, and property rates will keep falling until a bad season changes the arithmetic.
What is controllable is how fast you can act on any of it. The products launched during a soft market are still on the books when it hardens, and the cost of running them is set by decisions made now. If that is the problem in front of you, how the platform is put together is the more useful conversation, and it is priced in public.
The retirement forecast the industry has quoted for years named 2026, and it is 2026. The harder problem is that the junior roles which used to replace those people have gone, and the industry may be blaming the wrong cause.
Insurers spent 2025 and 2026 writing AI out of standard policies and building a new market to replace it. Here is what Armilla, Munich Re, AIUC, Klaimee, HSB and Testudo actually sell, and what it takes to launch a product like theirs.
A small commercial MGA integrates a new distribution partner in three weeks. Fortune 100 carriers have been working on the same problem for two years. Here is why the gap exists, what a working insurance API actually has to do, and why the bar is about to rise again.
Book a live, personalized demo with our product team - tell us your use case and see the platform work with your data. No commitment.