What the 2026 Climate Reports Mean for Insurance

On 28 October 2025, Hurricane Melissa came ashore in southern Jamaica with sustained winds of 185 mph and a central pressure of 892 millibars. No Category 5 storm had ever struck the island. Around 100 people died. Munich Re put the damage at $9.8bn, of which $3bn was insured; Aon counted $11bn and $2.5bn.

Either way, roughly three-quarters of what the storm destroyed was nobody’s claim.

Ten days later the World Bank confirmed that Jamaica’s catastrophe bond would pay in full. The trigger was the storm’s central pressure and track as reported by the National Hurricane Center, checked by an independent calculation agent. No adjuster visited a roof. The $150m reached the government’s account on 1 December.

That one storm holds the whole of this article: a climate signal, a protection gap that swallowed most of the loss, and a product that worked. 2026 is the year the reports caught up with it.

Three years above 1.5°C

The Copernicus Climate Change Service published its Global Climate Highlights on 14 January 2026. It ranked 2025 the third warmest year on record, at 1.47°C above the 1850–1900 baseline and 0.13°C cooler than 2024. The number that mattered was the average: 2023, 2024 and 2025 together came in above 1.5°C, the first three-year period ever to do so. Mauro Facchini, head of Earth observation at the European Commission, called it “a milestone none of us wished to reach.”

The agencies do not agree on the ranking. NASA has 2025 second; NOAA and Copernicus have it third. The World Meteorological Organization, which consolidates six datasets, settled on about 1.43°C in its State of the Global Climate 2025 on 23 March and called the year second or third. The last eleven years are the eleven warmest whichever series you pick.

The WMO report matters more for what it promoted than for what it ranked. For the first time it lists Earth’s energy imbalance, the gap between heat arriving from the sun and heat leaving, as a headline indicator, and it set a record in 2025. More than 90% of that surplus goes into the ocean.

For an underwriter, the imbalance is the line that separates weather from trend. Air temperature wobbles with El Niño and La Niña. Stored heat does not come back out.

In June, the annual Indicators of Global Climate Change update from Piers Forster’s group at Leeds put human-induced warming alone at 1.37°C, rising about 0.27°C a decade. The remaining carbon budget for a coin-flip chance at 1.5°C was 130 gigatonnes of CO₂ at the start of 2026. That is a little over three years of current emissions.

The next record year already has a date

The WMO’s Global Annual to Decadal Climate Update, published in May, gives a 91% chance that at least one year between 2026 and 2030 exceeds 1.5°C, and an 86% chance that one of them beats 2024 as the warmest on record.

The mechanism is already running. El Niño probabilities for August to October sit near 90%, and Munich Re’s half-year review talks of a possible record-strength event by year end.

Copernicus’s July bulletin showed the run-up. July 2026 tied July 2024 as the warmest month ever measured, 1.47°C above pre-industrial. The extra-polar ocean surface reached 20.96°C, a record for the month. Western Europe had its warmest June and July on record.

“July 2026 was the third consecutive month of exceptional heat in western Europe, bringing the combined temperature for June and July to a new record for the region.”
Samantha Burgess, Strategic Lead for Climate, European Centre for Medium-Range Weather Forecasts

Möckern, in eastern Germany, recorded 41.8°C. The Robert Koch Institute counted more than 5,000 heat-related deaths in Germany between April and June.

Dr Leon Hermanson, the update’s lead author, put a year on it: “There is an El Niño predicted for the end of 2026, which increases the chances of the following year, 2027, being the next record-breaking year.”

A quiet year, and why nobody in the industry believes it

Set against that, the loss ledger for 2025 looks almost reassuring. It depends on whose ledger.

The spread is scope: which events clear the threshold and how sub-billion losses are counted. Swiss Re’s figures are the spine here because the market quotes its trend projections back.

Inside the $107bn, a record 92% came from what the industry still calls secondary perils: wildfire, severe convective storm and flood. The Palisades and Eaton fires in Los Angeles produced $40bn of insured loss in January, the costliest wildfire in history. Severe convective storms added $51bn on Swiss Re’s count and $61bn on Aon’s, and Aon now ranks them the costliest insured peril of the 21st century.

For the first year since 2015, no hurricane made landfall on the US mainland.

The insured share of economic loss reached 49%, the highest in sigma’s records, and that headline hides where the money was. The United States took around four-fifths of insured losses. In Asia-Pacific, Aon counted $76bn of damage and about $7bn of insurance. In the European Union, EIOPA puts the long-run insured share at roughly a quarter, and below 5% in Greece.

Then the first half of 2026 came in lower still

Swiss Re’s preliminary estimate for the first half of 2026 is $42bn of insured natural catastrophe loss, the lowest first half since 2020 and 16% below the ten-year average. Munich Re says $44bn, Gallagher Re $46bn, Aon $47bn.

There was no $10bn event.

“A less costly first half does not mean risk disappeared. One major hurricane, earthquake or wildfire can quickly change the picture.”
Balz Grollimund, Head Catastrophe Perils, Swiss Re Institute

The second half of a year carries 58% of insured losses on average, mostly North Atlantic hurricanes, and El Niño tends to suppress them. What Swiss Re’s half-year note dwells on instead is heat: Europe has 64% more hot days than in the 1950s, and European wildfire losses have grown 8–11% a year in real terms since 1970.

The models moved the other way

On 1 September, Verisk published its Global Modeled Catastrophe Losses report. It raised the industry’s expected insured loss in an average year to $171bn, up $19bn on a year earlier and the highest figure the modeller has ever put out. A 1-in-100 year now models at $477bn.

Severe convective storm is 40% of that risk. Wildfire, for all the attention it gets, is 6%.

“A quiet hurricane season can lead markets to respond as if risk has eased: rates soften, insurers keep more risk on their own books, and more capital competes to write new business.”
Rob Newbold, President, Verisk Catastrophe and Risk Solutions

Swiss Re’s trend line says the same thing in different units: $148bn is a normal 2026, $186bn a normal 2030, and there is a one-in-ten chance of a $320bn year.

What the two reinsurers disagree about is why the line rises. Grollimund is precise: 2025 was “the result of favourable variability rather than any easing of underlying risk,” and the long-run growth of 5–7% a year in real terms is “structural” because “exposure keeps building.” Verisk gives that driver numbers: insured property exposure up about 7% a year since 2021.

Munich Re’s chief climate scientist Tobias Grimm puts the emphasis elsewhere, opening his firm’s review with “a warming world makes extreme weather disasters more likely,” and board member Thomas Blunck names “climate change and growing exposure” in that order. Both houses agree on the direction and the rate. They disagree about how much of the slope belongs to the atmosphere and how much to the people who keep building in front of it, and no honest reading of 2026 lets you drop either half.

Retreat, one policy at a time

The retail end of the market has stopped waiting for that argument to settle.

On 5 August the National Association of Insurance Commissioners released its first national analysis of homeowners insurance, built on seven years of market conduct data. Between 2018 and 2024, non-renewal rates rose between 96% and 216% depending on region, and more than tripled in the West. Inflation-adjusted premiums rose between 18.3% and 43.3%.

With 103 million policies and 715 companies in the data, this is not a story about a few carriers and a few states.

California is the sharpest case. The FAIR Plan, the state’s insurer of last resort, held 696,562 policies by mid-2026, up 157% since September 2022. The Los Angeles fires cost it around $4bn and forced a $1bn assessment on member insurers.

Its rates rise 29.1% on 15 October.

Federal Reserve chair Jerome Powell told the Senate Banking Committee in February 2025 that “if you fast-forward 10 or 15 years, there are going to be regions of the country where you can’t get a mortgage.” First Street puts a figure on the same idea: $1.47tn of US property value at risk by 2055.

“Certain locations and perils cannot be covered as we would wish them to be covered. We cannot help it.”
Günther Thallinger, Member of the Board of Management, Allianz SE

Thallinger, speaking to Bloomberg in July 2026, went further: heat, floods, storms and wildfires could become frequent enough that “risk-adequate pricing would not be affordable any longer.” That is a board member of Europe’s largest insurer describing the edge of the product.

Europe is answering with law. Italy made catastrophe cover compulsory for businesses, small firms included from 1 January 2026; go without and you forfeit public disaster aid. EIOPA, the ECB and the European Stability Mechanism have proposed an EU public-private reinsurance scheme, arguing that pooling across member states could cut the capital needed by up to 67%.

Meanwhile the cost of protection fell

Here is the part that makes 2026 strange.

Guy Carpenter’s Global Property Catastrophe Rate-on-Line Index is down 16% for the year after the July renewals, the steepest annual fall since the late 1990s. Howden Re measured Florida rates down as much as 25% at the 1 June renewal. Gallagher Re counts $838bn of capital available to the reinsurance market. Catastrophe bond issuance set a half-year record of almost $18bn, according to Artemis.

The rate-cycle mechanics behind that are covered in the piece on the two-speed soft market. The point here is narrower. In the same quarter, the scientists and the cat modellers marked climate risk up, and the capital markets marked it down. Dean Klisura, Guy Carpenter’s president and chief executive, notes that cedents are “exploring alternative options, such as parametric solutions and sidecars” alongside the cheaper traditional cover.

Of the two forces, capital turns fastest. Rate-on-line fell 16% in a year with no US hurricane landfall. It will not take a record El Niño year to send it the other way.

Products that pay on a data feed

Which brings the argument back to Jamaica. On 18 May 2026 the World Bank priced the replacement for the bond Melissa exhausted: $200m of cover to May 2030, a per-occurrence named-storm trigger, and a risk margin of 6.75% over SOFR. ILS funds took 69% of the paper. Finance minister Fayval Williams called disaster risk financing “a key pillar of our resilience building framework,” which is the kind of sentence a minister says after the cheque has cleared.

The same logic is arriving at building scale. On 15 June, Liberty and the satellite operator ICEYE launched a parametric wildfire cover that classifies each structure as destroyed or undamaged from synthetic aperture radar, which sees through smoke and at night. Results within hours, payouts within days, no site visit. It is available first in the United States and Australia, for homeowner associations, municipalities, public risk pools and portfolio holders.

“By pushing the boundaries of technology, we are delivering a new generation of parametric wildfire coverage that is faster, more responsive, and fully data driven.”
Jean-Christophe Garaix, Head of Parametrics & Agriculture, Liberty

Read that against Los Angeles, where adjusting capacity, not capital, was the bottleneck for months.

Flood is going the same way. In February, Floodbase and Liberty Mutual put parametric flood into a tool that quotes wholesale brokers in minutes, and in March Floodbase opened the same engine to MGAs through an API.

The smallest product here may be the clearest. The Self-Employed Women’s Association in India, with the parametric specialist CelsiusPro, the non-profit HERA and the premium charity Humanity Insured, covers women who work outdoors in Gujarat and Delhi against extreme heat. Forty thousand members were enrolled in 2025; the target for 2026 is 100,000. This summer’s trigger, two consecutive days above the threshold, released about 2.06 million rupees, roughly $21,700.

Tiny money. But the heat that fired it is the same heat in the Copernicus bulletin, and the policy paid on the data rather than on the loss.

None of this needs the mechanics of parametric insurance re-explained, and the basis-risk trade-offs are in the piece on parametric cover beyond catastrophe. What is new in 2026 is the distribution. Sovereigns, HOAs, wholesale brokers and informal workers are buying the same product shape.

Pricing the roof instead of the postcode

The other opening is on the indemnity side, and it starts with a regulation. On 6 March 2026 the Louisiana Department of Insurance issued Bulletin 2026-04: under the state’s new Regulation 136, premium discounts are mandatory for homes holding a FORTIFIED designation from the Insurance Institute for Business & Home Safety. IBHS passed 100,000 designations this year. Aon’s head of catastrophe insight, Michal Lorinc, put the shift in one line: “Resilience today must be both physical and financial.”

For a century, property insurance priced the postcode and ignored the roof. Mandated, portable, certified mitigation credits reverse that. They create a product family whose price moves with a verified physical fact about the risk rather than with the loss history of its neighbours.

Heat opens a second family. Munich Re estimates that ten extra days above 35°C cost an economy about 0.3% of annual productivity, a business interruption exposure that almost no employer, utility or event organiser currently insures.

Community-level parametric layers under indemnity deductibles, Jamaica’s structure at HOA scale, are a third. Weather covers sold at the point of booking, the embedded insurance route, are a fourth. And every peril the standard markets are leaving is a programme waiting for an MGA, in an MGA market where E&S property premium is already contracting.

Each of those is a rating problem before it is anything else. A wildfire or heat product lives or dies on whether a trigger feed can be wired into the rating engine and the cover defined in the product builder in weeks, and on whether the bordereaux reporting to the capacity providers Klisura describes is automatic rather than a spreadsheet. That is what Openkoda is built for, and it is the least glamorous part of the story.

The WMO gives an 86% chance of a new record year before 2030, and the El Niño for it is forming now. Verisk has already written the higher number into the models. Reinsurance capital has not. Between those two prices is roughly one cycle in which to get specialty insurance products that price adaptation into the market, before the quiet ends and the market prices it for you.