Secondary perils — the industry's shorthand for everything that is not a named hurricane or an earthquake — accounted for a record 92% of the $107 billion in global insured catastrophe losses last year, according to Swiss Re Institute's latest sigma tally.
The headline needs its context. Total insured losses came in below the long-run natural catastrophe trend, and 2025 was notable for something that did not happen: no major hurricane made landfall in the United States. A quiet year, in other words, mostly because nothing enormous hit Florida.
What actually cost money
The Los Angeles wildfires at the start of 2025 contributed roughly $40 billion — the largest wildfire loss event on sigma records. Severe convective storms added $51 billion, making 2025 the third-costliest year on record for that peril, behind 2023 and 2024. Floods went the other way: $3.4 billion against a five-year average of $15.4 billion.
That split is the whole story. One event in one state, in January, accounted for more than a third of the year's insured bill. Everything else was scattered hail, wind clusters, downpours and fire.
'Secondary' describes the model, not the damage
The word is a legacy of how catastrophe models were built: named storms and earthquakes were treated as peak perils and got the attention, the capital and the reinsurance layers. Everything else was modelled as a residual. In 2025 the residual was 92% of the losses — a record share, and a reminder that the label says more about the history of the industry than about the size of the cheque.
Four numbers that do not agree
Gallagher Re put 2025 catastrophe insured losses at $129 billion, Aon at $127 billion, Munich Re at $108 billion and Swiss Re at $107 billion. A $22 billion spread between top and bottom is not a rounding error. These are estimates from competing modelling houses with different event sets and different views of what counts as insured, not one audited total. Anyone quoting a single figure should say whose.
The trend underneath the quiet year
Balz Grollimund, who heads catastrophe perils at Swiss Re, was blunt about the below-trend result: it reflects "favourable variability rather than any easing of underlying risk". If losses revert to long-term norms, 2026 would come in at about $148 billion. On the firm's modelled peak-loss scenario, it could reach roughly $320 billion.
Urs Baertschi, chief executive of property and casualty reinsurance, made the same point from the capital side: a peak year "could be more than double the recent annual insured natural catastrophe losses and exceed $300 billion". That is the business model — absorbing low-frequency, high-severity events that turn a quiet year into a record one.
The structural number in the report is the 5–7% annual growth in long-term insured losses. Jérôme Jean Haegeli, head of the institute, attributes most of it to a simple reality: more valuable property is being built in harm's way, and rebuilding costs have risen. Then comes the caveat that matters for pricing — for some perils and regions, hazard and vulnerability appear to be evolving faster than exposure alone would explain. It is not only that there is more stuff to damage.
The gap that is not closing
Economic losses were $220 billion, of which about 49% was insured — the highest share on sigma records. That ratio is genuinely good news. It is also an average. In emerging economies, 80–90% of catastrophe losses are typically uninsured, which is where the protection gap is actually wide, and where adaptation and insurance coverage have to be built together rather than in sequence.
What to watch
The headline total for 2026 is the least interesting variable. Watch two things instead: whether a landfall-free year repeats, and whether the 5–7% growth trend holds in a year when it does not. The 92% secondary-peril share is the more durable finding. The losses that now dominate the industry's bill are the small, frequent, scattered ones that no single event explains, and they are the hardest to model, price and reinsure.