Swiss Re Sigma 1/2026: Secondary Perils Hit 92% of the $107B Nat Cat Bill, Crossing a Structural Threshold

Wildfires, severe convective storms and floods produced 92% of the $107 billion of insured natural catastrophe losses in 2025, according to Swiss Re Institute's sigma 1/2026. Tropical cyclones and earthquakes together accounted for about 8%.

The obvious objection is a quiet hurricane season. The number that answers it is the floor: SCS alone has cleared $50 billion for three consecutive years.

Key Takeaways

• 92% of $107 billion came from perils the industry classifies as secondary, with the January Los Angeles wildfires at $40 billion and US severe convective storms at $51 billion.
• Three consecutive years above $50 billion of US SCS insured losses, following $55 billion in 2023 and roughly $41 billion in 2024, which makes $50 billion a floor rather than a peak.
• 60% of North American wildfire loss growth is not explained by exposure growth or insurance penetration, against a 14% annual growth rate for the peril there.
• 11% reduction in Swiss Re's own nat cat volumes in Q1 2026, taken on peak-peril pricing adequacy, which is where the reinsurance discipline signal is pointed.
• 49% insured share of $220 billion of economic losses is the highest on sigma records, and still leaves $113 billion uninsured.

What the 92% Contains

The composition is concentrated in two perils and one country.

Peril Category	2025 Insured Losses	Share of Total	Trend (Annual Growth)
Severe Convective Storms	$51B	~48%	~7% per year
Wildfires	$40B	~37%	~12% per year
Floods	$3.4B	~3%	~6% per year
All Secondary Perils Combined	~$98B	92%	5-7% real terms
Tropical Cyclones + Earthquakes	~$9B	~8%	Episodic
Total Insured Nat Cat Losses	$107B	100%	5-7% per year

The Palisades and Eaton fires produced $40 billion of insured losses against roughly $53 billion economic, destroying more than 16,000 structures in some of the densest concentrations of residential property value in the United States. An insured-to-economic ratio around 75% reflects both high penetration in that market and the minimal role federal programs play in wildfire compared with hurricane or flood.

SCS reached $51 billion without a single catastrophic outlier. The March 10 to 12 outbreak produced $8 to $10 billion across 26 states, and no convective event that year exceeded $15 billion. The total accumulated through cadence rather than magnitude, which is the structural difference that matters downstream.

The quiet-hurricane objection does not survive the arithmetic. SCS and wildfire produce losses every year, with annual floors around $50 billion and $10 to $15 billion respectively. Tropical cyclones produce near-zero in quiet seasons and $30 to $100 billion in active ones. Over any multi-year average the persistent floor accumulates faster than the intermittent spike, which is why the secondary share has run above 70% for several years and the trend holds across both active and quiet cycles. Munich Re's independent estimate puts non-peak perils at roughly $98 billion insured out of $166 billion in total losses, consistent with Swiss Re's split.

A Frequency Loss Running Through a Severity Structure

The pricing consequence follows from a mismatch between how these losses arrive and how the reinsurance covering them is built.

Property catastrophe excess-of-loss is designed around a low-frequency, high-severity event piercing a high attachment every several years. SCS produces frequent moderate losses that erode retained earnings quarter after quarter without reaching that attachment. So the SCS trend is predominantly a primary carrier retention problem: reinsurers collect higher ceding premium as primary rates rise, and take limited loss unless aggregate or quota share structures are in place. Aggregate covers and aggregate-trigger cat bonds fit the loss shape that per-occurrence towers do not.

The trend selection is where this reaches a rate indication, and the headline growth rate is the wrong input for it. The 5% to 7% real annual growth in insured nat cat losses is a blended figure that already embeds exposure growth, which most rating plans capture separately through the premium base. Applying it whole double-counts.

The decomposition sigma provides is what makes the correction possible. North American wildfire grows at 14% a year with roughly 60% unexplained by exposure or penetration, meaning hazard intensification and concentration in the wildland-urban interface carry most of it. Gallagher Re attributes up to 90% of SCS loss growth since 2000 to non-weather factors: population migration, construction cost escalation, legal system cost inflation and insurance-to-value gaps. Flood runs 6% globally and 12% in Asia. A single trend factor applied across a mixed book misestimates every peril in it, in directions that do not offset.

Reserving inherits a related problem. SCS losses often fall below a carrier's internal catastrophe threshold and flow into attritional development, which makes the SCS trend hard to separate from general homeowners experience. Event numbering aggregates some multi-day outbreaks and splits others, distorting per-occurrence triggers and loss development factors. And in Texas, Colorado and Florida, attorney involvement gives residential property claims a tail that looks attritional at 12 months and accelerates at 18 to 36.

The Discipline Signal Points at the Other Perils

Reinsurance pricing has historically been the strongest external pressure on primary rate adequacy. On the perils now producing 92% of losses, that pressure is weakest.

Swiss Re cut nat cat volumes by 11% in Q1 2026 while posting $1.5 billion of net income, and Munich Re reduced retrocession volumes on underwriting discipline. Both moves target peak-peril pricing adequacy, tropical cyclone and earthquake, which is where the reinsurance risk-pricing signal is sharpest and where 8% of the 2025 loss sat. The perils doing the damage receive the least of that signal.

The models are the second constraint. Most commercially deployed SCS and wildfire models were calibrated on event sets predating the 2020 to 2025 escalation; research presented at CAS meetings documents a 4.6% variance between expected and actual SCS losses over 1990 to 2023, and the gap widened in the three most recent years. Moody's RMS released North America SCS HD models in December 2025 calibrated against more than $55 billion of location-level claims and validated over 2,700 damage curves, which is a direct response to resolution failures: hail that drops 3-inch stones on one subdivision and rain two miles away is smoothed away by coarse footprints.

Capital treatment is the third, and it is only now moving. NAIC proposal 2025-20-CR adds wildfire to the binding Rcat formula, the first time the peril has received treatment equivalent to hurricane and earthquake, and 2025-19-CR separates earthquake and hurricane loss reporting for 2027 annual statements. Both are corrections in the right direction and both arrive after six consecutive years of insured losses above $100 billion, which means the capital charge has been sized against a peril composition the loss data stopped matching some time ago.