From reviewing six consecutive years of sigma publications alongside NAIC data calls and AM Best supplements, the secondary peril composition trend has been building since 2019 but crossed a structural threshold in 2025. Swiss Re Institute's sigma 1/2026 quantifies what the industry has been circling around for several years: the perils that carriers, reinsurers, and cat modelers have historically classified as "secondary" now dominate the global insured loss bill.
In 2025, wildfires, severe convective storms (SCS), and floods generated 92% of the $107 billion in total insured natural catastrophe losses. That figure is not an anomaly caused by a quiet hurricane season. The secondary peril share has been rising steadily, and the 2025 result confirms a structural shift in where insured losses actually concentrate, one that demands corresponding changes in how actuaries model, price, and reserve for catastrophe risk.
The headline figures are stark. The January 2025 Los Angeles wildfires produced $40 billion in insured losses, making them the costliest wildfire event on record and one of the ten most expensive insured natural catastrophe events ever documented (Swiss Re sigma 1/2026). U.S. severe convective storms added $51 billion, ranking 2025 as the third costliest SCS year on record behind 2023 and 2024 in inflation-adjusted terms (Swiss Re). Meanwhile, the perils that have historically commanded the most cat modeling investment and reinsurance capacity, tropical cyclones and earthquakes, accounted for just 8% of the 2025 insured total.
The Sigma 1/2026 Data in Full: What Changed and What Didn't
Swiss Re's sigma research series has tracked natural catastrophe losses since the 1970s. The 2025 edition, published in March 2026, documents several records and near-records that collectively paint a picture of accelerating secondary peril dominance.
Global economic losses from natural catastrophes totaled $220 billion in 2025. Of that, $107 billion was insured, producing a 49% insured share, the highest proportion on sigma records (Swiss Re). This narrows the global protection gap, though the improvement is concentrated in mature markets, particularly the United States, which accounted for 83% of global insured nat cat losses in 2025, approximately $89 billion (Swiss Re).
Munich Re's parallel NatCatSERVICE estimate pegged total economic losses at $224 billion and insured losses at $108 billion, broadly consistent with Swiss Re's figures. Munich Re additionally noted that weather disasters accounted for 92% of all 2025 losses and 97% of insured losses, and that non-peak perils (floods, SCS, and wildfires) produced $166 billion in total losses, of which roughly $98 billion was insured (Munich Re 2025 Natural Disaster Figures).
| 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 flood figure stands out for its modesty. Global insured flood losses of $3.4 billion were well below the five-year average of $15.4 billion (Swiss Re). This reflects the absence of a major riverine or coastal flood event in a high-insurance-penetration market during 2025, rather than any structural improvement in flood risk. The protection gap for flood remains wider than for any other secondary peril, particularly in Asia, where insured flood losses are growing roughly 12% per year but the vast majority of economic losses still fall outside coverage (Swiss Re).
The LA Wildfires: A $40 Billion Inflection Point
The Palisades and Eaton fires that swept through Los Angeles County in January 2025 generated $40 billion in insured losses against total economic losses of approximately $53 billion (Munich Re). This makes them the costliest wildfire event by a wide margin, exceeding the previous record, the 2018 Camp Fire and associated events, by roughly fivefold.
The scale was driven by a convergence of hazard, exposure, and vulnerability factors. A prolonged Santa Ana wind season coupled with negligible winter rainfall created conditions for rapid fire spread. More than 16,000 structures were destroyed, many of them high-value single-family homes in some of the densest concentrations of residential property value in the United States (Swiss Re). The insured-to-economic loss ratio of approximately 75% reflects both the high insurance penetration of the affected market and the minimal participation of federal disaster programs in wildfire relative to hurricane or flood.
For actuaries tracking the wildfire peril specifically, sigma 1/2026 flags a compounding pricing problem. Wildfire insured losses have been growing at approximately 12% per year, faster than any other peril class (Swiss Re). In North America, where the growth rate exceeds 14%, roughly 60% of the increase is not explained by exposure growth or expanding insurance penetration. Hazard intensification and disproportionate population growth in wildfire-prone areas account for the unexplained portion (Swiss Re). This distinction matters for pricing actuaries because it means that exposure trend factors alone will systematically understate the forward-looking loss expectation. The hazard component must be trended separately, and legacy cat models calibrated before the 2020-2025 wildfire escalation will miss it entirely.
Severe Convective Storms: $51 Billion and the Third Consecutive $50B+ Year
Severe convective storms have quietly assembled a loss record that dwarfs tropical cyclones on a cumulative basis. The $51 billion in 2025 SCS insured losses followed $55 billion in 2023 and approximately $41 billion in 2024 (Swiss Re; Munich Re). This marks the third consecutive year that U.S. SCS losses exceeded $50 billion, a sustained level that Aon's January 2026 report confirmed has pushed the cumulative SCS insured loss total since 2000 past the cumulative tropical cyclone total for the same period.
The 2025 SCS season did not produce a single catastrophic outlier event on the scale of the LA fires. Instead, the losses accumulated through a relentless cadence of moderate-to-large events spread across the Midwest, Southeast, and Plains. The March 10-12, 2025 SCS outbreak produced $8-10 billion in damages across 26 states, but no single convective event exceeded $15 billion in insured losses. This frequency-driven loss pattern is precisely what makes SCS fundamentally different from hurricane in its implications for cat modeling and reinsurance treaty design.
For context, 2025 also saw fewer multi-billion-dollar SCS outbreaks than 2023, yet the aggregate loss remained above $50 billion. This suggests that the $50 billion threshold is becoming a baseline rather than a peak, driven by the same non-weather factors that Gallagher Re has identified as responsible for up to 90% of SCS loss growth since 2000: population migration into high-risk areas, construction cost escalation, legal system cost inflation, and insurance-to-value gaps.
Why 92% Is a Structural Threshold, Not a Statistical Artifact
An obvious objection to the 92% figure is that 2025 was a below-average year for tropical cyclone activity. If a Category 4 or 5 hurricane had made U.S. landfall, the secondary peril share would have been lower. This is true but misses the structural point that sigma 1/2026 documents.
The upward trend in secondary peril share has persisted across both active and quiet hurricane seasons. Swiss Re's data shows insured nat cat losses growing at 5-7% per year in real terms, a trend that holds when averaged across decades and multiple hurricane cycles. Within that trend, secondary perils have consistently gained share. The 92% figure is the current high-water mark, but the trajectory has been above 70% for several years running.
The reason is arithmetic. SCS and wildfire produce losses every single year, with an annual floor that has risen to roughly $50 billion for SCS and $10-15 billion for wildfire (excluding outliers like the LA fires). Tropical cyclones, by contrast, produce zero or near-zero insured losses in quiet seasons and $30-100 billion in active ones. Over any multi-year averaging period, the persistent floor of secondary peril losses accumulates faster than the intermittent spikes from tropical cyclones.
Swiss Re's own forward projections reinforce the structural interpretation. If 2026 insured nat cat losses revert to the long-term trend, they would total approximately $148 billion (Swiss Re). In a peak-loss scenario, defined as losses exceeding the trend by at least one standard deviation and occurring with roughly 10% annual probability, 2026 losses could reach approximately $320 billion (Swiss Re). By 2030, the baseline trend projects $186 billion in annual insured losses (Swiss Re). These projections embed the secondary peril growth rates described above and assume that the peril composition shift is permanent rather than cyclical.
Cat Model Implications: Where the Calibration Gaps Are Widest
The 92% secondary peril share creates a direct challenge for catastrophe models, which remain architecturally and financially tilted toward tropical cyclone and earthquake. Several structural factors drive the mismatch.
Calibration vintage. Most commercially deployed SCS and wildfire models were calibrated using event sets that predate the 2020-2025 loss escalation. Research presented at CAS meetings has documented a 4.6% variance between expected and actual SCS losses over the 1990-2023 period (CAS Actuarial Review), and that gap has widened in the three most recent loss years. Models calibrated through 2019 will systematically underestimate the frequency and severity of convective events at current exposure levels.
Hazard resolution. SCS damage varies dramatically at sub-kilometer scales. A hailstorm that drops 3-inch stones on one subdivision may produce only rain two miles away. Legacy models with coarse spatial resolution produce unrealistic event footprints that smooth over this micro-scale variability, leading to both overestimation in some territories and underestimation in others. Moody's RMS addressed this directly with its December 2025 North America Severe Convective Storm HD Models, which were calibrated against over $55 billion in location- and policy-level claims data and validated more than 2,700 property damage curves (Moody's RMS).
Wildfire model maturity. Wildfire catastrophe modeling is substantially less mature than tropical cyclone modeling. The NAIC's new Natural Catastrophe Risk and Resilience Task Force ran a wildfire model evaluation under ASOP No. 38 with four approved vendors (RMS, Verisk, KCC-CoreLogic), finding increasing model consistency but noting that wildfire model uncertainty remains wider than for hurricane, particularly for wildland-urban interface events where fire behavior interacts with structure density and defensive actions (NAIC Spring 2026).
Flood modeling gaps. Despite producing lower insured losses than SCS or wildfire globally, flood is the peril where the protection gap is widest and where model confidence intervals are the most generous. Inland flood model confidence intervals run two to four times wider than coastal surge models, driven by limited gauging station density, heterogeneous soil conditions, and rapid land-use change in flood-prone areas. Swiss Re's own data shows flood insured losses growing at 6% per year globally but at 12% in Asia, where insurance penetration is lowest.
Reinsurance Treaty Pricing: The Structural Mismatch Between Peril and Product
From tracking reinsurance renewal cycles, a pattern has emerged that sigma 1/2026 puts into sharp focus. Traditional property catastrophe excess-of-loss treaties are designed around low-frequency, high-severity events: the paradigmatic hurricane or earthquake that pierces a high attachment point every several years. SCS events, by contrast, produce frequent moderate-severity losses that erode retained earnings quarter after quarter but rarely trigger excess-of-loss coverage.
Swiss Re's own behavior illustrates the dynamic. The reinsurer deliberately reduced nat cat volumes by 11% in Q1 2026 while posting $1.5 billion in net income (Swiss Re Q1 2026 earnings). Munich Re similarly cut retrocession volumes and emphasized underwriting discipline over premium growth. These strategic pullbacks focus on peak-peril pricing adequacy, specifically tropical cyclone and earthquake, where the reinsurance market's risk-pricing signal is strongest.
The SCS loss trend is predominantly a primary carrier problem. Reinsurers benefit from higher ceding premiums driven by SCS-related rate increases on the primary side but bear limited loss exposure unless aggregate stop-loss or quota share structures are in place. The pricing signal from the reinsurance market, historically the strongest external pressure for primary carrier rate adequacy, is muted for SCS relative to its actual loss contribution.
This mismatch has practical implications for cedants evaluating catastrophe reinsurance programs. Aggregate excess-of-loss covers, which respond to the accumulation of smaller events across the year, may be more structurally appropriate for SCS exposure than traditional per-occurrence towers. The growth in cat bond capacity covering aggregate triggers rather than single-event triggers aligns with this structural need, and the ILS market's 2026 issuance pace suggests that investors are beginning to price aggregate secondary peril risk more explicitly.
NAIC Response: Task Force Consolidation Aligns with the Sigma Findings
The NAIC's institutional response to the secondary peril shift has been more substantive than the typical working group reshuffle. At the Fall 2025 National Meeting, the NAIC consolidated the Climate and Resiliency Task Force, the Catastrophe Insurance Working Group, and the FEMA Working Group into a single Natural Catastrophe Risk and Resilience Task Force. Two new subgroups emerged: the Pre-Disaster Mitigation and Risk Modeling Working Group, and the Severe Peril Working Group (NAIC Spring 2026).
The regulatory actions taken at the Spring 2026 meeting align directly with sigma 1/2026's findings on peril composition.
Wildfire Rcat charge adoption. Proposal 2025-20-CR adds wildfire to the binding Rcat formula within the P&C risk-based capital framework. Four vendor models have been approved for wildfire Rcat calculation, with own-model permissions available to carriers that meet specified data requirements at selected return periods. This is the first time wildfire has been given equivalent RBC treatment to hurricane and earthquake, a direct acknowledgment that the peril's loss contribution now warrants capital charges of comparable rigor.
Earthquake-hurricane reporting separation. Proposal 2025-19-CR separates earthquake and hurricane loss reporting in the PR100 schedule and updates the SRSS (square root of the sum of squares) aggregation formula for 2027 annual statement filings. By disaggregating these perils, regulators can assess how much of a carrier's cat risk charge is attributable to each peril individually, enabling more targeted solvency review (NAIC Spring 2026).
Severe Peril Working Group mandate. The new working group is specifically chartered to evaluate how updated SCS catastrophe models will appear in rate filings and whether current regulatory review processes can adequately assess model-driven rate indications for convective storm exposure. This is a notable shift from the NAIC's historical posture of relying on backward-looking experience data for SCS ratemaking.
IBNR Estimation: Why Frequency-Driven Perils Complicate Reserving
From working with catastrophe reserve development data, frequency-driven secondary perils produce a fundamentally different IBNR pattern than peak-peril events. A single landfalling hurricane generates a concentrated burst of claims that, while potentially large and slow to settle in complex commercial lines, forms a well-defined event that actuaries can track through development. The event date is known, the geographic footprint is mapped, and the claim count stabilizes within weeks.
Secondary perils, by contrast, produce a continuous stream of moderate-severity events throughout the year. In 2025, the United States experienced over 142 days with damaging hail of 2 inches or greater (Cotality), each potentially generating thousands of claims. State Farm received over 53,000 home and auto claims from storms in a single two-week period in March 2025 alone. Allstate reported $1.24 billion in pretax catastrophe losses for Q1 2025, with 15 wind and hail incidents in March generating approximately $925 million.
This claims volume creates several reserving complications for P&C actuaries.
Event definition ambiguity. Catastrophe event numbering systems (PCS serial numbers in the United States) sometimes aggregate multi-day SCS outbreaks into a single event and sometimes split them. This creates inconsistency in how losses are allocated across events, affecting per-occurrence treaty triggers and distorting loss development factor (LDF) calculations when events are analyzed individually.
Attritional vs. catastrophe classification. Many SCS losses fall below the individual carrier's internal catastrophe threshold, particularly for smaller events. These losses flow into attritional reserve development patterns rather than catastrophe reserves, making it difficult to isolate the SCS loss trend from broader non-catastrophe homeowners loss experience.
Adjusting capacity constraints. The sheer volume of SCS claims can overwhelm field adjusting capacity, particularly during spring outbreak seasons. Delayed inspections extend the reporting lag and compress development patterns later in the calendar year. Carriers relying on standard chain-ladder development factors may understate late-reported SCS claims in preliminary reserve estimates.
Litigation tail. In SCS-prone states with aggressive litigation environments, including Texas, Colorado, and Florida, attorney involvement in residential property claims adds a severity tail that extends settlement timelines and inflates ultimate claim costs beyond initial field estimates. This creates a development pattern that looks attritional in the first 12 months but accelerates at 18-36 months, a pattern that standard reserving methods may not capture without explicit adjustment.
What Drives the 5-7% Annual Growth: Exposure, Climate, and the Residual
Swiss Re's sigma 1/2026 decomposes the drivers of insured nat cat loss growth into several components. Economic growth, urbanization, and the associated accumulation of insurable assets have been the primary driver historically. Insurance penetration improvements, particularly in emerging markets, add a secondary layer. Climate change contributes a third component that varies substantially by peril and region.
The decomposition matters for actuarial trend selection because each component implies a different forward-looking trajectory. Exposure growth is somewhat predictable from demographic and economic data. Climate-driven hazard intensification is less predictable and may be nonlinear. The "residual" component, which Swiss Re attributes to factors like aging infrastructure, building code gaps, and land-use decisions that increase vulnerability, is the most difficult to project but may also be the most responsive to mitigation investment.
Regional decomposition reveals significant variation. In North America, wildfire loss growth of 14% per year is roughly 60% attributable to non-exposure factors, meaning that hazard intensification, population concentration in the wildland-urban interface, and vulnerability factors (aging roofs, vegetation management gaps) are doing most of the work (Swiss Re). For SCS, Gallagher Re's analysis attributes up to 90% of loss growth to non-weather factors: population migration, construction cost escalation, litigation, and insurance-to-value gaps.
In Asia, flood is the dominant secondary peril growth driver at 12% annual loss growth, with rapid urbanization and land-use change contributing the majority of the increase. In Oceania and Australia, the growth splits more evenly across SCS and flood with a smaller wildfire component (Swiss Re).
For pricing actuaries selecting catastrophe loss trend factors, the decomposition has a direct application. A trend factor based purely on historical loss growth (the 5-7% headline figure) embeds all components, including exposure growth that may already be captured separately in the premium base. Double-counting is a real risk. The actuarially correct approach isolates the frequency and severity trend from the exposure trend, then applies each to the appropriate base. But doing this requires peril-specific decomposition data of the type that sigma 1/2026 provides, data that most carriers do not generate internally.
The Protection Gap: 49% Insured Share Still Leaves $113 Billion Uninsured
The 49% insured share, the highest on sigma records, represents genuine progress in closing the global protection gap. But the aggregate figure conceals a stark regional divide. In the United States, where 83% of global insured losses occurred, the insured share is well above 50% for most perils. In emerging economies, 80-90% of catastrophe losses remain uninsured (Swiss Re).
For actuaries working in admitted markets, the protection gap is less about uninsured populations and more about underinsured properties. The LA wildfires exposed massive insurance-to-value gaps, with many homeowners carrying replacement cost limits that lagged actual reconstruction costs by 30-50%. In SCS-prone states, the gap manifests differently: percentage deductibles for wind and hail shift a significant portion of loss back to the policyholder, reducing the insured loss total but creating a different kind of protection gap that does not appear in the sigma statistics.
Swiss Re notes that adaptation and insurance must work together to narrow the gap. Every $1 invested in pre-disaster mitigation can save up to $13 in future disaster costs (Swiss Re sigma 1/2026). The economic return on mitigation investment is highest for secondary perils precisely because of their frequency: a fortified roof reduces losses from every hailstorm, not just a once-in-a-generation hurricane. This makes the mitigation credit framework more actuarially tractable for SCS and wildfire than for peak perils where the return period is measured in decades.
Why This Matters for Actuarial Practice
The sigma 1/2026 data compels a reassessment of how P&C actuaries allocate their catastrophe modeling, pricing, and reserving attention. Six consecutive years of $100 billion-plus insured losses, with secondary perils dominating the composition, is not a cyclical anomaly. It is the new baseline.
Cat model adoption urgency. Carriers still relying on pre-2025 SCS and wildfire models are pricing off calibration that predates three consecutive $50B+ SCS years and the costliest wildfire event in history. The transition to updated models from Moody's RMS, Verisk, and Cotality will produce discontinuities in indicated rates that pricing actuaries must communicate to underwriting and management. Delaying adoption does not reduce the risk; it accumulates the underpricing.
Reinsurance program design. The structural mismatch between frequency-driven secondary peril losses and traditional per-occurrence excess-of-loss treaties means that primary carriers absorb a disproportionate share of SCS losses. Actuaries advising on reinsurance program structure should evaluate aggregate covers, catastrophe quota shares, and ILS products with aggregate triggers as complements to traditional per-occurrence towers.
RBC capital charges. The NAIC's addition of wildfire to the binding Rcat formula is a first step, but the current RBC framework still allocates the majority of catastrophe risk charge to hurricane and earthquake. Carriers with significant SCS exposure, particularly those concentrated in the Midwest and Southeast where SCS is the dominant peril, should assess whether their internal capital models reflect the actual loss distribution documented in sigma 1/2026.
Reserve segmentation. Actuaries monitoring reserve adequacy should track secondary peril loss development separately from peak peril development. Blending SCS, wildfire, and hurricane into a single catastrophe reserve triangle masks deterioration in one peril with favorable development in another. The frequency-driven nature of secondary perils produces development patterns that differ structurally from the concentrated, event-specific development of hurricane reserves.
Trend factor decomposition. The 5-7% real annual growth rate in insured nat cat losses is not a single number to be applied uniformly. It decomposes into peril-specific growth rates (12% for wildfire, 7% for SCS, 6% for flood) and further into exposure, hazard, and vulnerability components. Pricing actuaries selecting catastrophe trend factors must match the decomposition to their own book's peril and geographic composition or risk systematic misestimation.
The industry's collective vocabulary is changing. "Secondary peril" is an increasingly misleading label for the loss drivers that now generate 92% of the insured nat cat bill. The actuarial response, in models, prices, reserves, and capital frameworks, needs to catch up to what the data has been signaling for the past six years.