Swiss Re: Protection Gap Hits $424B as Nat Cat Losses Climb Past the Structural Threshold

Swiss Re Institute's sigma 1/2026 puts the global natural catastrophe protection gap at $424 billion for 2025, up from $395 billion, the first year it has exceeded $400 billion. The same report records the insured share of economic losses at 49%, the highest on sigma records. Both facts are true at once, and the arithmetic connecting them is the story.

Key Takeaways

• $220 billion of economic losses across 190 events in 2025 against $107 billion insured, a sixth consecutive year above $100 billion and a record 49% insured share.
• The gap still widened by roughly $29 billion in that year, because the economic exposure base grew faster than the coverage expanding to meet it.
• 14% annual growth in North American wildfire insured losses, of which roughly 60% is unexplained by exposure growth or expanding coverage, which is the part attributable to the hazard itself.
• 27.3% global insurance resilience index in 2025 against 25.3% in 2015, two percentage points of improvement across a decade in which the dollar gap grew steadily.

A Record Insured Share and a Wider Gap

Region	Protection Gap (2025)	YoY Growth	Insurance Resilience Index	Key Driver
North America	$140B	6%	~42%	Wildfire, SCS exposure concentration
EMEA	$90B	11%	~41% (advanced)	Flood, urbanization in exposed areas
Asia-Pacific	~$130B	~8%	~29% (advanced)	Monsoon flood, low penetration
Latin America & Caribbean	~$35B	~5%	<15%	Hurricane, earthquake, minimal coverage
Africa	~$29B	~4%	<5%	Drought, flood, near-zero penetration
Global Total	$424B	7%	~27%	Climate, urbanization, reconstruction cost inflation

The ratio and the dollar figure move in opposite directions because they measure different things. The insured share is a fraction; the gap is what is left over from a base that is growing. A 49% share applied to $220 billion of economic loss leaves more uninsured than a 35% share applied to the loss base of the early 2010s did.

The decade view says the same thing more starkly. The Insurance Resilience Index rose from 25.3% in 2015 to 27.3% in 2025, and the absolute gap grew throughout, at a compound rate of roughly 5% a year, which is the same 5% to 7% real annual growth Swiss Re documents for insured nat cat losses themselves. Coverage is expanding at the rate risk is expanding, so the difference between them compounds.

Regional resilience shows where the base is growing hardest. North America's index has been essentially flat at 40% to 42% since 2015 on a $140 billion gap growing 6% a year. Advanced EMEA improved from 37% to about 41%, which Swiss Re credits partly to dikes, levees and land-use planning constraining flood loss growth in the UK, France, Switzerland and Austria. Asia-Pacific carries roughly 30% of global economic catastrophe losses and 5% of global insured losses, and its 2025 economic losses of $65 billion against about $5 billion insured leave a 92% uninsured share.

The Residual Is What Historical Calibration Cannot Carry

Exposure growth explains more than 80% of the long-term rise in weather-related insured losses since 1970, which is the finding usually quoted to argue the hazard has not changed. The peril-level residual is where the pricing problem lives.

North American wildfire insured losses are growing at roughly 14% a year, and about 60% of that increase is not explained by exposure growth or expanding coverage. European severe convective storm losses are growing at roughly 10% a year, approximately twice what exposure growth alone predicts. Those residuals are hazard intensification, and they are the component a loss distribution calibrated to historical experience does not contain, because the period it was fitted to had a different frequency.

Two other forces compound it in the same direction. Wildland-urban interface growth in the United States runs 1.8 times faster than non-WUI areas nationally and 1.9 times faster in California, where one-third of residents now live in the WUI. And U.S. reconstruction costs sit 37% above December 2019 levels, which raises the replacement value of insured and uninsured stock alike.

The reconstruction figure is the one that acts as a multiplier rather than an addend. It means an unchanged geographic coverage ratio produces a larger dollar gap every year, and it means the exposure in a model calibrated on pre-2020 replacement values is understated before any hazard adjustment is applied. Swiss Re's Jerome Jean Haegeli puts the combination plainly: "Most long-term loss growth comes from a simple reality: more valuable property is being built in harm's way, and rebuilding costs have risen."

Take-up moves the wrong way against all of this, because intensifying hazard raises the price in exactly the places where affordability is already the binding constraint. Only 12% of California residential policies carried earthquake coverage in 2024, down from 30% at the time of the 1994 Northridge earthquake. The $40 billion of insured losses from the January 2025 Los Angeles wildfires came out of a market with high homeowners penetration, and still left substantial uninsured economic damage behind them.

The Products That Fit the Gap Do Not Scale at the Gap's Speed

Conventional indemnity products have had decades to reach these populations and have not, which is why the response set is parametric, index-based and mitigation-linked rather than a rate change.

Parametric structures pay a fixed amount on an objective trigger, wind speed, rainfall or earthquake magnitude, rather than indemnifying actual loss. They remove loss adjustment cost, cut moral hazard and move payment from months to days, and they are growing at 15% to 20% a year against 5% for traditional agricultural cover. Swiss Re's Heilongjiang pilot covers 28 counties on satellite and weather triggers, which shows the model works at province scale.

Microinsurance has the harder economics and the more instructive failure. Mobile distribution reaches populations agency networks cannot serve, and several programmes have passed a million policyholders. Persistency is where they break: take-up drops sharply after the first year when no payout occurs, which is the predictable behaviour of a product whose value is invisible in a year without an event, sold to buyers with no capacity to hold a multi-year view.

Against that, the growth rate of the thing being closed. If insured losses continue on the 5% to 7% long-term trend they reach approximately $148 billion in 2026 and $186 billion annually by 2030, and Swiss Re assigns roughly 10% annual probability to a peak-loss year producing $320 billion insured. Economic losses in that scenario run substantially higher, pushing the protection gap above $500 billion in a single year.

That is the constraint on every product response. A parametric market compounding at 15% to 20% from a small base and a gap compounding at 5% from $424 billion do not converge on any timescale a product roadmap operates over, and a single peak-loss year moves the target by more than a decade of the innovation moves the coverage.