When Record Catastrophe Losses Meet Record Reinsurance Capital, National Premium Contagion Mostly Stops

The intuitive story about disaster and insurance pricing runs like this: a wildfire or hurricane destroys property in one place, insurers pay claims, reinsurance gets more expensive, and homeowners a thousand miles away find a larger number on their renewal notice. The story is mechanically coherent. It is also, on the evidence of the last eighteen months, mostly not what happened.

2025 produced the largest insured wildfire loss ever recorded. The reinsurance market responded by cutting prices at the steepest rate in more than a decade. Both statements are true and not in tension, and why they are not in tension is the most useful thing to understand about how localized disasters reach national pricing.

The Setup as of Mid-2026

Start with the loss side, because it is the part that makes the intuitive story feel obviously right. Swiss Re Institute's sigma count put global insured natural catastrophe losses for 2025 at USD 107 billion, against total economic losses of USD 235 billion. Wildfires, storms and floods accounted for 92% of insured losses, a record share for that group of perils. Within that total, the January 2025 Los Angeles wildfires produced roughly USD 40 billion in insured losses, the largest insured wildfire loss on record anywhere. Severe convective storms contributed about USD 51 billion. The insured share of economic losses reached 49%, the highest on record.

Those are not small numbers, yet the 2025 total sat just below the ten-year average of USD 111 billion and well below 2024's USD 141 billion. A year containing the largest wildfire loss in history still came in under trend, because the trend has moved.

Now the pricing side. At the January 1, 2026 renewal, the Guy Carpenter global property catastrophe rate-on-line index fell 12%. The United States fell 12%. Asia Pacific fell 12%. Europe fell 15%. Howden Re's measure of risk-adjusted property catastrophe pricing put the decline at 14.7%, the largest reduction since 2014, with property retrocession down 16.5% and direct and facultative business down 17.5%. Across the full 2026 renewal calendar including mid-year, the global index has fallen roughly 16%, which is the steepest annual decline since the late 1990s and steeper than any single year of the 2010s soft market.

This was not a one-year reversal. The same index fell 6.6% at January 2025 and 8.1% at the 2025 mid-year renewals. Softening has been running for two full years, through a period that included the costliest wildfire event ever underwritten.

What Was Absorbing the Losses

Capital. Aon put global reinsurer capital at a record USD 785 billion at the end of 2025, up from USD 715 billion a year earlier. Traditional capital reached USD 649 billion; alternative capital reached USD 136 billion, a rise of more than 18% in a single year. The catastrophe bond market issued USD 25.6 billion in 2025, up 45% year on year, taking outstanding notional to USD 61.3 billion at year end, a 24% expansion. Fifteen new sponsors came to the market, an annual record.

A market that adds roughly USD 70 billion of capacity in the same twelve months it pays out USD 107 billion of insured losses is not a market under capital stress. It is a market where supply is growing faster than demand can absorb it, and prices in such a market fall regardless of how dramatic the individual loss events were.

The effect shows up downstream too. The US effective approved homeowners rate change was 13.6% in 2024, 6.3% in 2025, and 1.8% through July 2026. Some states swung hard: Minnesota went from 17.8% in 2025 to 1.6% in 2026, Colorado from 16.6% to 0.8%. S&P Global Market Intelligence characterised the resulting environment as a fragmented, state-by-state answer set built on localized catastrophe experience, regulatory posture and earned performance rather than a uniform national resolution.

The Four Channels, and Which Ones Are Actually Open

The claim that a local disaster reaches distant policyholders is not wrong in principle. It is a claim about specific plumbing, and the plumbing has four distinct pipes that open and close independently. The intuitive story treats them as one.

Channel One — Paid Claims Against Capital

Direct claim payments reduce an insurer's surplus, and a smaller surplus supports less written premium, which in principle pushes pricing up across the whole book. This channel scales with the ratio of the loss to the capital base, not the size of the loss in isolation. Against USD 785 billion of dedicated reinsurance capital, a USD 40 billion regional event is a bad quarter for carriers concentrated in that region and a rounding adjustment for everyone else. The channel opens only when an event or clustered season consumes a visible fraction of sector capital.

Channel Two — Reinsurance Repricing at Renewal

This is the channel the intuitive story leans on hardest, and it is the one that has been most thoroughly throttled. Two things closed it. The first is the capital surplus described above. The second is structural: after the 2022 to 2023 hard market, reinsurers raised attachment points sharply, and Moody's Ratings expects them to stand firm on those elevated attachment points through the 2026 renewals, leaving primary insurers to retain a large proportion of secondary-peril losses themselves.

That structural change matters more than the price move. A higher attachment point means a mid-sized regional wildfire or a severe convective storm outbreak does not reach the reinsurance layer at all. It stops at the primary carrier's retention. If the loss never touches the reinsurance tower, it cannot be transmitted through reinsurance pricing to policyholders in other states. The reinsurance channel now carries peak events and largely ignores everything else.

Channel Three — Model Revision Entering Rate Filings

Catastrophe model vendors do revise their views after major events, and a revised view can raise indicated rates for a whole peril class rather than a single ZIP code. But a model output is not a premium. It becomes a premium only after it passes through a state rate filing, and the states differ enormously in how much friction they impose. California is the sharpest illustration: catastrophe modelling was introduced into ratemaking there only under the Sustainable Insurance Strategy, with the associated reinsurance and ratemaking regulations finalized for administrative approval on December 30, 2024, and carriers using them commit to writing policies covering at least 85% of properties in distressed areas. Before that, a revised wildfire model could not lawfully drive a California rate indication at all.

The practical consequence is a lag measured in quarters and a geography set by regulatory calendars rather than by the fire's footprint — a poor mechanism for fast national contagion.

Channel Four — Residual Market Assessments

This is the channel that stayed open, and it is the one the intuitive story usually omits. On February 11, 2025, the California Insurance Commissioner approved a USD 1 billion assessment on member insurers to fund FAIR Plan wildfire claims. The FAIR Plan had paid more than USD 900 million in wildfire claims as of February 9, 2025, and held more than 451,000 policies. Under the governing framework, insurers may recover half of that assessment from policyholders through a temporary supplemental fee expressed as a percentage of premium, subject to prior approval under Proposition 103, and may not fold the assessment into future base rates. It was the first assessment passed through to customers in more than thirty years.

Note precisely what this channel does. It reaches a homeowner in an inland county who filed no claim and faces no wildfire exposure — exactly the outcome the intuitive story predicts. But it stops at the state line. A residual-market assessment is a creature of state statute and cannot reach a policyholder elsewhere. The channel that actually functioned in 2025 is powerful and narrow at once.

Two Precedents That Ran in Opposite Directions

1992 — When a Single Event Did Rewrite the Structure

Hurricane Andrew produced USD 15.5 billion of insured losses in 1992 dollars, roughly USD 25 billion restated into 2011 dollars. Seven domestic insurers and one foreign insurer became insolvent as a direct result. The industry's method of estimating catastrophe exposure changed wholesale: before Andrew, insurers projected future losses largely from historical experience data, and afterwards probabilistic catastrophe modelling became standard practice. Florida's institutional architecture was rebuilt around the gap, with the Residential Property and Casualty Joint Underwriting Association formed in 1997 and merged with the Florida Windstorm Underwriting Association into Citizens Property Insurance Corporation in 2002.

Andrew is the strongest case for the intuitive story, and it is instructive about the conditions required: a loss large relative to the capital of the exposed carriers, a modelling framework demonstrably wrong, and no residual-market structure yet in place. None of those three holds in 2026.

2017 — Record Losses, Capital That Refilled

2017 set the then-record for insured natural catastrophe losses at USD 135 billion, against USD 330 billion of overall economic losses. The United States accounted for 50% of global losses that year, against a typical share closer to 32%. There were 710 recorded events versus an average of 605, and insured losses ran almost three times the then-prevailing USD 49 billion average. Hurricane Harvey alone caused around USD 85 billion in overall losses; Irma was the costliest for insurers at roughly USD 32 billion.

Yet 2017 registers in the pricing data as a soft market low. After two consecutive years of softening, the Guy Carpenter global index still sits more than 38% above that 2017 trough, and 19% below its 2024 peak as measured at January 1, 2026. The enormous 2017 loss year began from a pricing floor and produced no durable repricing regime; the hard market arrived in 2022 and 2023, driven by loss trend, repair cost inflation and a retreat of capital rather than by any single storm.

The Objection That Has Weight

The argument above is conditional, and the conditions are visible enough to monitor.

• A genuine peak-peril event. Swiss Re Institute frames 2026 as running to roughly USD 148 billion of insured losses at normal long-term levels, but flags a peak-loss scenario of about USD 320 billion. A loss of that magnitude would breach elevated attachment points across the market and reopen channel two immediately. Nothing in the current reading survives a USD 300 billion year.
• Alternative capital is not permanent capital. The USD 136 billion of alternative capital and USD 61.3 billion of outstanding catastrophe bonds behave differently from balance-sheet equity under stress. Collateral can be trapped after a loss event, and investors can decline to renew. Capital that arrived in eighteen months can leave in six.
• National averages conceal the cases that matter. A 1.8% national approved rate change is consistent with double-digit increases in individual states. A reading built on the national aggregate will systematically understate what a policyholder in a concentrated-peril market experiences.
• Softening reinsurance does not mechanically become cheaper insurance. Rate filings lag, approval is discretionary, and in some jurisdictions the net cost of reinsurance became an admissible ratemaking input only recently. Cheaper reinsurance improves carrier margins first and policyholder pricing second, if at all.
• The assessment channel is asymmetric. It operates precisely when the private market is failing, which is when the other three channels are least informative. Residual-market growth is therefore a leading signal that the intuitive story is becoming true again in a particular state — and it is measured in policy counts, not in reinsurance prices.

The claim that one disaster moves national pricing is not false so much as unconditioned. It described 1992 accurately and describes 2026 poorly. Which regime applies is answered by capital supply and attachment structure, not by the severity of the headline event.

What to Watch Next Week

• Catastrophe bond issuance and spread data. Primary-market spreads widening while issuance volumes hold indicates investors demanding more for the same risk — the earliest available signal that alternative capital is turning, well before any renewal quote.
• Residual-market policy counts. Month-on-month growth in state FAIR Plan or equivalent policy counts, against the 451,000-plus California figure as a reference point, shows whether the assessment channel is being loaded for a future event.
• State rate filing dockets. Approved versus requested rate changes, tracked against the 13.6% / 6.3% / 1.8% national series, reveal whether regulatory friction or carrier appetite is the binding constraint in a given state.
• Attachment point commentary in reinsurer disclosures. Any indication that reinsurers are willing to sell lower layers again would mark the reopening of channel two, and would matter more for national transmission than the headline rate change.
• Reinsurance broker commentary ahead of the mid-year cycle. The gap between January and mid-year renewal outcomes has widened in each of the last two years, and is now more informative than either point alone.

Concrete Framework — The Monitoring Sequence

A repeatable monitoring routine for the question "should a disaster in another state change the expected renewal here":

1. Size the event against sector capital, not against the news cycle. Divide the estimated insured loss by roughly USD 785 billion of global reinsurer capital. Below about 2%, channel one is closed by arithmetic. The USD 40 billion Los Angeles wildfire loss sits near 5% of that base and still did not move global pricing upward.
2. Establish whether the loss reached the reinsurance tower. If commentary describes the loss as retained by primary carriers, channel two is closed for that event regardless of headline severity. Elevated attachment points have been the standing condition since the 2022 to 2023 reset and are expected to hold through 2026.
3. Identify the affected state's residual market vehicle and check its policy count trajectory. This is the only channel documented to have reached non-claiming policyholders in 2025, and it is state-bounded. A rising policy count is a loaded assessment waiting for an event.
4. Read the rate filing series, not the premium anecdote. Track approved versus requested changes for the specific state. The national series of 13.6%, 6.3%, and 1.8% is the baseline against which a state's divergence becomes meaningful.
5. Set an explicit invalidation threshold before the season starts. A season approaching the USD 320 billion peak-loss scenario, or a visible contraction in the USD 136 billion alternative capital base, flips the regime. Deciding the threshold in advance prevents rationalising after the fact.
6. Re-run the sequence at each renewal date rather than after each disaster. Pricing regimes change at January 1, April 1, June 1 and July 1 renewals. Disasters are inputs to those dates, not events with independent pricing authority.

The distinction worth holding onto is between a mechanism existing and a mechanism being active. All four channels exist permanently in the structure of insurance markets. In 2026, three are throttled by an unusually large capital base and unusually high attachment points, and the fourth reaches widely but stops at a state border. That configuration is not permanent, and the thresholds above exist to catch the moment it ends.

This analysis is for general informational purposes and describes market structure and publicly reported data. It is not financial, insurance or investment advice, and it is not a forecast of any carrier's pricing decisions.