The Softening
Catastrophe-reinsurance pricing is falling faster than the loss trend — even as the loss floor keeps rising. One of those lines is wrong, and universal owners are on both sides of the trade.
The Frontier — the weekly deep dive from Universal Asset Owners · Saturday, July 11, 2026
Executive summary
For three years after Hurricane Ian, catastrophe reinsurance was the best-priced risk in institutional markets. Property-catastrophe rates jumped 27.5% at the January 2023 renewals — the hardest market in a generation — and the capital that piled in behind those prices earned record returns: the big four European reinsurers averaged a 19.6% return on equity in 2025, catastrophe bonds delivered their best vintages in a decade, and pensions from Florida to Zurich quietly built billion-dollar allocations to insurance risk.
That market is now softening at a pace that deserves the attention of anyone who owns it. The benchmark global property-catastrophe rate index fell 6.6% at the January 2025 renewals, another 12% at January 2026, and roughly 16% more through the June and July 2026 renewals — with the best-performing North American programs clearing at 20–25% below last year. Global reinsurance capital has never been higher, at $790 billion. Alternative capital has never been higher, at $141 billion. Catastrophe bond issuance just set an all-time first-half record of $17.3 billion, and the spread investors are paid above expected loss fell below 4% for the first time in five years. The price of catastrophe risk is falling because 2025's losses came in below trend, the first quarter of 2026 was the quietest since 2020, and CSU has issued its least-active seasonal forecast since 2013 — though NOAA's official outlook remains a broader below-normal range.
Underneath the pricing cycle, the loss line points the other way. Insured natural-catastrophe losses have exceeded $100 billion for six consecutive years. The long-run growth trend in those losses is 5–7% a year in real terms — a doubling roughly every decade — which puts the expected annual figure near $186 billion by 2030. Depending on the loss aggregator and peril scope, roughly $110–135 billion of 2025 natural-catastrophe economic losses were uninsured; a separate, far broader Swiss Re resilience measure ($424 billion, all insurance lines) should not be conflated with the annual catastrophe-loss gap. And the retreat of private insurance from the riskiest geographies continues: California's insurer of last resort carries over $600 billion of residential exposure, up more than fourfold since 2020.
We frame the collision as a resolvable forecast: will calendar-2026 global insured natural-catastrophe losses reach $100 billion for a seventh consecutive year? A 50,000-path simulation, anchored on verified first-quarter data and an estimated second-quarter range, plus this week's sharply suppressed hurricane forecast, puts the probability at roughly 25–30% — the streak more likely breaks than holds. The median simulated outcome is near $89 billion; there remains roughly a 3% probability of a $150 billion-plus year that would snap the softening overnight. Benchmark conflict: Swiss Re's long-run-trend estimate for 2026 is about $148 billion of insured losses — materially above this desk's season-conditioned median of $89 billion. The gap reflects the desk's estimated quiet first half, an El Niño adjustment and lower Atlantic assumptions; because the divergence is large, treat the ~25–30% result as an editorial scenario estimate, not an actuarial forecast.
The Desk view, in one paragraph. The most dangerous outcome for long-horizon owners is the most likely one: a quiet year. If 2026 closes below $100 billion — as our simulation narrowly favors — the market will read a lucky, El Niño-suppressed draw as vindication of the softening, capital will chase catastrophe risk harder through the January 2027 renewals, and pricing will detach further from a loss trend that has not bent. The mispricing is not this year's premium; it is the cycle's habit of extrapolating its luckiest year. A universal owner sits on every side of this trade at once — as an investor in insurers and catastrophe bonds, as an owner of the real estate, infrastructure and municipal credit that insurance makes financeable, and as the ultimate backstop when private cover retreats. The correct posture is to take the reinsurance cycle's income while it lasts but underwrite every real-asset position as if the insurance bill of 2030, not 2026, were already in the rent roll.
The situation as of today
Three data points, all from the past five weeks, define the moment.
Pricing. At the June 1, 2026 renewals — the Florida-heavy date that prices the sharpest edge of US hurricane risk — risk-adjusted property-catastrophe rates fell by up to 25% on a weighted-average basis. At July 1, the best-performing North American accounts renewed 20–25% down. The benchmark global property-catastrophe rate-on-line index now stands roughly 16% below a year ago, its second consecutive annual decline after the 12% fall recorded at January 2026. Pricing remains above the levels of 2014–2022 — the market has softened from a historic peak, not collapsed — but the direction and the speed are unambiguous.
Capital. Global reinsurance capital reached a record $790 billion at the end of the first quarter of 2026. The alternative segment — catastrophe bonds and other insurance-linked securities, the part institutional investors own directly — hit a record $141 billion, having grown 18% in 2025 alone. The first half of 2026 produced $17.3 billion of new catastrophe bond issuance, an all-time record, and May 2026 was the single most active month in the market's history. The outstanding market reached $65.6 billion at June 30. Demand is so strong that the average spread above expected loss fell to 3.74% in the second quarter — the first sub-4% reading in twenty quarters.
Hazard. On July 8, Colorado State University cut its 2026 Atlantic hurricane forecast to nine named storms, four hurricanes and one major hurricane, with accumulated cyclone energy around 40% of average — the least active forecast since 2013, driven by a strengthening El Niño. CSU puts the probability of a major hurricane striking the United States this year at 17%, against a climatological average of 43%. The first quarter of 2026 produced just $20 billion of global insured losses, 26% below the ten-year average and the lowest first quarter since 2020.
Cheap and getting cheaper; more capital than ever; a season forecast to be quiet. Every input tells the owner of catastrophe risk to relax. This report is about why the 25-year owner should not.
What changed — and what did not
What changed is the cycle. The 2023–2024 hard market did exactly what hard markets do: record prices attracted record capital, record capital produced record returns — reinsurers posted a combined ratio of 79.8% in 2025 — and record returns are now competing those prices away. Two below-trend loss years accelerated the turn. Swiss Re's final tally for 2025 came in at $107 billion of insured natural-catastrophe losses, down from $141 billion in 2024 (in 2025 prices) and well below the roughly $145 billion the long-run trend implied. Then the first half of 2026 came in quieter still.
What did not change is the trend. The 5–7% real annual growth in insured losses is not a weather statistic; it is an exposure statistic. It is driven by where the world builds (coasts, wildland interfaces, floodplains), what it builds (larger homes, denser cities, more valuable contents), what rebuilding costs (construction inflation has outrun general inflation for a decade), and — increasingly — how the hazard itself behaves. Severe convective storms, the unglamorous thunderstorm-and-hail peril, have quietly become the costliest insured peril of the 21st century: US SCS losses have exceeded $20 billion for eleven consecutive years and passed $22 billion again by mid-June 2026 even in this "quiet" year. The January 2025 Los Angeles wildfires produced roughly $40 billion of insured losses from a peril and a month that the market once treated as an afterthought. The floor is rising underneath the cycle, and the two lines — pricing, which moves in three-year waves, and losses, which compound — are now pointed in opposite directions.
The second thing that did not change is the retreat. The Q1 2026 European windstorm cluster was the costliest start for that peril since 1999 — about $22 billion in economic losses — of which barely $3.5 billion was insured. Globally, an estimated $110–135 billion of 2025 catastrophe losses were uninsured (Swiss Re's broader $424 billion resilience-gap figure covers all insurance lines and is not comparable). In California, the FAIR Plan — California's industry-funded insurer of last resort — carried over $600 billion of residential exposure by mid-2025, more than five times its 2020 book, and levied a $1 billion assessment on member insurers after projected Los Angeles fire claims exceeded its immediately available claims-paying resources. Florida's Citizens has moved the other way, shedding nearly three-quarters of its policies since 2023 as reforms and softer reinsurance pulled private capital back in — proof that the retreat is cyclical where policy allows it to be, and structural where it does not.
Source ledger
Source ledger — primary institutional releases and established industry reporting. Confidence: H(igh)/M(edium)/L(ow). "Moves model?" flags whether the item shifted the probability.
| # | Source (publisher, date) | Data point used | Conf. | Moves model? |
|---|---|---|---|---|
| 1 | Swiss Re Institute, sigma (Mar 2026, final 2025) | 2025 insured nat-cat losses $107bn; economic $220bn; 49% insured share, a record | H | Yes — base year |
| 2 | Swiss Re Institute, press release (Dec 16, 2025) | Sixth consecutive year above $100bn; 2024 $141bn in 2025 prices; LA wildfires ~$40bn, a wildfire record; ten-year average $111bn | H | Yes — base rate |
| 3 | Swiss Re Institute, sigma (2025, final 2024) | 2024 insured $137bn; long-run 5–7% real annual loss growth | H | Yes — trend |
| 4 | Gallagher Re, Natural Catastrophe and Climate Report (Jan 2026) | 2025 insured $129bn on broader basis; US $100bn; SCS $60bn; record $838bn industry capital | H | Yes |
| 5 | Munich Re, NatCat figures 2025 (Jan 2026) | 2025 economic $224bn / insured $108bn; LA fires costliest event ($53bn economic) | H | Corroborates |
| 6 | Aon, 2026 Climate and Catastrophe Insight (Jan 20, 2026) | 2025 insured $127bn; SCS now the costliest insured peril of the 21st century | H | Yes |
| 7 | Gallagher Re, Q1 2026 Natural Catastrophe and Climate Report (Apr 2026) | Q1 2026 insured $20bn — 26% below 10-yr average, lowest Q1 since 2020; fourth straight quarter under $40bn | H | Yes — H1 anchor |
| 8 | Gallagher Re, Q1 2026 report (Apr 2026) | European windstorm cluster ~$22bn economic vs ~$3.5bn insured; costliest start for the peril since 1999 | H | Yes — gap |
| 9 | Aon, Q1 2026 Catastrophe Recap (Apr 16, 2026) | US storms and European flood the Q1 drivers | H | Corroborates |
| 10 | Gallagher Re via Artemis (Jun 18, 2026) | US SCS insured losses >$22bn YTD; 11th consecutive year above $20bn; ninth-costliest H1 inflation-adjusted | H | Yes — H1 anchor |
| 11 | Gallagher Re via trade press (late Jun 2026) | June US storm outbreak added a mid-single-digit-billion insured loss | H | Yes — H1 anchor |
| 12 | Colorado State University, seasonal forecast update (Jul 8, 2026) | 9 named / 4 hurricanes / 1 major; ACE ~50 (~40% of average, lowest forecast since 2013); 17% US major-landfall probability vs 43% average | H | Yes — largest single update |
| 13 | NOAA Climate Prediction Center, Atlantic outlook (May 2026) | 8–14 named / 3–6 hurricanes / 1–3 major; 55% probability below-normal; developing El Niño | H | Yes |
| 14 | Guy Carpenter, Global Property Catastrophe Rate-on-Line Index (Jan 2023–Jan 2026) | +27.5% (2023), +5.4% (2024), −6.6% (2025), −12% (2026); index still above 2014–2022 levels | H | Yes — cycle state |
| 15 | Guy Carpenter via Artemis (Jul 2026) | Global and US property-cat indices −16%, Asia-Pacific −19% after July 2026 renewals | H | Yes |
| 16 | Howden Re, June 1 2026 renewals commentary (Jun 2026) | Risk-adjusted property-cat rates down up to 25% weighted average at June 1 | H | Yes |
| 17 | Gallagher Re, July 1 2026 renewals commentary (Jul 2026) | Best-performing North American accounts −20% to −25%+ | H | Yes |
| 18 | Aon, Reinsurance Market Dynamics mid-year 2026 (Jul 2026) | Record $790bn global reinsurance capital at Mar 31, 2026; alternative capital record $141bn | H | Yes — capital |
| 19 | Guy Carpenter / AM Best (Sep–Dec 2025) | Dedicated reinsurance capital $660bn for 2025: $540bn traditional + $123bn alternative | H | Yes |
| 20 | Artemis, Q2 2026 catastrophe bond market report (Jul 2026) | Record H1 issuance $17.3bn; record Q2 $11.3bn; outstanding $65.6bn; Q2 spread over expected loss 3.74%, first sub-4% in 20 quarters | H | Yes — pricing of tail |
| 21 | Artemis, 2025 cat bond year review (Jan 2026) | 2025 issuance record $20.6bn; market +24% to $61.3bn | H | Yes |
| 22 | Swiss Re Institute Resilience Index (Jun 3, 2026) | TOTAL protection gap, all insurance lines (NOT annual nat-cat) $424bn in 2025; ~73% of global exposure uninsured; on-trend insured losses ~$186bn by 2030 | H | Yes — trend |
| 23 | California FAIR Plan, key statistics (Jun 2025) | Residential exposure $603bn, +424% since Sep 2020; 450,000+ policies | H | Yes — retreat |
| 24 | California FAIR Plan (Feb 2025) | $1bn member-insurer assessment after ~$4bn of LA fire claims exhausted reserves | H | Yes |
| 25 | California Department of Insurance, Sustainable Insurance Strategy (2025–26) | Cat models and net reinsurance cost now allowed in rates; major insurer filings approved May 2026; FAIR Plan growth slowed to +2.4% in Q1 2026 | H | Yes — policy response |
| 26 | Florida Citizens / Florida Realtors (Dec 2025–Jan 2026) | Citizens policies down to ~395,000 from a 1.42m peak (−73%); rate cuts recommended for most policyholders | H | Yes — counter-case |
| 27 | US Treasury, Federal Insurance Office (Jan 2025) | 246m policies analyzed: non-renewal rates ~80% higher in highest-risk ZIP codes; premiums rose 8.7% above inflation 2018–22 | H | Yes |
| 28 | AM Best (Dec 1, 2025) | US homeowners segment outlook revised to stable from negative | H | Context |
| 29 | Federal Reserve Bank of Dallas, working paper (Jan 2025) | Rising premiums raise mortgage delinquency; ~203,000 additional delinquent mortgages per year projected 2025–2055 | H | Yes — spillover |
| 30 | First Street (Feb 2025) | $1.47 trillion net US property-value loss by 2055 from insurance-cost capitalization and demand shifts | M | Yes — spillover |
| 31 | BIS, Financial Stability Institute Insights No. 65 (Mar 2025) | "Mind the climate-related protection gap" — reinsurance pricing and underwriting as a stability issue | H | Context |
| 32 | IAIS, Global Insurance Market Report (Nov–Dec 2025) | Financial-stability implications of nat-cat protection gaps a designated supervisory priority | H | Context |
| 33 | EIOPA/ECB, protection gap dashboard and joint proposal (Dec 2025) | Only ~25% of European extreme-weather losses insured 1980–2024; EU-level reinsurance scheme proposed | H | Yes — Europe |
| 34 | Fitch via trade press (2026) | Big four European reinsurers record average ROE 19.6% in 2025; combined ratio 79.8% | H | Yes — cycle |
| 35 | Swiss Re, FY2025 results (Feb 27, 2026) | Record net income $4.8bn, ROE 19.6% | H | Corroborates |
| 36 | Florida Retirement System / City of Zurich pension disclosures via Artemis (H1 2026) | FRS ILS allocation ~$2.23bn (1% of fund); Zurich ~$1.58bn, +6.9% return in 2025 | H | Yes — owner exposure |
| 37 | NOAA NESDIS notice (May 2025) | Billion-dollar disaster database retired; no federal updates beyond 2024 | H | Context — data risk |
Where a report is cited via established trade press (Artemis, Insurance Journal, Reinsurance News), the underlying publisher and date are named and independently checkable. No figure in this report is invented; estimates are labeled as estimates.
Key data table
| Variable | Current reading | Prior reference | Direction | Confidence |
|---|---|---|---|---|
| Global insured nat-cat losses, 2025 (sigma basis) | $107bn | $141bn (2024, 2025 prices) | ↓ below trend | H |
| Consecutive years ≥ $100bn | 6 (2020–2025) | 0 before 2017 on a sustained basis | ↑ structural | H |
| Long-run insured-loss growth trend | 5–7%/yr real | — | ↑ compounding | H |
| Q1 2026 global insured losses | $20bn | 10-yr Q1 avg ~$27bn | ↓ 26% below | H |
| US SCS insured losses, 2026 YTD (mid-June) | >$22bn | 11th straight year >$20bn | → floor intact | H |
| Property-cat rate-on-line index, Jan 2026 | −12% y/y | −6.6% (Jan 2025), +27.5% (Jan 2023) | ↓ softening | H |
| Mid-2026 renewals (Jun/Jul) | −16% to −25% | — | ↓ accelerating | H |
| Global reinsurance capital | $790bn (record) | $660bn dedicated (2025) | ↑ | H |
| Alternative (ILS) capital | $141bn (record) | $136bn (end-2025) | ↑ | H |
| Cat bond H1 2026 issuance | $17.3bn (record) | $20.6bn full-year 2025 (record) | ↑ | H |
| Cat bond spread over expected loss, Q2 2026 | 3.74% | first sub-4% in 20 quarters | ↓ | H |
| 2026 Atlantic forecast (CSU, Jul 8) | 9/4/1, ACE ~50 | least active since 2013 | ↓ hazard (this year) | H |
| Total protection gap, all lines (Resilience Index), 2025 | $424bn | widening | ↑ | H |
| California FAIR Plan residential exposure | $603bn (Jun 2025) | +424% since 2020 | ↑ | H |
The forecast question
Question. Will calendar-year 2026 global insured natural-catastrophe losses reach or exceed US$100 billion — a seventh consecutive $100 billion year — on the Swiss Re Institute sigma basis (current prices)?
Horizon and resolution. Resolves on Swiss Re Institute's estimate for calendar 2026: the preliminary sigma estimate typically published in December 2026, confirmed by the final estimate in the first quarter of 2027. If Swiss Re's basis is unavailable, the average of the Gallagher Re and Aon full-year figures, adjusted for their broader peril scope, substitutes.
Base rate. Six of the last six years resolved Yes (2020–2025, in 2025 prices), and nine of the last ten on the broadest basis. A naive base rate is therefore 85–90%. But conditioning matters: years that entered July with a bottom-quartile first half and a strong-El Niño Atlantic forecast are a different reference class — 2019, the last analogous setup, closed near $52–54 billion.
What would change it. A single major US hurricane landfall (a 17% probability per CSU, but it carries most of the tail), a Japanese typhoon season running hot in an El Niño year, a Q4 wildfire outbreak, or a major insured earthquake — any one of which can add $10–60 billion in a quarter.
Prior and analogues
Four reference points discipline the prior.
2019 is the quiet-year camp's exhibit A: a below-average first half rolled into a full year near $52–54 billion, the last sub-$100 billion print. It shows sub-$100 billion years still happen — but 2019's exposure base was roughly 30–40% smaller in real terms than 2026's. The same weather in 2026 costs meaningfully more.
2023 is the floor camp's exhibit A: a strong El Niño year — the same suppression argument being made today — that nonetheless closed at $108 billion, because record severe-convective-storm losses replaced the missing hurricanes. The lesson: the Atlantic is no longer the swing factor it was. Thunderstorms, hail and floods now carry the floor.
2017 is the tail's exhibit: Harvey, Irma and Maria turned a mid-year lull into a then-record $144 billion in nine weeks. CSU's 17% major-landfall probability is not zero, and El Niño years have produced landfalling majors before.
2005–2006 is the cycle's exhibit: Katrina's record losses hardened the market violently; the quiet 2006–2007 that followed softened it just as fast — into the build-up that made 2008–2011 vintages the ILS market's worst. Soft markets are made in quiet years and paid for in loud ones.
The probability model — evidence update
We start from the naive base rate of ~85% (six consecutive Yes years) and update on the season-specific evidence. Direction is the effect on P(≥ $100bn).
| Evidence item | Direction | Strength | Effect on probability | Confidence |
|---|---|---|---|---|
| Q1 2026 = $20bn, lowest since 2020; four straight quarters < $40bn | ↓ | Strong | 85% → ~65% | H |
| US SCS >$22bn by mid-June but below 5-/10-yr H1 pace | ↓ | Moderate | 65% → ~55% | H |
| CSU Jul 8 cut: ACE ~50, 17% US major-landfall probability (vs 43%) | ↓ | Strong | 55% → ~35% | H |
| NOAA 55% below-normal season probability, El Niño strengthening | ↓ | Moderate (correlated with above) | 35% → ~30% | H |
| Exposure trend 5–7%/yr: the same weather costs more each year | ↑ | Moderate | 30% → ~33% | H |
| 2023 precedent: strong El Niño still closed above $100bn on SCS | ↑ | Moderate | 33% → ~35% | M |
| No major insured event Apr–Jun beyond SCS (verified through early July) | ↓ | Weak-moderate | 35% → ~30% | M |
Posterior before simulation: ~30%. The 50,000-path simulation below, which models the components rather than the aggregate, lands at 27% — consistent with the hand-built update. We publish the Desk probability as ~25–30% (point estimate 27%).
Scenario tree (12-month horizon, probabilities sum to 100%)
| Scenario | Probability | Narrative | Key assumptions | Trigger events | Market implications | What falsifies it | Confidence |
|---|---|---|---|---|---|---|---|
| A. The Quiet Year — 2026 closes below $100bn | 73% | El Niño delivers: no US major landfall, SCS finishes near floor, Q4 stays calm. The streak breaks at six. Softening reads as vindicated; January 2027 renewals price down again; cat bond spreads compress further | H2 hurricane losses < $15bn; no $10bn+ single event | CSU forecast verifies; wetter California autumn | Reinsurer earnings strong again; ILS returns 10%+; but 2027 vintage priced at the cycle's worst risk-reward | A single major landfall | H |
| B. The Floor Holds — $100–150bn | 24% | Exposure growth does what it did in 2023: secondary perils, a moderate landfall or a typhoon/quake surprise carry the year over $100bn despite the quiet forecast | One $15–40bn event OR SCS/flood overshoot | Late-season Gulf hurricane; Japan typhoon; insured EQ | Softening stalls at Jan 2027; spreads stabilize; no capital event | H2 comes in under $55bn | M |
| C. The Snap — above $150bn | 5% | The 17% tail fires: a Milton-class landfall or multi-event cluster. Some 2026-vintage cat bonds attach; June-2026 pricing instantly looks like the cycle's error | Major metro landfall or $50bn+ single event | Category 4–5 US landfall Sep–Oct | Hard market returns at Jan 2027 (+15–30%); ILS marks down; FAIR-plan and Citizens stress returns to headlines | Season ends without a major event | M |
Within Scenario C sits a roughly 0.3% sub-tail — a $225 billion-plus "capital event" (a Katrina-scale metro hit at 2026 exposure values) that would impair alternative capital broadly and reprice every coastal asset class at once. We flag it separately on the watch dashboard because its portfolio consequences are categorically different, but it does not warrant its own line at 5% rounding.
Simulation results
Design summary. We ran a 50,000-path Monte Carlo on calendar-2026 global insured losses. The first half is anchored on observed data: Q1 verified at $20 billion; the second quarter estimated at $18–26 billion from verified components (US SCS running above $22 billion year-to-date by mid-June including the first quarter's $7 billion; a mid-single-digit-billion June outbreak; no major tropical-cyclone or earthquake insured event identified) — H1 is therefore drawn uniformly between $38 billion and $46 billion and is explicitly an estimate pending the mid-July H1 reports. The second half is built bottom-up from five peril components, each calibrated to the 2015–2025 loss history and adjusted for the 2026 seasonal forecasts: US severe convective storm and global secondary perils (lognormal, median $24bn); Atlantic hurricane losses modelled with a frequency-severity distribution informed — but not determined — by CSU's 17% major-US-landfall probability (heavy-tailed when a major landfall fires, median $4bn when it does not); West Pacific typhoon (El Niño modestly raises intensity; median $5bn); autumn wildfire and December European windstorm (fat-tailed, medians $3.5bn and $2bn); and a Poisson earthquake/other jump (~7% per half-year, $8–60bn class). Components are drawn independently — a limitation discussed below.
Results (50,000 paths, seed disclosed in the methodology box):
| Percentile | Calendar-2026 insured losses |
|---|---|
| P10 | $74bn |
| P25 | $80bn |
| P50 | $89bn |
| P75 | $101bn |
| P90 | $119bn |
| P95 | $135bn |
| Threshold | Probability |
|---|---|
| ≥ $100bn (seventh consecutive year) | 26.7% |
| ≥ $150bn (the snap) | 3.0% |
| ≥ $225bn (capital event) | 0.3% |
| < $100bn (the streak breaks) | 73.3% |
Main drivers. The Atlantic mixture contributes most of the variance: conditional on the 17% major-landfall branch firing, the mean path adds roughly $40 billion and the probability of a $100 billion year jumps above 70%. The SCS floor contributes most of the certainty: no simulated path with plausible SCS parameters finishes below $60 billion.
Limitations. Peril components are drawn independently, understating clustered-year risk (2017-style sequences) — this biases the extreme tail modestly downward. The H1 anchor is an estimate until the H1 reports publish in mid-July. Basis matters: Gallagher Re and Aon run $15–25 billion above the sigma basis in recent years; on their basis, the same simulation shifts up by roughly that amount and the ≥$100bn probability rises toward 45–50%. The resolution criterion is deliberately pinned to one basis to keep the question gradeable.
Market pricing vs. the Desk view
What the market prices. Mid-2026 catastrophe pricing embeds three consecutive benign outcomes: rate-on-line down ~16% year-over-year, cat bond spreads below 4% over expected loss for the first time in five years, and record capital chasing the risk. That is a market pricing the quiet year as the central case and, implicitly, pricing the continuation of quiet as the regime.
Where the Desk agrees. On the twelve-month question, the market is probably right: we put roughly 73% on the quiet year, and the softening has a genuine fundamental input — record capital, El Niño, reformed Florida.
Where the Desk disagrees — the highest-conviction mispricing. The market is extrapolating a cycle trough into a trend. The loss line grows 5–7% a year in real terms; the expected-loss models beneath cat bond spreads update slowly and dispute-prone perils (wildfire, SCS) are still being re-parameterized after 2023–2025 surprised them. A 3.74% spread over modeled expected loss is thin compensation if the models understate a compounding trend — and the trend line says the "average" year of 2030 looks like the bad year of 2024. The asymmetric trade is not this year's event risk; it is the January 2027–2028 renewal pricing that a quiet 2026 will produce. Owners adding ILS exposure today at spot spreads are selling insurance at the cheapest point of the cycle against the most expensive decade of hazard on record.
What would prove the Desk wrong. Three more quiet years (hazard genuinely plateauing rather than trending); expected-loss models demonstrating they have caught up (spreads holding while modeled losses rise); or exposure growth decelerating as capital and people actually leave high-hazard geographies at scale.
The universal-owner portfolio map
Strategic implications by asset class — Base = Scenario A (quiet year, softening extends), Upside = disciplined-cycle variant of A/B, Tail = Scenario C/D (snap or capital event). Horizons: 3 months / 12 months / 5 years. This is scenario analysis, not personalized advice.
| Asset class | Base (73%) | Upside (24%) | Tail (3%) |
|---|---|---|---|
| Catastrophe bonds / ILS | 3m: carry strong; 12m: spreads compress further — trim adds; 5y: worst vintage risk since 2011 building | Spreads stabilize; 2026–27 vintages adequate | Attachment losses; marks −10–30%; then the best entry since 2023 |
| Listed reinsurers / P&C | 3m: earnings beats; 12m: ROE fade as pricing declines compound; 5y: reserve adequacy of soft-market years becomes the question | Discipline holds, buybacks | Initial drawdown, then hard-market windfall for the strongly capitalized |
| Real estate (coastal/WUI) | Insurance costs still capitalizing into values — a slow repricing, not a crash; underwrite 2030 premiums into today's cap rates | Policy reform (risk-based pricing accepted) steepens but clarifies the repricing | Post-event non-renewal waves; forced sales in exposed submarkets; $1.47tn long-run value-at-risk estimate moves forward |
| Municipal credit (exposed states) | Spreads underprice FAIR-plan/assessment contingent liabilities | Reform reduces state backstop risk (Florida model) | Assessments and disaster bonds hit exposed issuers; spread decompression |
| Mortgages / RMBS | Premium pass-through raises delinquency at the margin (~200k/yr projection) | Stable | Escrow shock post-event; servicer advances spike in hit geographies |
| Infrastructure (grids, water, ports) | Rising insurance line-items in O&M; resilience capex accelerates — an investable theme | Public co-insurance schemes (EU proposal) de-risk revenue | Uninsured damage to unlisted assets marks down NAVs with a lag |
| Insurance-adjacent private credit | Sidecar/quota-share economics still attractive but thinning | Selective vintages fine | Collateral trapped; liquidity gates |
| Public equities broad | Second-order only | — | Consumer/regional-bank exposure in hit states |
| Sovereign/reserve portfolios | EU-level reinsurance proposal advances — watch for sovereign risk-transfer issuance | New cat-bond sovereign sponsors (record 2025–26 issuance already) | Fiscal backstops activate; contingent liabilities surface on state balance sheets |
| Gulf / Asia sovereign owners | ILS as diversifier at the wrong price point — patience | Build ILS capability for the post-event entry | The entry arrives |
Second- and third-order effects
Risk migrates toward the public balance sheet. Where private cover retreats, exposure concentrates in publicly mandated residual-market pools, industry assessments, policyholder surcharges and — in the most severe cases — explicit government backstops; California's FAIR Plan alone carries over $600 billion of exposure. The next capital event does not just reprice insurance; it converts a private-market problem into a fiscal one, with assessments, post-event bonds and contingent liabilities surfacing in municipal and sovereign credit. The ECB–EIOPA December 2024 proposal for a union-level reinsurance scheme — and a further EIOPA–ESM proposal in April 2026 — is the template: catastrophe risk migrating up the capital structure of the state. For reserve managers and sovereign funds, that is issuance supply — and a new asset class of sovereign risk transfer — arriving within the decade.
Insurance is the pricing mechanism for climate adaptation. The premium is the only annually-repriced, forward-looking climate signal in most real-asset cash-flow models. As pricing softens cyclically, that signal temporarily mutes — precisely when underwriting decisions (a 30-year mortgage, a 50-year port concession, a permanent real-estate allocation) need it loudest. Owners who read the 2026 premium as the 2030 premium are embedding the softest point of the cycle into the longest-dated assets they hold.
The data layer is thinning as the risk layer thickens. The US federal billion-dollar disaster database was retired in 2025; attribution and exposure data are migrating to private and philanthropic providers. For a market that prices tail risk off shared public data, the privatization of loss data is itself a slow-burn systemic issue — models diverge, disputes rise, and the expected-loss number under a cat bond spread becomes more contested, not less.
Adaptation becomes the investable response. A rising loss floor with softening insurance is an arbitrage on resilience: hardened grids, flood defense, wildfire-resistant construction, parametric products, and the data/analytics layer itself. The capital that owns both the risk (ILS) and the mitigation (infrastructure) internalizes the externality — a genuinely universal-owner trade.
What we're watching — the dashboard
| # | Indicator | Current | Threshold that changes the model | Direction |
|---|---|---|---|---|
| 1 | H1 2026 insured-loss reports (Gallagher Re/Aon/Munich Re, due mid-late July) | Q1 $20bn verified; H1 est. $38–46bn | H1 > $55bn → raise P(≥100bn) above 40% | ↑ |
| 2 | CSU/NOAA August forecast updates | ACE ~50 forecast | Upward revision or El Niño stall → raise | ↑ |
| 3 | ENSO index (ONI) through autumn | Strong El Niño developing | Rapid decay toward neutral → raise | ↑ |
| 4 | Any US hurricane landfall watch, Aug–Oct | None | Major (Cat 3+) landfall → Scenario C live | ↑↑ |
| 5 | US SCS running total | >$22bn (mid-June) | > $35bn by Sep → floor scenario strengthens | ↑ |
| 6 | West Pacific typhoon intensity (El Niño year) | Season starting | Jebi-class Japan landfall → +$10–15bn | ↑ |
| 7 | California autumn fire-weather outlooks | El Niño wetter signal | Offshore-wind events with dry fuels → tail risk | ↑ |
| 8 | Cat bond spread over expected loss (quarterly) | 3.74% (Q2 2026) | < 3.25% → mispricing thesis strengthens; > 4.5% → market re-rating | ↕ |
| 9 | January 2027 renewal guidance (Oct–Dec commentary) | — | Double-digit further softening guided → cycle-risk thesis strengthens | ↓ |
| 10 | Cat bond issuance pace H2 | Record H1 $17.3bn | > $25bn full year → capacity glut | ↓ |
| 11 | California FAIR Plan quarterly policy/exposure growth | +2.4% Q1 2026 (slowing) | Re-acceleration → private retreat resuming | ↑ |
| 12 | Florida Citizens depopulation | ~395k policies | Reversal/growth → reform stall | ↑ |
| 13 | EU reinsurance-scheme legislative progress | Proposed Dec 2025 | Adoption → sovereign risk-transfer supply | ↕ |
| 14 | Insurer Q2/Q3 2026 earnings: reserve development on 2023–25 wildfire/SCS | Favorable so far | Adverse development → models behind trend | ↑ |
| 15 | Swiss Re preliminary 2026 sigma estimate (December) | — | The resolution reading | — |
Visual appendix
The two lines (the report in one chart). Property-catastrophe rate-on-line index, 2023–2026: +27.5% → +5.4% → −6.6% → −12% → −16% (mid-year). Against it, insured losses: 2020–2025 all above $100bn, trend 5–7%/yr, expected ~$186bn by 2030. Pricing falls while the floor rises.
The probability bands (simulation output). <$60bn: ~0%; $60–80bn: 25%; $80–100bn: 48%; $100–120bn: 17%; $120–150bn: 7%; $150–225bn: ~2.6%; >$225bn: ~0.3%.
The capital wall. Total global reinsurance capital: ~$790 billion in Q1 2026 (Aon), including ~$141 billion of alternative capital. A separate dedicated-capital series (Guy Carpenter/AM Best) stood near $660 billion in 2025. Different bases — not a single continuous trend.
The retreat map. California FAIR Plan exposure $115bn (2020) → $603bn (2025); Florida Citizens 1.42m policies (2023) → 395k (2026) — the two directions the state-backstop story can go.
The heatmap. The universal-owner portfolio map above, rendered as a color grid.
How this could be wrong
The Desk's probability could be too low. Our H1 anchor is an estimate; if the mid-July reports print H1 near $55 billion (one unverified trade-press figure suggested ~$52bn — we excluded it), the posterior moves to ~40% mechanically. Component independence understates clustering. And the 2023 lesson — El Niño years can still clear $100bn on secondary perils alone — is exactly the kind of regime change that punishes models calibrated to older seasonality.
The cycle thesis could be too bearish on pricing. The softening is from the highest levels in a generation; even −16% leaves rate adequacy above every year from 2014–2022. If reinsurers hold attachment points (the structural win of 2023) rather than competing them away, soft-market vintages may stay adequately priced longer than the 2005–2011 analogue implies.
The retreat narrative has a live counterexample. Florida's reform-driven depopulation shows policy can pull private capital back in. If California's Sustainable Insurance Strategy achieves the same by 2027–28, the "state becomes the reinsurer" chain weakens materially.
The weakest assumption: that expected-loss models under cat bond spreads lag the trend. The ILS market repriced hard after 2022 and again after the 2025 wildfires; spread compression may reflect genuinely better risk selection (higher attachment, more remote layers) rather than complacency. We hold the mispricing view at medium confidence only.
Methodology box
The probability in this report comes from the Desk's editorial scenario model combining base-rate evidence (nine years of published industry loss data across three institutional loss aggregators), seasonal-forecast priors (NOAA, Colorado State University), verified year-to-date loss components, and a 50,000-path scenario-weighted Monte Carlo simulation (seed 20260711) whose design summary — component distributions, calibration sources, the H1 anchor and its status as an estimate, and known limitations — is set out in the Simulation Results section. Inputs are weighted by recency, reliability and institutional materiality. No probability is published unless the evidence that moved it is shown. Signals that did not meet the publication threshold (one unverified H1 trade-press figure; prediction-market pricing, where no liquid market in this cluster met the desk's threshold) were monitored qualitatively and excluded from the formal ensemble.
This report is for informational and research purposes only and does not constitute investment, legal, tax, or financial advice. Editorial scenario analysis only — not investment, actuarial, or geopolitical advice.