Prolepsia forecastYes / no

Will at least one US Atlantic hurricane cause more than 20 billion dollars of insured losses in 2027?

As of 16 November 2026, Prolepsia puts the chance of yes at 27%.

27%chance of yes

Yes 27%No 73%
Prolepsia forecast made

How Prolepsia reasons

The report Prolepsia wrote with this forecast.

How forecasts are made · Track record

TL;DR

The forecast is 26.8% Yes and 73.2% No. A qualifying loss is a substantial minority risk, but the more likely outcome is that no individual hurricane exceeds $20 billion in nominal U.S. insured losses.

Key evidence

  • Industry benchmarks establish a meaningful baseline risk. Karen Clark’s May 16, 2022 account places a $20 billion U.S. insured hurricane loss at approximately a five-year recurrence interval, using then-current replacement costs. That is a useful anchor, not a directly measured probability for this exact 2027 question (Clark).
  • Historical losses support the scale of the risk. An Aon-sourced table identifies eight seasons during 1990–2025 containing a storm above $20 billion after adjustment to 2025 purchasing power. These are historical comparisons, not nominal-dollar qualifying counts or full reconstructions at today’s exposure (Insurance Information Institute).
  • The fixed nominal threshold becomes easier to cross as loss potential grows. Replacement costs, development and insurance coverage support an upward adjustment from older benchmarks. But building resilience, deductibles and coverage contraction can offset that pressure; Swiss Re’s Katrina reconstruction illustrates that modernization does not raise every storm’s insured loss (Swiss Re).
  • Winter El Niño does not establish a quiet 2027 hurricane season. NOAA’s November outlook gives June–August 2027 probabilities of 33.0% La Niña, 61.0% neutral and 6% El Niño. That argues against carrying winter suppression straight into summer, but does not establish peak-season U.S. landfall risk (NOAA).

Uncertainties

Storm tracks and the insured property they encounter dominate the outcome. Basin-wide hurricane counts are an imperfect guide to a single costly U.S. event.

Resolution also depends on measurement. Exactly $20 billion fails; multiple storms cannot be combined. U.S. territories and public insurance count, while foreign and uninsured losses do not. Aon’s applicable estimate takes precedence, using information available by June 30, 2028. Loss development, geographic scope and rounded estimates can therefore decide a borderline result. Peak-season climate conditions, future rebuilding costs and storm-specific coverage remain unknown.

Read the full report

TL;DR

The forecast is 26.8% Yes and 73.2% No. No single qualifying hurricane exceeding $20 billion in nominal U.S. insured losses is the more likely outcome. Treat this as a substantial single-event risk, driven chiefly by where storms travel and how much insured property they damage.

Context

As of November 16, 2026, the 2027 hurricane season has not begun. The official Atlantic season runs from June 1 through November 30, but the question also admits storms outside those dates (NOAA). The relevant event is a North Atlantic cyclone first attaining hurricane intensity during 2027 and causing strictly more than $20 billion in insured losses within the United States, including its territories. Damage during the storm’s later tropical or post-tropical stages counts. A U.S. hurricane landfall is not required.

This is a single-storm threshold, not a test of annual hurricane losses. Private insurance, government-backed insurance and covered business interruption count. Uninsured damage, foreign losses and separately added reinsurance recoveries do not. Resolution uses applicable estimates published by June 30, 2028, with Aon’s event-specific point estimate taking precedence when available. Those rules make insurance coverage and reporting part of the event being forecast, not just details to settle afterward.

Evidence

The historical backbone supports a substantial minority risk. The Aon-sourced U.S. loss table reproduced by the Insurance Information Institute contains the following ten events above $20 billion when expressed in 2025 purchasing power. Its scope includes government insurance, Puerto Rico and the U.S. Virgin Islands. These values are CPI-adjusted historical losses, not losses reconstructed against 2027 buildings and policies (Insurance Information Institute).

Hurricane Year Insured loss, billions of 2025 dollars
Katrina 2005 107.495
Ian 2022 58.854
Ida 2021 42.742
Sandy 2012 42.124
Harvey 2017 39.678
Irma 2017 39.544
Maria 2017 38.822
Andrew 1992 36.883
Ike 2008 27.019
Milton 2024 20.579

These events occupy eight seasons during 1990–2025, or 22% of seasons. Restricting the comparison to 2000–2025 gives seven positive seasons out of 26, or 27%. The windows overlap and are not independent evidence. The distinction between events and seasons also matters: Harvey, Irma and Maria count as one positive season for this question’s annual occurrence test. The table misprints Milton’s year as 2014; Aon’s original report identifies it as 2024 (historical table; Aon report).

This history is a cross-check, not a literal count of past successes under the question’s rules. Inflation-adjusted losses do not establish that a storm exceeded the nominal threshold when it occurred. Nor does CPI adjustment capture the location of new construction, changing insurance take-up or stronger buildings. Older normalized reconstructions also produce lower threshold-crossing counts, showing that the exposure basis and coverage assumptions materially affect the answer (KCC historical reconstruction).

Industry catastrophe-loss benchmarks provide a more direct anchor near the threshold. In her May 16, 2022 account, Karen Clark described U.S. insured hurricane losses of $10 billion approximately every other year, $20 billion at least every five years, $40 billion at a ten-year return period and $150 billion at approximately a hundred-year return period. These benchmarks used then-current replacement costs and climate conditions. They establish that a $20 billion storm is neither routine nor an exceptionally remote disaster (Clark).

A reconstruction by Comola and colleagues adds support. It examines approximately 150 storms during 1950–2022, combines private insured losses with NFIP claims and indexes losses to 2022 conditions. Its comparison with 10,000 simulated years broadly corroborates the national loss scale. But the precise threshold crossing comes from a graph, not a published numerical entry, and readings differ. That limits precision rather than supplying a firm reason to move the forecast. The study also covers the contiguous United States, not the full territorial scope required here (Comola and colleagues).

A separate AIR/Verisk disclosure gives national occurrence-loss benchmarks of $49.3 billion, $151.4 billion and $222.3 billion at ten-, fifty- and hundred-year return periods, using end-2022 exposures. These support the broad scale of the loss tail, but do not directly answer the $20 billion question. Some nearby probabilities in that disclosure are conditional on Jamaican catastrophe-note triggers and are not national annual base rates (World Bank prospectus).

The main reason to look above an older nominal-loss benchmark is growth in loss potential. A fixed $20 billion threshold buys less rebuilding over time, while development adds property at risk. Verisk reports annual North American nominal property-exposure growth of 6.9% during 2020–2024, comprising 1.6% new construction and 5.3% price inflation. That is historical regional evidence, not a forecast of hurricane losses in 2027 (Verisk). The analysis assumes continuing growth in loss potential, but that assumption is not a measured 2027 probability. Mitigation works in the opposite direction: Swiss Re’s Katrina replay shows that current-condition losses need not exceed an inflation-adjusted historical result (Swiss Re).

Climate evidence argues against simply extending winter El Niño suppression through the next hurricane season. NOAA’s November 2026 phase outlook shows the following progression. These are overlapping three-month forecasts, not independent observations or U.S. loss probabilities (NOAA phase outlook).

Three-month period La Niña Neutral El Niño
October–December 2026 0% 0% 100%
November 2026–January 2027 0% 0% 100%
December 2026–February 2027 0% 0% 100%
January–March 2027 0% 0% 100%
February–April 2027 0% 2% 98%
March–May 2027 0% 20% 80%
April–June 2027 1% 61.0% 38.0%
May–July 2027 13% 75% 12%
June–August 2027 33.0% 61.0% 6%

El Niño generally suppresses Atlantic hurricane formation through stronger wind shear. The projected transition removes a strong basis for assuming that suppression persists. It does not establish an unusually damaging season. The outlook stops short of the full August–October peak and crosses the spring predictability barrier (NOAA discussion; predictability explanation).

Insurance changes provide a smaller signal. Citizens’ flood-insurance requirement expands to most remaining eligible personal-residential policies at issuance or renewal from January 1, 2027, subject to exemptions. More coverage can turn otherwise uninsured flood damage into counted losses. This is a limited portfolio change, not a nationwide expansion (Citizens).

What's non-obvious

The question is not mainly about how many hurricanes form. It is about whether one storm intersects enough insured exposure to exceed a fixed threshold. An active basin can avoid expensive U.S. losses. A quieter basin can still deliver the decisive storm. Seasonal climate signals therefore deserve less weight here than their prominence in hurricane coverage suggests.

The reported loss can be decisive near the threshold. Aon’s January 2025 figures put Milton at exactly $20 billion and Helene at $17.5 billion insured. The former does not qualify under a strict “more than” rule. Aon’s later report lists Helene at $21 billion nominal, but that entry includes foreign geography and cannot automatically establish U.S.-only exceedance. This is evidence of loss-development and scope uncertainty, not grounds to select the highest available estimate (Aon 2025; Aon 2026).

Uncertainties

The largest unknown is storm track. Intensity, rainfall, surge and the concentration of insured property along that track determine whether a storm enters the relevant loss range. Basin activity and ENSO do not resolve those details this far ahead.

Coverage is another major gap. Flood insurance, deductibles, limits, business interruption and territorial exposure affect counted losses. Some national benchmarks do not establish their exact NFIP or territorial treatment. Florida residential wind models are useful comparisons, but their geography, deductible conventions and exclusion of direct flooding prevent treating them as national all-lines estimates.

The evidence also lacks a complete annual maximum-loss history reconstructed consistently at modern exposures and at the question’s reporting deadline. Such a series, together with a national occurrence-loss curve covering public insurance and territories, would improve the baseline. Peak-season climate forecasts and updated rebuilding-cost and coverage data would narrow other gaps.

Finally, estimates available by June 30, 2028 may differ from losses established years later. Rounding can leave an otherwise severe event at exactly $20 billion. A major storm with insufficient reliable reporting can lead to annulment rather than an automatic No. These uncertainties remain part of the assessment; the forecast makes no separate deduction for annulment.

Forecast history

How Prolepsia’s forecast has movedOne forecast so far: 27% on 16 Nov 2026.
DateForecast
27% chance of yes

Sources Prolepsia read

  1. Facts + Statistics: Hurricanes - Triple-I®www.iii.org
  2. United Statesassets.aon.com
  3. Presentationwww.casact.org
  4. The Evolution of Catastrophe Modeling Since Hurricane Andrewwww.insurancejournal.com
  5. Increase in insurance losses caused by North Atlantic hurricanes in a warmer climated-nb.info
  6. World Bank prospectusthedocs.worldbank.org
  7. Modeling Insured Catastrophe Losses: Global Perspective for 2025www.verisk.com
  8. 2 **Swiss Re Institute**www.swissre.com
  9. Climate Prediction Center - Official NOAA CPC ENSO Probabilitieswww.cpc.ncep.noaa.gov
  10. The Spring Predictability Barrier: we’d rather be on Spring Break | NOAA Climate.govprod-01-asg-www-climate.woc.noaa.gov

The question

This question asks whether at least one hurricane occurring in the North Atlantic basin during the 2027 Atlantic hurricane season will cause more than US$20 billion in insured losses within the United States, including its territories. The official Atlantic hurricane season runs from June 1 through November 30, although qualifying storms may develop outside those dates. A qualifying hurricane must reach Category 1 intensity or higher at some point during its lifetime, as classified by the National Hurricane Center (NHC), and cause insured losses exceeding US$20 billion from a single storm event. The hurricane need not make landfall in the United States or retain hurricane intensity when causing the damage. Insured losses include covered property damage, business interruption, and other insured losses attributable to the hurricane, including damage from wind, storm surge, and flooding. Losses will be measured in nominal US dollars, without inflation adjustment, using estimates available by June 30, 2028. The threshold applies to each individual hurricane, not the combined losses from multiple storms. Historical experience demonstrates that individual Atlantic hurricanes can generate insured losses of this magnitude. For example, Aon's catastrophe data estimated approximately US$20 billion in insured losses from Hurricane Milton in 2024 and US$17.5 billion from Hurricane Helene. The 2025 Atlantic hurricane season, by contrast, produced no hurricane landfalls in the United States, illustrating the substantial variation in annual hurricane impacts. As of November 16, 2026, the 2027 Atlantic hurricane season has not begun. NOAA identifies June 1 through November 30 as the official Atlantic hurricane season, while the National Hurricane Center maintains the authoritative historical records of Atlantic tropical cyclones. Sources: NOAA (https://www.noaa.gov/prepare-before-hurricane-season), NHC (https://www.nhc.noaa.gov/data/tcr/), Aon (https://www.aon.com/), and Insurance Information Institute (https://www.iii.org/).

How it resolves

The question resolves YES if at least one Atlantic tropical cyclone that attains hurricane intensity during calendar year 2027 causes more than US$20 billion in insured losses attributable to damage within the United States, including Puerto Rico, the US Virgin Islands, and other US territories. A qualifying storm must have been classified as a hurricane (maximum sustained winds of at least 64 knots, equivalent to approximately 74 mph) at some point in its lifetime, according to the NHC's final best-track data. The hurricane does not need to make US landfall, and losses resulting from its subsequent tropical storm, tropical depression, or post-tropical stages count toward the same event total. The insured-loss threshold must be strictly exceeded: exactly US$20 billion does not qualify. Losses are measured in nominal US dollars and include insured property damage, business interruption, and insured flooding or storm-surge losses. Both private insurance and government-backed insurance programs, including the National Flood Insurance Program (NFIP), are included. Uninsured losses, broader economic losses, government disaster assistance, and losses occurring outside US territory are excluded. Reinsurance recoveries are not added separately to underlying insured losses, to avoid double counting. The primary resolution source will be Aon's annual Global Catastrophe Recap or Climate and Catastrophe Insight reporting, using the most recent event-specific insured-loss estimates available by June 30, 2028. If Aon does not publish an applicable figure for a potentially qualifying hurricane, estimates from Gallagher Re, Swiss Re, Munich Re, or recognized catastrophe-modeling organizations such as Verisk or Moody's RMS may be used. Where credible estimates differ, preference will be given to comprehensive estimates covering both private and public insurance, with consistent geographic coverage and without double counting. If the primary source publishes an explicit point estimate, that estimate will take precedence over ranges or estimates from secondary sources. If only a range is available, the midpoint will be used, provided its scope is consistent with the definition above. The question resolves NO if no qualifying hurricane has an insured-loss estimate exceeding US$20 billion by the resolution deadline, provided sufficient reporting exists to assess potentially qualifying storms. If a major potentially qualifying hurricane occurs but reliable insured-loss estimates are unavailable or remain too uncertain to determine whether the threshold was exceeded, the question will be annulled rather than automatically resolving NO. The evaluation will take place after June 30, 2028, using information published by that date. Subsequent revisions will not change the resolution.

Fine print

The question concerns losses caused by individual hurricanes, not the total insured losses from all hurricanes during 2027. A storm that forms outside the official June 1–November 30 season can qualify if it reaches hurricane intensity during calendar year 2027. A storm that forms in 2027 and continues into early 2028 will qualify if it reached hurricane intensity in 2027; insured losses from its entire continuous lifecycle will count, even if some damage occurs in 2028. Storms first attaining hurricane intensity in 2026 or 2028 do not qualify. Losses must be attributable to the qualifying storm, including its associated wind, rainfall, inland flooding, coastal flooding, and storm surge. Losses from unrelated weather events occurring around the same time must be excluded. The geographical definition of the United States includes all 50 states, Washington, DC, Puerto Rico, the US Virgin Islands, and other US territories, but excludes foreign countries affected by the same hurricane. Financial figures are expressed in nominal US dollars and will not be adjusted for inflation. Published insured-loss estimates are acceptable; actual final claims settlements are not required. If a source reports global insured losses for a storm, only the portion attributable to US locations will count. If a report provides a range that straddles US$20 billion and no reliable point estimate is available, the midpoint of that range determines the result. If multiple credible estimates with comparable coverage remain materially inconsistent, the primary source takes precedence when available; otherwise, the question may be annulled if a defensible determination cannot be made. Sources published after June 30, 2028, will not affect the result.