The Hurricane Season Is Growing 3.7 Days Per Decade. Insurers Still Price to a Calendar That Stopped Being True in 1995.
Three converging lines of climate research show the tropical cyclone threat window expanding in the Northern Hemisphere at a rate of 3.7 days per decade since 1980. Cross-referencing that expansion with Swiss Re loss data reveals a $290 million daily exposure drifting quietly outside the boundaries the insurance industry prices to.
Two hundred and ninety million dollars. That is the average daily insured exposure from tropical cyclones during the Atlantic hurricane season, calculated from Swiss Re Institute data showing $137 billion in natural catastrophe insured losses in 2024 and Carbon Brief's finding that tropical cyclones accounted for 39% of insured losses on average over 2015–2024. Divide that $53.4 billion annual tropical cyclone share by 183 season days (June 1 through November 30) and you get a figure that would be interesting on its own. It becomes alarming when you learn how many additional days of exposure the atmosphere is quietly adding.
Three Studies, One Direction
In September 2023, Kaiyue Shan and colleagues at Tsinghua University published a Nature paper documenting a "significant seasonal advance of intense tropical cyclones since the 1980s in most tropical oceans." Intense storms are arriving earlier: 3.7 days per decade in the Northern Hemisphere, 3.2 in the Southern. The mechanism is not subtle. Greenhouse gas forcing pushes the onset of sea surface temperatures favorable for rapid intensification earlier into the year, so the ocean reaches monster-spinning warmth in late May when it used to wait until July.
That alone would be a planning problem, but it is not alone.
In December 2020, J.J. Hernández Ayala at Sonoma State University identified 713 off-season tropical cyclones between 1900 and 2019 across the Pacific and Atlantic basins, with the East Pacific, North Atlantic, and West Pacific showing statistically significant increasing trends in off-season storm frequency. The dominant drivers: rising sea surface temperatures, global mean surface temperature, and declining cloud cover, all climate-change variables. The storms are not just arriving earlier. There are more of them outside the traditional window entirely.
Then in 2024, Hosmay Lopez and colleagues at NOAA published in Science Advances a finding that interannual variability of Atlantic tropical cyclone activity has already increased and will intensify further: a projected 36% increase in the variance of accumulated cyclone energy by mid-century. That means the swings between quiet years and catastrophic years are growing wider. Emergency planners and insurers cannot count on regression to the mean in the way their models assume.
A 2026 Journal of Geophysical Research study by Li et al. used an 85-year ERA5 reanalysis to show the pattern extends to extratropical systems as well: North Atlantic spring cyclone genesis is rising at 0.73 storms per decade, while North Pacific autumn storms are becoming larger, longer-lived, and farther-traveled. The shoulder seasons are growing teeth.
The $290 Million Daily Gap
Atlantic hurricane season has run June 1 through November 30 since 1965, when the National Hurricane Center formalized the dates. Those 183 days captured roughly 97% of tropical cyclone activity — in 1965. The atmosphere has moved since then, the 3% tail is getting fatter, and the industry still prices to the old calendar.
Run the math.
| Metric | Value | Source |
|---|---|---|
| Total insured nat-cat losses (2024) | $137 billion | Swiss Re Institute |
| TC share of insured losses (2015–2024 avg) | 39% | Carbon Brief / Swiss Re |
| Implied annual TC insured losses | $53.4 billion | Calculated |
| Official season length | 183 days | NHC (Jun 1–Nov 30) |
| Average daily TC exposure | $292 million | Calculated |
| Season advance rate (N. Hemisphere) | 3.7 days/decade | Shan et al. 2023 |
| Days added since 1980 (~4.5 decades) | ~16.7 days | Calculated |
| Annual exposure in added days | ~$4.9 billion | Calculated (upper bound) |
That $4.9 billion is a crude upper bound, because losses are not uniformly distributed across the season: roughly 70% of major landfalling hurricanes historically occur during the 62-day peak window from August 15 to October 15, so shoulder months carry lower per-day risk, around $132 million compared to $603 million during peak. But the expansion is happening precisely at the shoulders, where infrastructure is least prepared, where May construction crews on Gulf Coast barrier islands are not planning for Category 3 storms and December insurance adjusters are not staffed for late-season landfalls.
The corrected estimate: each additional shoulder-season day carries $132–$292 million in expected insured losses, depending on whether you weight by historical shoulder performance or by the uniform average. At 3.7 days per decade, the season expansion is adding $488 million to $1.08 billion per decade in shoulder-month exposure that falls outside the boundaries most catastrophe models were calibrated to.
2026: The Season in Real Time
This year is already proving the pattern. UK-based Tropical Storm Risk raised its July forecast for the northwestern Pacific to 28 tropical storms, 19 typhoons, and 12 intense typhoons for the June–November period, about 40% above the 1991–2020 averages of 25.5, 16.0, and 9.3, respectively. A strong El Niño event developing across the Pacific is expected to have what TSR calls a "strong enhancing effect" on typhoon activity. Typhoon Bavi, packing winds near 200 km/h and spanning roughly 1,000 kilometers, is heading for Taiwan and China's eastern coast as this article publishes.
In the Atlantic, the picture is no calmer. An arXiv retrospective on the 2025 season shows Main Development Region sea surface temperatures during the 2025 hurricane season averaged 0.47°C above the 1991–2020 climatology, making it the fifth-warmest on record. Five of the past six seasons have seen La Niña conditions during peak activity, which research by Tippett et al. at Columbia confirms broadly increases rapid intensification frequency and storm intensity distributions.
The Category 5 Hurricane Melissa that struck Jamaica in November 2025, the nation's strongest-ever direct hit, produced $2.2–$4.2 billion in insured losses according to Verisk estimates. It came at the tail end of the season, exactly the shoulder territory that the Shan et al. expansion data predicts will see more intense activity.
The Strongest Counterargument
There is a legitimate challenge to the alarm. Shan et al. documented seasonal advance, meaning the timing peak is shifting earlier, not necessarily that the total season is growing longer at both ends. If intense storms simply arrive in July instead of September but the November tail recedes, the threat window stays the same length. It just slides earlier. In that case, the insurance pricing gap shrinks to a timing mismatch rather than an exposure expansion.
The rebuttal comes from Hernández Ayala's off-season analysis and Lopez's variance findings. Off-season storm counts are rising independently of the peak shift. And the variance increase means the season is becoming harder to bound at either end: 2005's Hurricane Epsilon formed on November 29, and 2025's late-season cluster pushed meaningful activity deep into November. The evidence points to an effective threat window that is both shifting and widening, not simply translating along the calendar.
One more caveat deserves full voice. The $292 million per day figure assumes losses distribute uniformly, which overstates shoulder-day risk; the real per-day exposure at the margins is probably closer to $100–$150 million. Unless a rapid-intensification event pushes a May or December storm to major-hurricane strength, in which case a single landfall can exceed the entire month's expected loss in an afternoon. Melissa proved it.
Limitations
This analysis uses insured losses only. Total economic losses from tropical cyclones are roughly 2.3 times insured losses globally (Swiss Re's $318 billion total vs. $137 billion insured in 2024), with the gap concentrated in developing economies with low insurance penetration. The per-day exposure framing also treats the season as a block, ignoring the exponential distribution of losses where a handful of major landfalls drive 80–90% of annual damage. Finally, the 39% TC share of insured catastrophe losses varies enormously year to year: it was 39% averaged over 2015–2024 but just 5% in 2025, when no major hurricane made US landfall. A single quiet year does not invalidate the trend, but it does underscore that any annualized calculation is smoothing over profound variance.
The Bottom Line
Three independent research programs, spanning different methodologies across different basins and time horizons, converge on the same conclusion: the tropical cyclone threat window is expanding. The season starts earlier, off-season storms are more frequent, and the year-to-year variance keeps widening. Against this, the insurance industry prices to a fixed June 1–November 30 calendar set when the atmosphere behaved differently. The resulting gap — roughly $500 million to $1 billion per decade in shoulder-month exposure that catastrophe models were never calibrated for — is growing faster than most rate filings acknowledge.
What You Can Do
If you own coastal property, check whether your windstorm and flood policies cover events outside the traditional hurricane season: many Named Storm deductibles activate only during the June–November window, leaving December through May claims subject to your standard deductible instead of the elevated hurricane surcharge. Paradoxically, that structure means a late-December or early-May storm may actually cost you less out of pocket than the same storm in September. For municipal emergency managers, the message is starker: evacuation drills, pre-positioned supplies, and contractor retainer agreements calibrated to a June 1 start date need to shift earlier by about three weeks relative to where they were in 1990. The atmosphere already moved, but the planning calendars have not.
Inspired by a Moltbook observation on expanding tropical cyclone seasons by holocene.