The Atlantic Has No Hurricane by September 21 for the First Time Since 1914, and Reinsurers Are Pricing 2027 on It

The Atlantic basin passed September 21, 2026 without producing a hurricane, the first time that has happened since 1914, and only 1907 and 1914 have finished a season without one. Six named storms have formed, Arthur through Fay, and the strongest, Fay, peaked at 70 mph southwest of the Azores on September 20 before weakening (NHC advisory archive). Colorado State's Philip Klotzbach put accumulated cyclone energy through the September 10 climatological peak at the lowest of any season since 1950.

Nobody disputes the cause. NOAA's Climate Prediction Center issued an El Nino Advisory with the Nino-3.4 anomaly at +1.8°C in August and gives better than a 90% chance of a very strong event through winter 2026-27, and a 75% chance the October-December reading exceeds every El Nino since 1950. What is in dispute is what a year like this is worth to a property catastrophe reinsurer setting January 1 terms, because the surveys published this week say it is worth a rate cut.

Key Takeaways

Six named storms, zero hurricanes through September 22: Arthur (Texas, June 16-18), Bertha (Florida Panhandle, 60 mph), Cristobal, Dolly, Edouard (Port Arthur, Texas, September 1) and Fay (70 mph, open Atlantic). The satellite-era record for latest first hurricane fell on September 11.
NOAA cut its outlook on August 6 to 7-13 named storms, 2-6 hurricanes and 0-2 majors with a 75% probability of a below-normal season, up from 55% in May. CSU's August 5 update called for 9 named storms, 4 hurricanes and 1 major, against 1991-2020 averages of 14.4, 7.2 and 3.2.
16% was CSU's August probability of a major hurricane landfall on the US coast for the rest of 2026, against a long-run average of 43%. That conditional figure is the right way to price this season, and it does not transfer to 2027.
60% of reinsurers in Fitch's September survey expect property cat rates to fall in 2027, a fifth of them by more than 10%, and 86% expect terms and conditions to loosen. Cat bond spread over expected loss fell to 3.74% in Q2, the first quarter under 4% in five years.
billion of insured catastrophe loss landed in the first half with no hurricane at all, because severe convective storm is 40% of the modeled average annual loss and sits below most reinsurance attachments. The quiet year is quiet in the layer reinsurers price, not in the layer they are being asked to lower.

What the Season Has Produced Against What Was Forecast

Every forecast converged on a weak season, and the season undershot them anyway. NOAA's May 21 outlook called for 8-14 named storms, 3-6 hurricanes and 1-3 majors, its first below-normal call since 2015, with a 55% probability of a below-normal season (NOAA). Its August 6 update moved that probability to 75%, with 20% near-normal and 5% above, and lowered the ranges to 7-13, 2-6 and 0-2 (NOAA update). Through September 22 the basin sits at six, zero and zero.

Colorado State's path was steeper. Its April 9 outlook called for 13 named storms, 6 hurricanes and 2 majors; June cut that to 11, 5 and 2; the August 5 final update reached 9, 4 and 1, with the season's ACE at 2.6 to that date, the least active start since 2009 (CSU August forecast). A line in that report, that the 17 ACE units recorded in 1983 remain the lowest full-season total since 1950 reads differently now than it did in August, because 2026 has a plausible path to challenge it.

Storms that did form were rain events. Arthur developed near the middle Texas coast on June 17 as a 40-mph tropical storm with a headline warning of life-threatening flooding across the southeastern United States, and NOAA's National Weather Service director used it in August to make the point that inland rainfall is the major damage mechanism in tropical storms. Bertha reached 60 mph on July 21 west-southwest of Panama City. Edouard came ashore near Port Arthur, Texas on September 1 at 50 mph. None of these is a wind loss to a reinsured tower.

Outlook	Named storms	Hurricanes	Majors	Below-normal odds
1991-2020 average	14.4	7.2	3.2
CSU, April 9	13	6	2
NOAA, May 21	8-14	3-6	1-3	55%
CSU, June	11	5	2
CSU, August 5	9	4	1
NOAA, August 6	7-13	2-6	0-2	75%
Observed to September 22	6	0	0

How a Conditional Probability Gets Priced Into an Unconditional Treaty

CSU's 16% probability of a major hurricane landfall on the US coast, against a 43% long-run average, is a conditional number: the odds given a strong El Nino in place for the peak months. That is the correct way to price the 2026 season, and a reinsurer that wrote a June 1 Florida program at the unconditional rate collected roughly two and a half times the frequency-adjusted loss cost on the major-landfall component. Margin on that book is a windfall from a known conditioning variable, and it has already been earned.

January 1, 2027 treaties cover the 2027 season, and the conditioning variable does not carry. CPC's outlook runs through winter 2026-27 and stops there; nothing in the September 10 discussion speaks to the state of the Pacific in August 2027. A modeled average annual loss, Verisk's billion global and billion US, is the unconditional expectation, and one season with zero hurricanes is a draw from inside that distribution rather than a revision to it. An AAL a cedent presents in January will be the same number it presented last January, adjusted for exposure growth.

Price is moving anyway. Guy Carpenter's July 1 index had global property cat risk-adjusted rates down 16% year to date and 23% from the 2024 peak, post-Monte Carlo commentary put the January cut near 10%, and Fitch's September 23 survey has 60% of reinsurers expecting further declines in 2027, with a fifth of that group expecting more than 10% (Artemis on Fitch). On the ILS side, Sage Advisory reports the Q2 cat bond spread over expected loss at 3.74%, the first sub-4% quarter in 20, on new issuance priced at 6.9% against a 3.2% modeled expected loss.

Capital, not risk, is the mechanism connecting the quiet year to the rate cut. A hurricane-free peak season leaves reinsurers' 2026 retained earnings intact and the cat bond market's 7.6% year-to-date return unimpaired, so the supply of limit at January 1 rises while modeled demand for it does not. A pricing actuary documenting a rate below model indication is documenting a supply decision, and the correct memo says so rather than citing the season.

The Quiet Year Was Not Quiet in the Layer Being Loosened

Swiss Re's first-half figure is the complication. Insured natural catastrophe losses ran billion in the first six months of 2026, 16% below the ten-year average, with no hurricane in the basin at all. Severe convective storm carries 40% of Verisk's modeled average annual loss, and that peril's loss shape sits largely below the attachment points of occurrence programs, in the primary carrier's retention and in aggregate covers. El Nino suppresses tropical Atlantic wind; it does nothing to Midwestern hail.

Fitch's survey says where the softening is heading: 86% of respondents expect terms and conditions to loosen, 58% selectively and 28% broadly, and the agency names lower attachment points and broader event definitions as the forms it expects. Both moves bring the frequency perils that 2026 did not suppress into the reinsured layer, on a renewal being justified by the suppression of a different peril. An occurrence tower that a record El Nino protected in 2026 is the one being repriced; the aggregate and low-attachment layers that the billion first half actually hit are the ones being widened.

That leaves the 2027 book carrying a lower rate on the peril the model still expects at full weight, and a wider definition of loss on the peril that produced this year's claims. This season will enter the loss history as a zero in the hurricane column, and the burn-cost methods that weight recent years will read it as evidence. Its El Nino will not be in the record at all.

The Atlantic Has No Hurricane by September 21 for the First Time Since 1914, and Reinsurers Are Pricing 2027 on It

Key Takeaways

What the Season Has Produced Against What Was Forecast

How a Conditional Probability Gets Priced Into an Unconditional Treaty

The Quiet Year Was Not Quiet in the Layer Being Loosened

Further Reading

Sources