Global property catastrophe rates on line fell 16% through the July 1, 2026 renewal and now sit 23% below the 2024 hard market peak, the steepest annual decline since the late 1990s. Reinsurer ROE is tracking 14-15% for 2026, down from roughly 19% in 2025, against a sector cost of equity of 10-12%. Howden Re has put the next step on record: a comparable decline in 2027 would push large market segments below cost of equity.

Key Takeaways

  • Reinsurer ROE of 14-15% against a 10-12% cost of equity leaves 200 to 500 basis points of economic value added, and that spread assumes a normalized second half.
  • The global index sits 32% above its 2017 low, but APAC is roughly 3% below its own 2018 regional floor, so the cushion argument does not travel.
  • Another 16% cut compresses the combined ratio buffer by roughly 450 to 600 basis points, landing 2027 ROE at 9-11% against a 10-12% cost of equity.
  • Alternative capital closed 2025 at $136 billion, up 18% and about 17% of the $785 billion total, and it does not withdraw the way traditional capacity does.
  • First-half insured catastrophe losses were $38 billion, the lowest in more than a decade, against full years above $100 billion in each of the past four.

The Cushion Is Not One Number

The case for continued adequacy rests on the global Guy Carpenter property cat index still sitting 32% above its 2017 soft market low. That floor is a meaningful reference: at the 2017 trough several well-regarded reinsurers were earning at or near cost of capital on property cat, and the post-Irma correction built up from there. Two years of double-digit declines have not returned to it.

The composite hides three different positions. The US index fell 16% in 2026 and 22% from the 2024 peak but sits approximately 62% above its own 2017 low, with considerably more runway than the global figure suggests. Florida Citizens gave the cleanest single-program benchmark, its net property cat rate on line falling to 8.46% at the June 2026 renewal from 11.95%, a 29% decline in one year.

APAC is the case that qualifies every global average. The APAC index fell 19% in 2026 and now sits roughly 3% below its prior 2018 regional soft market low. For a book concentrated in Japan, Australia, or Southeast Asia, the bull case built on a 32% cushion does not apply, because that market has already crossed its own floor heading into peak West Pacific typhoon season.

Region / Index 2026 YTD ROL Change Change from 2024 Peak vs. Prior Soft Market Low
Global (Guy Carpenter) -16% -23% +32% above 2017 low
United States -16% -22% ~+62% above 2017 low
Asia Pacific -19% n.a. ~3% below 2018 regional low
Europe (Jan 1 basis) -15% n.a. Above prior soft market

Dean Klisura of Guy Carpenter attributed the environment to "benign loss activity, ample reinsurer capacity and strengthening risk appetite." The reductions came broadly across tower levels rather than concentrating in upper layers as early soft markets usually do, which says the capital overhang is competing at every attachment point at once.

Running the 2027 Arithmetic

The 2026 position is comfortable. At 14-15% ROE against a 10-12% cost of equity, with Gallagher Re's own estimate at 11.7%, the sector is generating roughly 200 to 500 basis points of positive economic value added, conditional on a normalized second half.

The forward step is where it tightens, and it tightens through the combined ratio rather than only the top line. Another 16% reduction shrinks premium on the same exposure base by about 16% while the expected loss on that exposure is unchanged, so the margin above expected loss cost compresses rather than simply scaling down. Run against current levels, that is roughly 450 to 600 basis points of combined ratio buffer depending on the leverage of the book, putting projected 2027 ROE at 9-11%.

Set that against a 10-12% cost of equity and the 2027 outcome sits between a 100 basis point surplus and a 300 basis point deficit. Howden Re's formulation is precise about this: a comparable rate decline in 2027 would push large market segments below cost-of-equity estimates. It is not a claim that the whole market crosses at once. It is a claim that the distribution of 2027 outcomes centers on the threshold.

Attachment discipline is the offset holding the arithmetic up. Per-occurrence attachment points rose sharply through 2022 and 2023 and have stayed nominally unchanged through subsequent renewals, so falling rates reduce the cost of protection without proportionally increasing the loss cost the layer absorbs. KBW notes that structural floor is why 2026 projects among the better years in catastrophe reinsurance history on absolute underwriting income. The exposure is a 2027 renewal that delivers rate pressure and attachment pressure together, which would compress capital returns faster than the rate arithmetic alone implies.

The Supply Floor Does Not Behave Like Capital

Alternative capital closed 2025 at $136 billion, up 18% on the year and roughly 17% of the $785 billion total. How that capital responds to a mediocre year is what makes this cycle different.

A 10-11% ROE year at a traditional reinsurer starts a board conversation about redeployment and appetite. The same return on a dedicated ILS fund produces an investor letter against a committed multi-year allocation held for portfolio diversification. The uncorrelated return stayed positive and the correlation thesis is intact, so the investor does not leave. That means the supply side at January 2027 looks close to today's configuration whether the second half of 2026 delivers $50 billion or $120 billion of losses.

The consequence is a correction that arrives slowly if it arrives through capital rather than losses. Institutional investors in committed multi-year structures cannot reduce appetite at a renewal the way a traditional reinsurer can, so below-cost-of-capital pricing can persist longer once established than a cycle model calibrated on traditional capacity would predict.

Retrocession is the fragile part of that structure rather than the sturdy one. Retro fell 16.5% risk-adjusted at January 1, 2026, tracking primary property cat in near-parallel rather than diverging as constrained retro capacity sometimes does, so the relative economics of buying versus running net barely moved while the absolute cost fell. KBW flagged the concentration directly: a relatively small number of retro players changing their view of profitability after a significant unmodeled catastrophe loss could upset dynamics that would ripple into the primary reinsurance market.

That is the H2 exposure in one sentence. First-half insured losses were $38 billion, concentrated at $22 billion in US convective storm with no US hurricane landfall, against full years above $100 billion in each of the past four. KBW's reversal thresholds are a $60-70 billion single event, a $35 billion event concentrated in lower layers, or $50-60 billion distributed across several. Those sit in the central portion of the second-half conditional distribution, not its tail, and a retro market that reprices inside that window does so when replacement capacity is hardest to find.

Further Reading

Sources

  1. Guy Carpenter, “Global and US Property Cat Rates Down 16%, APAC 19% After July Renewals,” Artemis, July 2026
  2. Guy Carpenter, “Global Property Cat Rates Down 16% as Softening Extends into July Renewals,” Reinsurance News, July 2026
  3. Gallagher Re, “Record Capital Drives Softer Reinsurance Pricing at July Renewals,” Insurance Business, July 2026
  4. Aon, “Alternative / ILS Reinsurance Capital Grew 18% to $136bn in 2025,” Artemis, 2025
  5. Howden Re, “Property Cat Reinsurance Down 14.7%, Retrocession Down 16.5% at Jan 2026 Renewals,” Artemis, January 2026
  6. KBW, “KBW Expects Property Cat Rate Declines Approaching 20%, Highlights Retro Vulnerability,” Artemis, 2026
  7. Gallagher Re, “Reinsurers Face Pricing Pressure After Capital Climbs to $648 Billion,” Insurance Business, 2026
  8. Swiss Re, Global Insured Natural Catastrophe Losses Review, Insurance Journal, 2026
  9. Guy Carpenter, U.S. Property Catastrophe Rate-On-Line Index, Artemis (current)