AM Best: Record Capital Tests Reinsurance's 87% Combined Ratio

The US and Bermuda reinsurance combined ratio improved from 104.4% in 2020 to 87% in 2025, and dedicated global reinsurance capital is estimated at roughly $705 billion for 2026 (AM Best, August 10, 2026).

That combination looks like durable discipline. It is closer to a reserve cushion and a quiet catastrophe year doing the work that current-year rate no longer can.

Key Takeaways

• AM Best puts the US and Bermuda reinsurance combined ratio at 87% for 2025, down from 104.4% in 2020, as dedicated global reinsurance capital reaches an estimated $705 billion for 2026.
• The 87% is a lagging indicator. Hard-market premium written in 2023 and 2024 is still earning through, while property catastrophe rates fell 15% to 20% at the April and midyear 2026 renewals.
• Benign catastrophe experience is masking net price erosion. Swiss Re's large natural catastrophe losses ran nearly $670 million below budget in H1 2026, while its effective net price change was negative 4.6%.
• The real question is whether casualty and property loss picks for accident years 2023 through 2025 hold genuine margin. Munich Re's 24.9% H1 2026 reinsurance ROE, flagged by its own management as unsustainable, shows how much reported results lean on that cushion.
• Capital growth has come from retained earnings rather than new entrants, a partial brake on softening. And AM Best's 79.2% European Big Four figure is flattered by IFRS 17 discounting, so it is not comparable to the US GAAP 87%.

What AM Best's Report Covers

"Global Reinsurance at an Inflection Point: Can Discipline Survive the Temptation of Record Capital?" opens AM Best's run of segment reports ahead of the Monte Carlo Rendez-Vous de Septembre (AM Best, August 2026). It reads as a multi-year series rather than a single data point.

The trajectory it tracks begins with a violent two-year swing. The segment's return on equity was roughly negative 3% in 2022, when the inflation shock and rising rates hammered bond portfolios and cost carriers underwriting margin at the same time. It reached roughly 23% in 2023, the first full year of post-Ian, post-inflation-shock hard-market pricing.

Since then the pattern has been consistent: combined ratios falling, capital rebuilding, and now a fourth consecutive year of double-digit returns. That is exactly the environment that pulled the last hard market back into competitive behavior.

Michael Lagomarsino, a senior director at AM Best, put the open question directly: "If underwriting discipline and pricing integrity can be maintained despite record levels of capital, the industry may indeed be in the midst of a meaningful evolution of the reinsurance market" (AM Best, August 2026).

What Is Actually Carrying the 87%

The Earned Book Lags the Written Book

An 87% combined ratio sounds like a market still pricing well above its cost of capital. On a raw earned-premium basis it is. But a reinsurer's earned book in any given year is a blend of vintages.

Premium written in 2023 and 2024 at genuinely hard-market rates is still running off through the earned side of the ledger in 2025 and 2026. The rate being written today has already turned soft.

Guy Carpenter's renewal data shows January 1, 2026 property catastrophe placements posting double-digit rate declines, with dedicated reinsurance capital growing an estimated 9% to $663 billion in 2025 on the back of retained earnings (Guy Carpenter, December 2025). AM Best's April and midyear 2026 renewal data pushes the same point further: property catastrophe rate reductions, Florida in particular, widely estimated at 15% to 20%.

An earned 87% calculated against that blended book is not the number a reinsurer would post if every dollar of premium were written at 2026 rates. It is a lagging indicator flattered by the hard-market vintages still working through the earned line.

A Quiet Catastrophe Half-Year

Benign catastrophe experience is doing a second, separate piece of the work. Swiss Re's Property & Casualty Reinsurance business posted a 76.7% combined ratio for the first half of 2026, improved from 81.1% a year earlier (Swiss Re, H1 2026 results). Large natural catastrophe losses came in at $169 million against a budget of $836 million, a shortfall of nearly $670 million relative to plan.

At the same time, Swiss Re raised its loss assumptions by 4.4% at renewal against a nominal price decrease of 0.2%. That combination pushed the effective net price change to negative 4.6%. The rate actually being charged on renewing business is falling faster than the headline renewal figures suggest.

A quiet catastrophe half-year is masking a renewal book whose net-of-loss-trend pricing is already moving the wrong direction. That is the exact mechanism actuary.info flagged when Swiss Re's loss-pick increases first surfaced this cycle: see Swiss Re's H1 2026 loss assumption increases against a softening price environment.

The Reserve-Cushion Mechanism

The part of the AM Best series that deserves more scrutiny than the headline ratio is what sits underneath it: the adequacy of the casualty and property loss picks booked for accident years 2023 through 2025. Those are the vintages a hard market was supposed to reserve conservatively, building in margin against the kind of severity surprise that hit the 2019-and-prior casualty years across the industry.

Two outcomes follow from that reserving question:

• If the margin is real, reinsurers can release it gradually over the next two to three years as those accident years mature, propping up reported combined ratios even as newly written business earns in at thinner margins.
• If the 2023 through 2025 casualty picks were set closer to best estimate than to a genuinely conservative hard-market load, the 87% figure has less runway than it appears to. There is no reserve cushion left to smooth the transition once current-year rate adequacy erodes further.

Dan Hofmeister, a director at AM Best, framed the stakes in long-tail terms: "Casualty exposures often develop over many years, meaning that decisions being made today may not be fully understood until well into the next decade" (AM Best, August 2026).

Munich Re Shows the Dependence

Munich Re's Reinsurance division generated a 24.9% return on equity in the first half of 2026, a figure management itself flagged as unsustainable at that level. It did so while cutting full-year reinsurance revenue guidance to roughly €38 billion from €40 billion on softer renewal volumes (Munich Re, H1 2026 results, and Investing.com, August 2026).

A 24.9% return generated in a half where the company is simultaneously cutting its own revenue outlook for softer pricing signals that current results run well ahead of what current-year underwriting alone would deliver. Reserve releases and low catastrophe activity fill the gap. actuary.info traced the same mechanism from the volume side in Munich Re's July 2026 revenue guidance cut.

IFRS 17 Discounting Flatters the European Numbers

AM Best puts the discounted combined ratio for the European "Big Four", Munich Re, Swiss Re, Hannover Re and SCOR, at 79.2% for 2025 under IFRS 17. The US and Bermuda group reports in the mid-80s to low-90s range under US GAAP. That gap is partly real underwriting performance and partly an artifact of how the two accounting regimes treat the time value of money, a distinction the report's framing does not fully separate.

SCOR's H1 2026 disclosure quantifies the mechanism. The IFRS 17 discount effect on its property and casualty result ran to negative 10%, reflecting benign catastrophe activity and higher locked-in interest rates, against an attritional loss and commission ratio of 76.8% before that discount benefit (SCOR, H1 2026 results, July 2026).

A 10-point discount benefit on a book already running an attritional ratio in the high 70s is not a rounding error. It is a meaningful share of the reported margin, and it is baked into every European Big Four combined ratio AM Best compares against the undiscounted US GAAP figures. Reinsurance News has previously estimated the discounting gap between IFRS 17 and comparable non-discounted bases at roughly eight points across the peer group in prior years, broadly consistent with what SCOR just disclosed directly.

The practical consequence: the 79.2% European figure and the 87% US and Bermuda number are not measuring the same thing. The European figure is flattered by a mechanical accounting benefit that has nothing to do with underwriting discipline. actuary.info's earlier read on SCOR's discount mechanics laid out the arithmetic in more detail: see SCOR's H1 2026 IFRS 17 discount effect.

Retained Earnings, Not New Capacity

Dedicated reinsurance capital has grown from $607 billion in 2024 to an estimated $705 billion for 2026. But the report notes that, unlike prior hard-market cycles, the growth has come almost entirely from retained earnings compounding inside existing organizations rather than from a wave of new entrants raising fresh third-party capital. Gallagher Re's own tally puts global reinsurance capital at a similarly elevated record; actuary.info covered that figure and its cedant-facing implications in Gallagher Re's record dedicated capital estimate.

The distinction matters for where pricing floors sit:

• New capacity chasing returns tends to price aggressively to win share, then exits just as quickly when returns compress. That is the classic soft-market overshoot pattern.
• Capital built from existing carriers retaining their own hard-market profits behaves differently. Those balance sheets do not need to write business at inadequate rates to justify their existence, and a management team defending a return on equity in the 20s has less incentive to chase volume at thin margins than a new sidecar or start-up trying to prove a business model.

That is a real, if partial, brake on how far the current softening runs, distinct from the underwriting-discipline question AM Best's headline poses.

What the Fitch and Guy Carpenter Numbers Add

Two adjacent data series corroborate the reserve-cushion read rather than contradict it:

• Fitch put the Big Four's average return on equity at 21.5% for the first half of 2026, matching the record set a year earlier, while flagging that consecutive rounds of renewal price reductions have not yet fully worked through into reported earnings. actuary.info's coverage of that report is at Fitch's H1 2026 Big Four ROE and price feedthrough.
• Guy Carpenter's own forecast shows the industry-wide return on equity easing from 17.6% in 2025 to a projected 16.6% for 2026, a modest decline consistent with a market still running well above its cost of capital rather than one already in distress (Guy Carpenter, 2026 outlook).

None of these figures show discipline breaking down in 2026 itself. Taken together with AM Best's own series, they show a market whose reported profitability is still elevated enough to absorb one or two more years of rate softening without returns turning negative, provided the reserve cushion booked in 2023 through 2025 turns out to be real. actuary.info's broader look at where that softening runs into a genuine cost-of-capital floor is at the reinsurance soft-cycle cost-of-capital threshold.

The Casualty Parallel

The parallel to primary casualty lines is not incidental. General liability writers posted an industry combined ratio near 114% in 2025 on picks that regulators and rating agencies increasingly view as inadequate for the accident years being written (see actuary.info's general liability rate-inadequacy coverage).

Reinsurers sit on top of that same casualty severity trend through their treaty books. That is exactly why the 2023-through-2025 reserve vintages carry more analytical weight in this report than the headline combined ratio does. A primary market still finding its casualty picks wrong in real time is not a market whose reinsurers can safely assume their own casualty reserves from the same period are conservative by default.

What the Series Will Show Next

AM Best's own framing puts the real test at twelve months out, and that is the right horizon. Reserve adequacy for 2023-through-2025 casualty accident years will not be visible in a single renewal season. Three markers will show sooner:

1. Whether the 15% to 20% property catastrophe rate reductions AM Best documents at the April and midyear 2026 renewals extend into the January 2027 renewal at a similar pace.
2. Whether reinsurers respond by holding attachment points and terms even as headline rate falls, the discipline signal AM Best's report is explicitly asking about, or by loosening both rate and terms simultaneously, the pattern that characterized the last soft-market cycle.
3. The combined ratio the industry reports for accident year 2026, once enough of that year has earned through and early development shows through the casualty triangles. That number will say more about underwriting discipline than the 87% figure calculated on a book still carrying 2023 and 2024 vintages ever could.