Lloyd's reported a 90.8% combined ratio for the first half of 2026, improved from 92.5% a year earlier (Lloyd's, September 2026). The attritional loss ratio implied by its own disclosed components moved the other way, to 51.1% from 48.3%. The entire 1.7 point headline gain came from a major claims ratio that fell to 6.8% from 10.4%.

Lloyd's own underlying combined ratio, which it defines as the combined ratio excluding major claims, also deteriorated, from 82.1% to 84.0%. That measure leaves prior-year reserve releases inside it, and those releases grew to 3.5 percentage points from 2.0. Strip them out as well and the accident-year cost of the book rose 3.4 points, not 1.9. Syndicates absorbed a 6.7% risk-adjusted rate reduction and wrote 15.8% more volume anyway.

Key Takeaways

  • 3.6 points of major claims relief carried the headline improvement, with the major claims ratio falling to 6.8% from 10.4%. Against that, the underlying ratio deteriorated 1.9 points and the expense ratio rose 0.6 points to 36.4%.
  • A 51.1% attritional loss ratio, derived from Lloyd's disclosed expense, underlying and prior-year figures, sits above every half-year and full-year reading the market has published since 2024, the nearest of which was 49.2% in H1 2024.
  • 3.5 points of prior-year release are still inside the 84.0% underlying ratio, because Lloyd's underlying measure excludes major claims only. On an ex-release basis the ratio moved from 84.1% to 87.5%.
  • Risk-adjusted rates fell 6.7% against 3.5% in H1 2025, while volume grew 15.8%. Price per unit of risk fell at nearly twice last year's pace and the market bought substantially more units (Insurance Business, September 2026).
  • Profit before tax fell to £3.5 billion from £4.2 billion despite the better combined ratio, because investment return halved to £1.8 billion (1.6%) from £3.2 billion (3.1%) as yields widened.

Rebuilding the 1.7 Point Move

The arithmetic is Lloyd's own, and it closes exactly. The 2025 half-year release states the definition plainly: "An underlying combined ratio is the combined ratio excluding major claims" (Lloyd's, September 2025). Subtract 10.4 from 92.5 and you get 82.1; subtract 6.8 from 90.8 and you get 84.0. Both hold to the decimal, which confirms that the underlying figure is a catastrophe-adjusted number rather than an accident-year one.

That distinction is what the headline coverage skipped. Laid out as a bridge, the half-year move separates into three components pulling in two directions.

ComponentHY 2025HY 2026Change
Attritional loss ratio48.3%51.1% (derived)+2.8
Expense ratio35.8%36.4%+0.6
Accident-year ratio, ex major claims84.1%87.5%+3.4
Prior-year reserve releases(2.0)(3.5)(1.5)
Underlying combined ratio82.1%84.0%+1.9
Major claims ratio10.4%6.8%(3.6)
Combined ratio92.5%90.8%(1.7)

Read down the change column and the 1.7 point improvement resolves into 3.6 points of catastrophe relief plus 1.5 points of additional reserve release, less 3.4 points of accident-year deterioration. Two of the three drivers are items an actuary would strip out of a rate indication. The one that survives that treatment got worse.

Chief Executive Patrick Tiernan set the tone in the release: "Underwriting discipline and innovation are the keys to maintaining outperformance and quality of earnings" (Reinsurance News, September 2026). Quality of earnings is precisely what the bridge measures.

Deriving the Attritional Loss Ratio

Lloyd's did not print an attritional loss ratio for the 2026 half year, but the identity that recovers it is fixed by the definitions it did publish: attritional equals the underlying combined ratio, less the expense ratio, plus the prior-year benefit. Run it on the 2025 comparatives and the result is 82.1 less 35.8 plus 2.0, or 48.3%, matching the attritional figure Lloyd's disclosed that year exactly. Run it on 2026 and it gives 84.0 less 36.4 plus 3.5, or 51.1%.

That reading sits above the 49.2% recorded at the 2024 half year, the 48.3% at the 2025 half year, and the full-year figures of 47.1% for 2024 and 47.9% for 2025 (Lloyd's, March 2026). The attritional book has spent two and a half years in a 47% to 49% band. It has now left it.

The split inside the 3.4 point accident-year move is where the pricing consequence lives. Loss cost accounts for 2.8 points of it and acquisition and administration expense for 0.6. Distribution cost explains under a fifth of the erosion; the rest is the 6.7% rate reduction feeding through into loss ratio. Lloyd's attributes the expense move to "higher acquisition costs and increased profitability-driven commissions" (Lloyd's, September 2026), which means commission is being paid off a profitability signal the current accident year is no longer producing.

The market context makes the volume decision legible. Global non-life alternative capital rose 9% in the half to a record USD 147 billion and total dedicated reinsurance capital reached USD 688 billion, with reinsurers posting a 19.9% return on equity (Gallagher Re, September 2026). The broker framed the resulting problem as one where "the challenge facing the industry is increasingly becoming one of capital deployment rather than capital generation."

Lloyd's answer to that challenge shows up in its own top line: of the 6.9% GWP growth to £34.7 billion, volume contributed 15.8 points against negative 6.7 from price and negative 2.2 from currency. The same trade shows up differently at syndicate level, where Beazley let cyber premium fall 15.4% rather than chase rate down.

Where the 3.5 Points of Release Came From

The release is a net figure, and Lloyd's says so: the 3.5 points reflect "favourable movement across multiple classes, partly offset by reserve strengthening on the Baltimore Bridge loss and updated Ukraine estimates" (Lloyd's, September 2026). Gross favourable development therefore exceeded 3.5 points, and two named items ran against it.

That construction makes the release harder to extrapolate than a broad one. Marine, aviation, political risk and trade credit carry concentrated exposure to both strengtheners, and the Baltimore Bridge total insured loss has climbed to roughly USD 2.8 billion from the USD 1.5 billion working assumption that shaped January renewal pricing. A net number that already absorbs two adverse movements is a mix outcome, not a statement about reserve margin across the market. The same class-dependence showed up in the Q2 2026 Schedule P data, where short-tail property redundancy masked flat-to-worse casualty development.

The deeper problem is what the release is funding. Risk-adjusted price fell 3.5% in H1 2025 and 6.7% in H1 2026, a compound reduction of roughly 10% in two years. The redundancy being released was earned on business written before those reductions, and it is subsidising a headline ratio for a book priced about a tenth cheaper per unit of risk than the vintages generating the margin.

Nothing else in the result covers that gap. Investment return halved to £1.8 billion, so the earnings cushion that offset underwriting softness a year ago is gone, and total capital, reserves and subordinated loan notes fell to £48.4 billion from £49.8 billion at year end. The central solvency ratio still rose to 503% from 496%, but solvency coverage is measured against a capital requirement, not against the accident-year cost of the business sitting behind it.

Further Reading

Sources