Arch's Buyback Machine Meets a Thinning Casualty Cushion

A 16% ROE quarter, a $1.4 billion repurchase, and a marine-war premium tailwind mask a softer underwriting market.

ACGL · Earnings Call · 2026-10-28

The quarter that capital return built

Arch Capital's third quarter looked clean on the surface: $910 million of after-tax operating income, $2.67 per share, a 16% annualized operating ROE, and an 18.1% net income ROE. The combined ratio was 94.4 in insurance and 78.3 in reinsurance, helped by light catastrophe activity and $179 million of favorable prior-year development. Premium growth was positive in insurance but negative in reinsurance, a split that says as much about market capacity as company execution. The number that really matters is the buyback. Arch repurchased 14.8 million shares for $1.4 billion in the quarter, and through nine months had returned roughly 109% of net income through repurchases. “Through nine months of 2026, we have repurchased approximately 109% of our net income in our own shares.” — François Morin, Chief Financial Officer · 2026-10-28 That is not a side dish. With growth harder to find, cycle management has become capital management. Repurchases of common stock were $1.2 billion in the second quarter and then $1.4 billion in the third. The stock is not statistically cheap: 7.2x trailing net income. But management has been explicit that price is part of the calculus. On the prior call, François Morin said, “And the current prices, we like the stock still. We think it is very attractive.” — François Morin, Chief Financial Officer · 2026-07-29 The fourth-quarter buyback slowed to about $100 million through late October, but that was partly a blackout and 10b5-1 plan artifact, not necessarily a change in philosophy. The June debt raise added roughly $1.1 billion of incremental capital, and debt to capital remains low enough that buybacks are still the preferred return tool. This is capital management as strategy, not cleanup.

Casualty: the cushion is gone

The underwriting story is less comfortable. For several quarters casualty was the growth engine. In April, Nicolas Papadopoulo was still arguing that casualty pricing would stay above trend: “I think we still optimistic on the casualty... continue to sustain price increases above trend.” — Nicolas Alain Papadopoulo, President and Chief Underwriting Officer · 2026-04-29 By October, the language had changed.

First correction is, I think I said the pricing in casualty has deteriorated, and now it is about loss trend to be loss. I think that margin now is zero or turning negative.
Nicolas Papadopoulo, President and Chief Underwriting Officer · 2026-10-28

That is the central change in this report. Arch still likes the absolute return in E&S casualty, construction, and national account casualty, but the margin of safety is thinner. Management also repeated that it will not chase growth for its own sake. That matters because insurance net premiums written grew only 2.7%, and reinsurance net premiums written fell 9.3% as abundant capacity pushed property and short-tail pricing lower. If casualty margin is at zero or turning negative, 2027 growth has to come from mix, middle-market expansion, and selective E&S opportunities rather than from rate.

Marine war and Bermuda: the earnings levers

A fresh company-specific theme this quarter is marine war risk, tied to the broader Iran conflict. Insurance premium growth benefited from increased writings of marine war risk coverage, and the ex-cat accident-year loss ratio improved 170 basis points to 55.8%, with most of the improvement coming from incremental marine war premium. On the fourth quarter, Morin was direct: “No, you should absolutely expect similar levels in the fourth quarter.” — François Morin, Chief Financial Officer · 2026-10-28 That is a real tailwind, but it is event-driven and priced for risk. It supports Q4; it does not create a durable growth engine. The quieter margin lever is Bermuda tax. The QRTCs lowered the insurance expense ratio and helped corporate expense fall to $22 million. Morin called them “sustainable” — François Morin, Chief Financial Officer · 2026-10-28 and said the credits should rise in 2027 on a percentage basis. That is helpful, but it is also policy-dependent; the deferred-tax-asset question remains unresolved and management wisely runs capital models with and without it.

What they stopped talking about — and what they started

The call added admitted markets to the risk list: some business is moving from non-admitted/E&S back to admitted carriers, which hurts the E&S franchise, and Arch says it will not chase that business into retail because of channel conflict. That is a subtle negative for a company whose recent growth came partly from E&S dislocation. Data centers got a question, and the answer was refreshingly unsensational. Management estimates roughly $10 billion of new industry premium, but Papadopoulo said, “For us, it is an opportunity, but it is not a big opportunity.” — Nicolas Papadopoulo, President and Chief Underwriting Officer · 2026-10-28 Arch writes some of it out of London, but it is not a game changer. In a market where every insurer wants an AI data-center narrative, that is a useful anti-hype signal. Mortgage remains a strong but constrained business. The 2.2% delinquency rate and roughly 80% persistency are excellent, and home affordability policy efforts are not a near-term threat. But mortgage rates above 7% limit new originations. Reinsurance, meanwhile, is squeezed by reinsurance capacity; Arch is leaning harder on third-party capital with a lower cost of capital, which protects returns but also caps net growth.

Bottom line

Arch is still a high-quality compounder, but the story has rotated. The 16% ROE is real, yet it is flattered by low cats, favorable development, marine-war premium, and tax credits. The durable engine is now buybacks plus selective underwriting, not broad market growth. The stock's recent 9.4% ninety-day move says investors like that trade, even with shares still below the 2024 peak. The risk is that the casualty cushion keeps thinning while the buyback shrinks the share count and flatters per-share optics. If casualty margins are truly at zero or turning negative, 2027 will test how much of Arch's return is underwriting skill and how much is capital return.