Arch Capital, the Bermuda-based specialty property and casualty insurer and reinsurer with a separately-managed mortgage insurance arm, entered the 2026 second quarter mid-cycle: priced near its 52-week high with the insurance and reinsurance markets softening after a long hard-market stretch, and under a management team led by CEO Nicolas Papadopoulo that has made capital return as much a signature as underwriting. There are two ways to read the numbers. The first reads as a company stalling: net premiums earned fell 8.1% year over year, and GAAP net income available to common shareholders slipped to $1.0 billion, or $3.00 a share, from $3.23 a year earlier. The second reads as a company harvesting. Nearly all of the top-line decline was engineered - non-renewed program business, retreats from competitively-priced E&S property, deliberate reinsurance retrocessions, and cedant retentions all shed premium Arch no longer wanted at the margin it demands.
The stronger read holds up. Excluding catastrophe activity and prior-year loss reserve development, the combined ratio was 82.5%, still an excellent underwriting margin, against 80.9% a year ago. Underwriting income came to $657 million. And the centerpiece was the capital decision: Arch repurchased $1.2 billion of its own stock in the quarter and roughly $1.9 billion in the first half - about 94% of first-half earnings - sharply shrinking the diluted share count to 348.8 million from 379.9 million a year earlier. Book value per share still rose 2.8% in the quarter to $68.04, and is up roughly 15% year over year, even though management notes the stock was bought at a price above book value per share.
This matters because the quarter reframes what Arch is doing with a softening market: giving up top-line scale to protect underwriting margin, then returning most of the retained earnings to shareholders. The investing question is whether that trade keeps working - whether margin discipline plus a shrinking share count can compound book value even as earned premium declines, and whether the insurance segment, whose ex-catastrophe margin is eroding, can hold the line. The thesis is the shrink; the test is whether the margin holds while the scale comes down.