ARCH CAPITAL GROUP LTD. (ACGL) stock intrinsic value fair value estimate & owner earnings analysis

Arch Capital Group is a Bermuda-based specialty insurer and reinsurer with three businesses. Insurance (about $7.8 billion of net premiums earned in 2025) writes specialty commercial lines in North America, the United Kingdom, Europe and Australia, including the U.S. middle-market and entertainment book bought from Allianz in August 2024. Reinsurance (about $8.1 billion) sells property-catastrophe, casualty and specialty cover to other insurers worldwide. Mortgage insurance (about $1.2 billion, through Arch MI in the U.S. plus Australian and European operations) is small by premium but very profitable: it earned $1.0-1.1 billion of underwriting profit in each of 2023-2025. How we measured earnings. We start from operating cash flow and take out the cash that comes from insurance float: the growth in loss reserves and unearned premiums, net of receivables and other timing items. That money belongs to policyholders and will be paid out, so we do not count it as profit. We also take out stock-based pay, a small amount of equipment spending and the $40 million a year of preferred dividends. On that basis Arch produced about $1.36 of owner earnings per share in 2016, $3.07 in 2019, $1.03 in 2020 (pandemic claims and mortgage delinquencies), $5.76 in 2022, $8.63 in 2023, $9.86 in 2024 and $10.09 in 2025. Three accounting choices matter. First, investment gains and losses and income from investment funds carried at equity value are left out; they follow markets, not underwriting, and are mostly not received as cash. Second, the 2023 profit included a one-time, non-cash $1.16 billion tax credit that Arch recorded when Bermuda brought in a corporate income tax. We leave it out of 2023. We do count the lower cash taxes it will bring each year until about 2034, which is worth roughly $150 million a year. Third, amortization of acquired intangibles (about $190 million in 2025) is a non-cash charge and is added back. The wearing-out of the acquired business is instead reflected in premiums that stay flat rather than grow. Setting the normal level. Arch's premium base roughly tripled between 2019 and 2025 during one of the strongest pricing markets in decades, so neither the small per-share figures from 2016-2021 nor the peak 2023-2025 figures describe a normal year for today's business. We rebuilt a normal year from each segment's median combined ratio over the nine years 2017-2025. That period covers the soft market, the 2017 hurricanes, 2020, Hurricane Ian in 2022 and the recent hard market. The medians are 96.7% for insurance, 92.2% for reinsurance and about 25% for mortgage. We applied them to a premium base that is already shrinking: first-half 2026 premiums earned fell 6.5% from a year earlier, and second-quarter premiums fell 8%. We held investment income near 2025's $1.6 billion. After interest on its debt, tax at about 16% and preferred dividends, that gives about $2.8 billion a year, or roughly $8.10 a share on the 349 million diluted shares outstanding in mid-2026. It falls to about $7.65 once the Bermuda tax credit is used up. This is about 17-18% of premiums, below the 2016-2025 median of 22%, because the high-margin mortgage business is a smaller part of Arch today than it was in 2018-2019. Near-term path. Owner earnings per share grew 2.4% in 2025. Our base case assumes about -1% per share for the coming year: owner earnings of about $3.5 billion, spread over 7% fewer shares after heavy 2025-2026 buybacks. We then step down to $3.2 billion and $2.95 billion before reaching the normal level in year four, because reinsurance margins (an 80.8% combined ratio in 2025 against a 92% median) and the shrinking premium base both point to a softer market. We credit no premium growth in any case. Arch's past growth came from hard-market expansion and acquisitions, and neither can be counted on today. Base case, about $113 per share. Normal-level earnings of about $2.8 billion a year through 2034 and about $2.67 billion after the tax credit runs out. They are held flat for 25 years, which matches Arch's record of profitable underwriting through several cycles since its 2001 relaunch, and then fade to zero by year 40, since no record we can point to supports going further. The business is worth about $39.4 billion. Bear case, about $71 per share, close to book value. A soft market as deep as 2017-2020: insurance breaks even at a 100% combined ratio, reinsurance runs at 97%, and premiums shrink to about $14 billion. A mortgage credit shock like 2020's takes that segment to a 59% combined ratio and then settles at 30%. Year two combines that shock with a heavy catastrophe year and earns only about $1.0 billion. After that, earnings of about $2.0 billion a year run for 20 years and fade out by year 30. The filings already show adverse reserve development on 2021-2024 casualty and program business, which supports this case. The business is worth about $24.8 billion. Bull case, about $160 per share. Arch holds premiums near 2025 levels and keeps margins close to its recent record: insurance at 95%, reinsurance at 88% and mortgage at 20% (all achieved in 2021-2025). Normal earnings of about $3.3-3.4 billion follow three years that stay near today's $3.5-3.75 billion. Arch's underwriting franchise keeps them going indefinitely at that level, without growth. The business is worth about $55.9 billion. Debt. Arch owes $2.75 billion of senior notes: $500 million at 4.011% due December 15, 2026, $300 million at 7.35% due 2034, $500 million at 5.144% due 2043, $450 million at 5.031% due 2046 and $1.0 billion at 3.635% due 2050. It also has $830 million of preferred shares. For an insurer, borrowed money and policyholder funds are part of how the business operates. So we value the earnings left after the roughly $148 million of annual interest and the $40 million of preferred dividends have been paid, and we do not subtract the debt a second time. In each case, the value of the business to shareholders, after its lenders and preferred holders, is therefore the figure shown: $24.8 billion in the bear case, $39.4 billion in the base case and $55.9 billion in the bull case. Even in the bear case's worst year, earnings before interest cover the interest bill about eight times. The December 2026 maturity is small against roughly $24 billion of shareholders' equity. Much of the debt is fixed at rates below 6% until 2043-2050, which works in shareholders' favour. Our previous analysis also valued Arch after interest but did not set out what it owed. This analysis does. What changed since our last analysis. Our estimates are lower: about $113 base against about $166 before. Three things moved them. We no longer count the one-time 2023 tax credit as earnings. We no longer assume modest growth that lasts forever after year ten. Instead, the base case holds earnings flat for as long as Arch's record supports and then lets them end. And we now set the normal level from segment margins over a full cycle. Not everything moved down. Our normal level (about $7.65-8.10 a share) is well above the roughly $5.50 our last analysis started from. The roughly 7% fall in the share count since 2025 also raises value per share. Margin-of-safety price. Arch is cyclical and exposed to catastrophes, reserve surprises and the housing credit cycle. So the margin of safety is wide, set against the bear case rather than the base. The margin-of-safety price of $74 is about the bear-case value and roughly 1.1 times book value. From that price, a move back toward the base value over two to three years implies about 20% a year. What would make this estimate wrong: margins falling below their long-run medians because of a deeper price war in reinsurance; a U.S. housing downturn that pushes mortgage losses above 2020 levels for several years; or large reserve increases on 2021-2024 casualty business. On the other side, a hard market that lasts longer than history suggests would make even the bull case look cautious. Income from Arch's stakes in investment funds and affiliates (about $500 million a year) is left out of every case. That is a deliberate conservative choice.