AMERISAFE INC (AMSF) stock intrinsic value fair value estimate & owner earnings analysis

AMERISAFE sells workers' compensation insurance to small and mid-sized employers in hazardous industries — construction, trucking, logging, agriculture, manufacturing, oil and gas, and maritime. It has about 10,200 policyholders, almost all reached through independent agents. No single agency accounts for more than 2% of premium, and the ten largest customers together are only about 2%. The company has no borrowings. It earns money two ways: underwriting (premiums less claims and expenses) and investing roughly $800 million of premiums and capital, mostly in high-grade bonds, until claims are paid. The central fact about this business today is where its underwriting profit comes from. Each year AMERISAFE books new claims at a conservative first estimate — about 72 cents per premium dollar in 2025 and again in the first half of 2026. Add about 30 cents of expenses and a cent of policyholder dividends, and the current year's business shows a loss on that first estimate. Profits have come from older years' claims settling for less than reserved. That was worth about 12 cents of every premium dollar in 2025, and roughly $7 million a quarter so far in 2026. The record is real and long: every accident year from 2016 through 2023 has developed favorably. But the company's own claim tables show the cushion is thinner on recent years. Claims from 2016 ultimately cost about 22% less than first booked, and 2019 about 16% less. By contrast, 2021 is only about 7% below its first estimate after four years, and 2022 about 4% below after three — both well behind where older years stood at the same age. Meanwhile, workers' compensation prices fell for a decade. Premiums earned dropped from about $376 million in 2014 to $267 million in 2023, while the yearly cost of new claims stayed near $190–205 million. Owner earnings per share fell from about $4.79 in 2016 to $2.10 in 2025. How we measured earnings: owner earnings start from net income. We add back non-cash items (depreciation, bond amortization, deferred taxes), remove investment gains and losses, and subtract ongoing capital spending. Swings in claim reserves, unearned premium and reinsurance balances are treated as timing, the way working capital is for an industrial company. Because reserve releases are running down, we start from the three years at today's business size (2023–2025, middle value about $2.59 a share). We then carry the run-off explicitly in each case rather than assuming recent releases repeat. The first half of 2026 sets the starting point. Premiums earned rose about 10%, on more policies, 93% renewal retention and higher payroll-audit premium. Investment income held near $13 million. But the combined ratio rose to 94.3% from 90.5%. That pace works out to about $1.90 of owner earnings per share for the year. Base case, about $25.70 per share. No premium growth is credited beyond today's level, since neither better pricing nor further share gains is secured. Reserve releases fade over three years, from about 10 points of premium to about 6½ points — what recent accident years appear to carry on the company's own development pattern. That leaves an underwriting margin of about 3%, plus about $26 million of investment income, or roughly $1.50 a share after tax from 2030 onward. We hold that level flat rather than growing it. The business renews annually in a line employers are legally required to buy, and it has kept this niche for four decades — evidence the earnings persist, but not evidence they grow. Bear case, about $15.90 per share. Two things go wrong together. First, recent accident years carry only about 3 points of cushion. That leaves underwriting at roughly break-even and earnings of about $1.00 a share from investment income. Second, a payroll recession repeats 2009–2011. Then, AMERISAFE's premiums earned fell from about $290 million in 2008 to $219 million in 2010, its first-estimate claim ratio rose to 81.8%, and its combined ratio reached 100.5% in 2011. Expenses now run about 30 points of premium versus 20–24 then, so a repeat would hit harder. We model three years of near-zero earnings before recovery. The company holds about $250 million of equity against roughly $300 million of annual premium, which is conservative. Even this case does not require raising capital, and with no debt there is nothing for lenders to call. The bear value sits above book value of about $13.50 a share for two reasons: the company does not consume its capital in the downturn, and afterward it keeps earning a return on the float and capital it holds. Bull case, about $32.60 per share. Recent accident years prove nearly as over-reserved as older ones (about 9 points). Expenses settle near 30% of premium, and today's premium base holds. That produces about $1.95–2.00 a share indefinitely. What would make this wrong: claim reserves are about $594 million against $250 million of equity. A 10% reserve shortfall would cost about $3 a share of capital, and the claims tables, not management commentary, are the thing to watch. Lower interest rates would also reduce investment income; the portfolio averaged 3.3% in 2025 and ended the year at 3.8%. Dividends from the insurance subsidiaries to the parent are limited by state regulators. Nothing material in the business has changed since our previous analysis; the earlier figures were prepared on a different basis and are not comparable. Margin of safety price: about $17. We are fairly confident in the earnings level, because this kind of insurance profit is visible quarter by quarter and the company is debt-free. We are less sure how much reserve cushion remains, and that is the biggest swing between our cases. The worst plausible event is a payroll recession arriving while the cushion is thin, perhaps with some strengthening of older reserves. The business would survive it: no debt, thousands of small policyholders, a low ratio of premium to capital, and regulated subsidiaries. Special dividends would stop. If we are wrong about the cushion alone, the shares are worth about $18. If we are wrong about the cushion and the cycle together, about $16. If bond yields fell a full point, the base would be nearer $20. The price assumes two to three years for value to show. Over that period the regular dividend of $1.64 a year makes up a large part of the implied return. From about $17, the base case plus dividends implies near 20% a year over three years, and the bear case implies a value slightly below that price before dividends. Above that price, the bear case starts to dominate the outcome.