$63.00
$156.00
bear $63.00
base $123.00
bull $156.00
margin of safety $52.00
market price* $89.16
Selective Insurance Group is a New Jersey-based property and casualty insurer, founded in 1926, that sells almost entirely through independent insurance agents. About 79% of its premiums come from Standard Commercial Lines (general liability, commercial auto, workers compensation and commercial property for small and mid-sized businesses), with the rest from Excess and Surplus lines and Standard Personal Lines (home and auto). In 2025 it wrote $4.87 billion of net premiums, up from $2.24 billion in 2016, while the diluted share count barely moved (58.7 million to 61.1 million). Its earnings come from two places: underwriting profit, when claims and expenses come in below premiums, and investment income on roughly $11-12 billion of mostly investment-grade bonds funded by shareholders' capital and by premiums held until claims are paid. How we measured earnings. We start from operating cash flow and then remove the growth in loss reserves, unearned premiums, receivables and other balance-sheet items. For a growing insurer those items are large: in 2025 operating cash flow was $1.23 billion but $715 million of it was simply more premium money held for future claims, not profit. Removing it leaves essentially net income plus depreciation and other non-cash items, less stock compensation and ordinary capital spending, less the $9.2 million a year owed to the preferred shareholders. Investment gains and losses are excluded; in 2022 this added back $115 million of mostly mark-to-market investment losses. Depreciation and amortization includes a small amount of bond premium amortization (about $6 million in 2025) that we did not remove. On this basis owner earnings per share were $3.50 (2016), $3.34, $4.32, $4.56, $4.64, $5.71 (2021), $6.22, $6.08, $3.40 (2024) and $7.37 (2025). 2024 was hit by $311 million of reserve increases for older years, mostly general liability claims from accident years 2020-2023, driven by larger jury awards and legal costs ('social inflation'). 2025 recovered with a 97.2% combined ratio, and its loss estimates for the current year were set 3.5 points higher than 2024's to reflect that trend. Earnings level. The median of 2021-2025, a window that covers the business at roughly its current size and includes a heavy catastrophe year and a major reserve charge, is $6.08 per share. We lift the starting level to $6.60 because two facts have now held for six straight quarters: investment income is structurally higher as the bond portfolio rolls into higher-yielding securities (after-tax investment income was $232 million in the first half of 2026, up 18%), and premiums are about 18% larger than in 2023. We do not start at the 2025 figure of $7.37, because a business this exposed to casualty reserve surprises should expect a year like 2024 roughly once every five years. Owner earnings rose 117% in 2025 as it recovered from 2024; our base case for 2026 assumes $6.60, a decline of about 10%, even though first-half 2026 results ran ahead of that (combined ratio 98.1%, no reserve development for older years). Growth we credited, and the cost of it. Premiums written fell 3% in the first half of 2026 because management cut back on new business and on accounts it judged underpriced (renewal price increases slowed to 6.5% from 8.9%; personal lines new business fell 36%). We therefore credit no growth in 2026 or 2027. From 2028 the base case credits 3% a year, well below the roughly 9% a year premium growth of 2016-2025, reflecting the company's long record of adding independent agents and following insured values higher. Growth is not free for an insurer: each extra dollar of premium needs capital behind it, so in growth years we deduct about 60 cents of retained capital per added dollar of premium from what owners receive. Expansion into new states, Excess and Surplus growth beyond its current size and margin gains were not credited. After 2035 earnings are held flat with no growth: what keeps them going is a 100-year-old independent agency franchise and an investment portfolio that keeps earning as long as the company keeps writing policies. Bear case, about $63 per share. Insurers have been through long underwriting downturns: from 1999 to 2002 the U.S. property and casualty industry ran combined ratios of roughly 107% to 116% as reserves set in a soft market proved too low. The bear case puts Selective through a similar five-year episode starting now: a 2024-sized reserve charge in the second half of 2026 (owner earnings about $3.30), combined ratios of about 108% and 106% in 2027-2028 with credit losses of about 1% of the bond portfolio, as investment-grade portfolios suffered in 2008-2009 (owner earnings about $0.60 and $1.20), then 103% and 101% as prices catch up ($3.40 and $4.00). Afterwards it settles at a 100% combined ratio, with bond yields falling back toward the 2010s level so investment income drops about 20%, leaving $4.25 a year with no growth. The company does not need new capital in this case: earnings stay positive in every year and shareholders' equity is about $3.7 billion. The bear value is above tangible book value (about $58 per share at June 2026) because even break-even underwriting leaves a large, steady stream of investment income on premiums held, and Selective's worst recent year, 2024, still produced $3.40 per share. Base case, about $123 per share. Earnings start at $6.60, which assumes a combined ratio near 98% on average over time, including occasional bad years, and investment income at today's level. Flat for two years, then 3% growth to 2035 less the capital needed to support it, then flat. Bull case, about $156 per share. Starts at the $7.40 level of 2025 and the first half of 2026 (combined ratio about 97%, no reserve surprises, investment income holding), with premium growth of 5% a year from 2027 to 2035, still below its demonstrated pace, less the capital it needs, then flat. Debt. Selective owes $901 million of senior notes (5.90% due 2035, 6.70% due 2035, 7.25% due 2034, 5.375% due 2049) and small home-loan-bank borrowings, plus $200 million of 4.60% preferred stock. For an insurer, policy reserves and premium funds are the raw material of the business, so we value earnings after interest and preferred dividends rather than subtracting the debt again. In each case the value shown ($63, $123, $156) is what remains for common shareholders after lenders and preferred holders; interest of roughly $50 million a year is already charged against earnings. No debt matures before 2034, and in every case, even the bear case's worst year, earnings cover interest. Changes since our previous analysis. The earlier figures were built on a different basis and are not comparable. What changed in the business: the first half of 2026 showed no reserve increases for older years, investment income rose sharply, and the company deliberately shrank premiums to protect margins. Margin-of-safety price. We are moderately sure of the earnings level. The investment income half is solid, earned on high-quality bonds whose yields are locked in for several years. The underwriting half can swing by $4 per share in a single year, as 2024 showed. The worst realistic event is a repeat of the industry's 1999-2002 experience, with general liability and commercial auto reserves proving too low for several years running. The business would survive it: debt is only about 20% of capital, nothing matures before 2034, the assets are mostly investment-grade bonds, no customer or agent dominates, and the main constraint would be New Jersey and other state regulators limiting dividends from the insurance subsidiaries to the parent. The two things we are most likely to be wrong about are the normal combined ratio and the level of bond yields. If underwriting settles near 101% instead of 98%, the shares are worth roughly $85. If yields fall back toward 2010s levels, roughly $93. If both happen, about $55-60. The price assumes a four-to-five-year holding period, during which the dividend of $1.72 a year is paid. At $52, slightly below tangible book value, the base case implies more than 20% a year if it is recognized within about five years. If both of the main risks go wrong, that outcome implies roughly the original price back plus dividends, and the bear case still implies a modest positive return. The margin-of-safety price is $52.