ARROW FINANCIAL CORP (AROW) stock intrinsic value fair value estimate & owner earnings analysis

Arrow Financial is the holding company for Arrow Bank, a community bank in upstate New York. Before its July 1, 2026 acquisition of Adirondack Bancorp it had roughly $3.4 billion of loans, about one third of them consumer loans, mostly automobile loans bought from dealers in New York and Vermont, with the rest in residential, commercial and commercial real estate lending. It funds itself mainly with local deposits, which are the raw material of a bank rather than debt to be subtracted, so no debt is deducted below. HOW WE MEASURED EARNINGS. We start from cash earnings, then make the choices a lender needs. Loan-loss provisions are kept as a real recurring cost (the cash-flow statement adds them back). Swings in loans held for sale and in other assets and liabilities are removed. Stock compensation, including shares paid to directors, is deducted. Capital spending is charged in full except the small share that supports growth. In 2025 the bank paid $11.2 million in cash for federal energy tax credits; the credits are worth $15.8 million against a $14.8 million price, so the real gain is about $1.0 million, and we treated the cash payment as timing rather than a cost. Per-share figures are restated for the 3% stock dividends paid through 2023. On that basis owner earnings per share were $1.84 in 2016, $1.85 in 2017, $2.12 in 2018, $2.13 in 2019, $2.41 in 2020, $3.03 in 2021, $2.36 in 2022, $1.69 in 2023, $1.86 in 2024 and $2.70 in 2025. The slide of about 44% from 2021 to 2023 shows what a deposit-funded lender does when rates rise quickly: deposit costs repriced faster than loan and securities yields. We anchor on 2020 through 2025, which covers the zero-rate year, the rate peak, the squeeze and the recovery, and take the median, $2.38 per share. The 2016-2019 years describe a smaller bank and are left out of the level. THE ACQUISITION. Adirondack Bank had about $942 million of assets, $848 million of deposits and $624 million of loans at the end of 2025. Its holders received 1.8610 Arrow shares plus $18.72 in cash per share, an announced value of about $89.1 million; by our arithmetic that is about 2.0 million new Arrow shares (about 12% more) and about $20 million of cash. We treat the deal as worth what was paid in the base case. Adirondack's own earnings were not disclosed, and management's stated target of roughly 18% earnings accretion in 2027 is a target we cannot verify, so it is not credited. Integration of the banking systems is scheduled for later in 2026. BASE CASE, $56.43 per share. Year one is $2.57, below the $2.70 earned in 2025. For reconciliation, owner earnings grew 45% in 2025, and the first year of this case is 8% above our $2.38 anchor, so it assumes less than last year's pace. The step up from the anchor is supported by recent quarters: net income was about $13 million a quarter in the second half of 2025 and $13.5 million in the first quarter of 2026, when the margin was 3.47% against 3.17% for all of 2025 and 2.72% in 2024. Second-quarter 2026 net income was $11.0 million after a $2.8 million provision, including a $1.6 million reserve for one commercial real estate borrower that entered bankruptcy, and $1.0 million of merger costs; the margin eased to 3.42%. The first half of 2026 earned $24.4 million, or $1.48 per share, against $17.1 million a year earlier. From year two we credit 3% annual growth through year 10, then growth fading to zero by year 20, after which earnings are held flat. The growth is in the bag because loans have grown every year (about $320 million in 2022, $229 million in 2023, $188 million in 2024 and $75 million in 2025), funded from deposits and retained profit through the 2023-24 squeeze, and owner earnings per share compounded about 4.3% a year from 2016 to 2025; we credit less. Flat earnings after year 20 rest on a franchise that earned a profit in every year we can see (2014 to 2025). We did not credit margin expansion beyond what has been reported, acquisition savings, or buybacks. With deposits as the funding, there is no debt to subtract from this value. BEAR CASE, $27.81. Three things go wrong. First, the margin gives back its gain: earnings run $2.40 in year one, $1.95 in year two and settle at the $1.77 average of 2023 and 2024, growing 1% a year to year 10 and flat after. Second, a full credit cycle arrives in years two to five, modeled on 2008 to 2011, when the Federal Reserve's charge-off rate on all bank loans rose from 0.79% in 2007 to 2.91% in 2009 (real estate loans 0.37% to 2.82%, consumer loans up to 5.75%) and was still 1.34% in 2011. Arrow's own losses run far lower (0.19% of loans in 2025, with reserves of $34.3 million, 0.99% of loans), so we take about 40% of the industry's increase: losses above the normal level of 0.4%, 0.8%, 0.6% and 0.3% of roughly $3.5 billion of loans, about $73 million before tax and $58 million after tax, or $3.52 per share spread over four years. Earnings stay positive in every year of the case. The common equity ratio was 13.2% at June 30, 2026, before the acquisition closed, so we assume no dilution. Third, the acquisition is worth only 75% of what was paid, a cost of about $1.06 per share. Funding is not a separate problem here: deposits fund the bank, and the roughly $300 million of brokered certificates replaced by Federal Home Loan Bank borrowings in the second quarter of 2026 is a funding change rather than a maturity wall. This value is above the $25.44 tangible book value at June 30, 2026 because the bank keeps earning through the downturn and its deposit franchise survives it. If credit losses were twice as large, the case would be about $24.92. BULL CASE, $73.68. The bank earns what the first half of 2026 implies, $2.96 per share a year, grows 4.2% a year to year 10 (just below the 4.3% it has delivered) and then fades to zero by year 20. The acquisition is worth 10% more than was paid, from cost savings on the combined branch network; this is the only credit given to acquisition benefits. WHAT COULD MAKE THIS WRONG. A large part of the value rests on the franchise still earning in later years: in the base case about $21 per share comes from the first 10 years and about $36 from the first 20. A rapid change in rates can compress the margin again, as in 2022-24. Indirect automobile loans are the most loss-prone part of the book; loans 30-89 days past due were $28.9 million, 0.84% of loans, at December 31, 2025, up from 0.60%, and $21.0 million of that was consumer loans. The 2022 annual report was filed in July 2023, months later than usual. The previous analysis (base about $30.55) was built on a different basis and its figures are not comparable; what has changed in the business since then is that the Adirondack deal closed, the margin rose to about 3.4%, and first-half 2026 earnings rose 43% from a year earlier. MARGIN-OF-SAFETY PRICE, $16.00. We are fairly sure of the earnings level: it has ranged from $1.69 to $3.03 per share over ten years, and the median of the last six is $2.38. The worst plausible outcome is a recession that brings heavy losses on the auto and commercial real estate books while the margin falls and the new acquisition disappoints. The bank is deposit-funded and had a 13.2% common equity ratio before the deal, so it should come through that with lower earnings rather than needing rescue. If we are wrong about the three things most likely to be wrong, which are the margin level, the size of a credit cycle and the franchise lasting past year 20, the shares would still be worth about $24.92 after twice the credit losses, and about $17.95 if the bear case earned nothing after year 20. The price assumes about a five-year holding period, during which the dividend is $0.30 a quarter ($1.20 a year). At $16.00 the shares would be about 6.7 times the $2.38 anchor earnings, the dividend would equal about 7.5% of the price, and the price would sit at about 63% of tangible book value. We estimate that in five years the shares could be worth around $31, roughly today's tangible book plus retained profit, which with dividends implies about 20% a year from that price; a stressed value of $28 implies about 18%. The margin-of-safety price is not higher because a bad credit cycle combined with a troubled integration could leave the shares in the low $20s for years, and the implied return would then fall short.