UNION BANKSHARES INC (UNB) stock intrinsic value fair value estimate & owner earnings analysis

Union Bankshares is the holding company for Union Bank, a community bank chartered in Vermont in 1891 and based in Morrisville. It lends and takes deposits in Vermont and northern New Hampshire (a loan office in Littleton became a branch in 2006, and three New Hampshire branches were bought in 2011). At year-end 2025 it had $1.62 billion of assets, $1.18 billion of loans (about 41% commercial and commercial real estate, most of the rest residential real estate loans) and $1.21 billion of deposits, topped up with $286.5 million of Federal Home Loan Bank advances and $16.5 million of subordinated notes due 2031. It also sells many of its home mortgages (about $146 million in 2025), earning a gain on each sale. Because borrowing is the raw material of a bank, the interest it pays is treated as a normal cost of doing business and no separate debt is subtracted from any case below. HOW WE MEASURED EARNINGS. For a bank we start from operating cash flow, then strip out the swings from mortgages waiting to be sold and from accrued interest and other balance-sheet timing, and then deduct stock compensation, ordinary maintenance spending on buildings and equipment, and two bank-specific costs that cash flow adds back: the provision for loan losses (a recurring cost of lending) and the losses on tax-credit housing partnerships (the bank keeps buying new partnership interests, $1.9 million in 2025, to earn the tax credits that lower its tax bill). On that basis owner earnings per share were $2.61 in 2021, $3.01 in 2022, $2.60 in 2023, $2.02 in 2024 and $2.42 in 2025. The 2020-2022 stretch was flattered by cheap deposits and, in 2020-2021, a mortgage-refinancing boom, both of which have reversed, and 2024 was the squeeze year when deposit costs jumped. We use the 2023-2025 median, $2.42 per share, which covers the squeeze and the recovery under today's funding costs. The first half of 2026 supports that level: net income was $5.9 million against $4.9 million a year earlier, and net interest income in the second quarter was $11.6 million against $10.4 million. Per share, though, the gain is smaller than it looks, because the bank also sold about 332,000 new shares in 2026 (shares outstanding rose from 4.61 million in February to 4.95 million in late July), raising $7.4 million. All values below are per share after that sale; we treated the $7.4 million as new capital added to the business. GROWTH WE CREDITED. Owner earnings per share rose 19.9% in 2025, but that was a rebound from the 2024 squeeze (the margin on earning assets improved from 2.77% to 2.93%), not a new trend. The base case therefore starts at the $2.42 level with year-one growth of 2.5%, below the 2025 figure. The 2.5% rests on a mechanism with a record: owner earnings per share grew about 3.2% a year from $1.82 in 2016 to $2.42 in 2025 while the bank paid out about 60% of its earnings ($1.44 a share a year) and kept the rest to build capital. We taper the rate to 2.0% from year six because capital is thin, and we deduct the capital that must be kept to support that growth, so only the remainder counts as distributable. We did not credit loan growth (net loans actually fell from $1.17 billion to $1.11 billion in the first half of 2026), a return of the 2020-21 mortgage boom, acquisitions, or further share sales. BASE CASE: about $38.99 per share. Earnings grow 2.5% a year for five years and 2.0% for the next five, with one mild credit episode in years five and six costing $0.40 a share in each year, similar in size to the $2.2 million loan-loss provision the bank took in 2020. After year ten we assume the earnings level continues at 85% of its year-ten figure rather than growing: a bank in business since 1891 with a deposit base that has produced a profit every year of the last ten is a structural reason for earnings to persist, and the 15% haircut reflects that banks in this region have gone through credit cycles in 1990-92, 2008-11 and 2020. About 40% of the value comes from the first ten years and the rest from the years after. BEAR CASE: about $28.75 per share. This builds in a full credit cycle of the 2008-2011 kind, which is not in today's filings. Commercial and commercial real estate loans (about $480 million) lose roughly 4% over four years (about 1% a year, the order of magnitude of commercial real estate losses at US banks in that cycle), and the other loans (about $700 million, mostly residential) lose about 0.7%, for roughly $24 million of extra loan-loss cost spread $4 million, $8 million, $7 million and $5 million over 2027-2030 (about $20 million after tax at an assumed 18% rate). A $13.6 million commercial real estate relationship that has been on nonaccrual since the first quarter of 2025, and nonaccrual loans that grew to $15.0 million by June 2026, are the visible start of this. In addition, funding costs squeeze margins again in the first two years, costing $0.40 a share in each of those years, about two-thirds of what happened in 2024. Earnings fall to roughly $0.61 a share in year two. Because the bank's equity is only about 5-6% of assets, we assume it raises $15 million of new equity around year two at roughly $13 a share (about three-quarters of tangible book value, a discount to tangible book value that a bank in distress should expect), which adds about 1.15 million shares; the new capital earns 8% after tax. Earnings per share then return to about $2.14 and stay flat, and the same 85% rule applies after year ten. Bear value remains above the equity on the balance sheet (about $18 a share after the 2026 share sale) because the bank's deposit franchise keeps earning after the losses are absorbed; if losses were much larger than we assume, that would not hold (see the margin-of-safety price). BULL CASE: about $43.95 per share. The levers are ones already visible: deposit costs have started to fall (deposit interest expense was $10.4 million in the first half of 2026 against $11.0 million a year earlier) and the margin improved from 2.77% to 2.93% in 2025, so earnings per share rise to $2.58 in year one and $2.75 in year two (the bank earned $3.01 in 2022 on cheaper deposits), then grow 3.2% a year, its demonstrated rate. The same mild credit episode in years five and six and the same 85% rule after year ten apply. WHAT WOULD MAKE US WRONG. Losses much larger than the bear case would hurt a lot given thin capital: losses of about 5% of all loans (roughly $55 million, around $45 million after tax) would consume about half of the $90.4 million of equity, require a large raise at a deep discount and could leave value in the mid-teens per share or below. Deposits fell $121 million in the first half of 2026 (to $1.09 billion), and the quarterly report's narrative was not available to us, so we could not establish why; a continued outflow would force more costly borrowing. We also could not confirm how much of the problem real estate relationship is covered by collateral: at year-end 2025 the loan-loss reserve of $8.4 million covered about 62% of nonaccrual loans. We challenged the growth credit against the fact that loans are shrinking and kept it tied to retained earnings only. We previously valued this company on a different basis, so our earlier figures ($19.50 / $36.50 / $60.00 and a $13.25 margin-of-safety price) are not comparable; what has changed in the business since is that the company added $7.4 million of equity (equity to assets rose from 5.0% to about 5.8%), first-half earnings rose about 21%, and nonaccrual loans grew from $13.6 million to $15.0 million. MARGIN-OF-SAFETY PRICE: $16.50. We are fairly but not fully sure of the earnings level: owner earnings per share ranged from $2.02 to $3.01 over the last four years, and $2.42 is the middle. The worst plausible outcome is a commercial real estate downturn on a balance sheet where equity is about 5-6% of assets; it has no large fee business to cushion it, and it leans on $286.5 million of Federal Home Loan Bank advances and on deposits that have recently been leaving. Its loans are secured by real estate, which gives hard collateral, but that collateral is exactly what falls in a downturn. If we are wrong about the three things most likely to be wrong, the loss rate, funding costs and how long earnings persist after year ten, shares could be worth well under the bear figure, perhaps in the mid-teens. The price assumes a four-to-five-year holding period, during which the business pays $0.36 a quarter ($1.44 a year), which at $16.50 is an 8.7% cash yield, covered about 1.7 times by owner earnings, and the price would be below the roughly $18 of equity per share on the balance sheet, so even a grinding bear outcome leaves capital behind each share. At that price, owner earnings alone are about 15% of the price each year before any growth, so a 20% annual return requires only a modest improvement in how the business is valued over several years. A higher price would leave too little room for the tail outcomes no case models.