United Bankshares

IThe business

A bank takes deposits, lends them out, and keeps the difference. United Bankshares does that through one subsidiary, United Bank, which has more than 240 offices in West Virginia, Virginia, Maryland, Washington, D.C., North Carolina, South Carolina, Ohio, Pennsylvania and Georgia.² It has no trading desk and no large card book. Fees from trust, brokerage, deposit and card services made 11% of its revenue in the first nine months of 2026, and the rest is net interest income.² At 30 September 2026 it had $34.2B of assets, $25.5B of loans and $27.7B of deposits, and shareholders owned $5.41B of equity, of which $3.37B is tangible, meaning what is left after $2.02B of goodwill and $0.03B of other intangibles.² The goodwill is 37% of equity, which is why a return of 9.2% on equity is a return of 14.7% on tangible equity.²

It had 2,812 full-time-equivalent employees at the end of September, and it calls itself the 35th largest US banking company by market value.² One 2026 ranking by market value lists it as the largest company in West Virginia; it is a secondary source and I use it for nothing else.⁹ The Piedmont Bancorp deal, which added Atlanta, closed on 10 January 2025; it was the company’s 34th acquisition.³⁴ It brought $2.4B of assets and $2.1B of loans, and United paid about $281M in stock for a company with $202M of equity.³

Net interest income is the difference between what the bank earns on loans and securities and what it pays for deposits and debt. It was a record $294.3M in the third quarter, up 5% in a year, on a net interest margin of 3.86%, up from 3.80%.² The rate paid on interest-bearing deposits fell 33 basis points in a year to 2.51%, while the yield on loans fell 18 basis points, so the bank gained more from cheaper funding than it lost on repricing.² Cheaper funding explains more of the rise than growth does: loans grew 3.8% over the year and deposits 3.1%.²

What it sells

The loan book is the product, and it is real estate. Here is how the $25.5B splits.⁶

Three things stand out. First, the commercial real estate book that the market worries about is 35% of loans, and with construction, 49%.⁶ Second, the bank is no longer mainly a West Virginia bank: 44% of loans are in the Southeast, 34% in the Washington and Baltimore area and 18% in the home markets.⁶ Third, the mix inside the commercial real estate book is not what the headline suggests; I come back to it below.

Revenue grew 10.7% a year from 2016 to 2025, from $0.50B to $1.24B, and net income 13.6% a year, from $147M to $465M C (calculated).⁵³ Almost all of that is acquisitions. The company bought Cardinal in 2017, Carolina Financial in 2020, Community Bankers Trust in 2021 and Piedmont in 2025, and the share count went from 81M at the end of 2016 to 140M at the end of 2025, up 73% C (calculated).⁵³ Per share, earnings grew 5.7% a year, from $1.99 to $3.27.⁵ That is the number to keep in mind. Net income more than tripled and earnings per share rose 64%.

Where the deposits are

Deposits are the other half. The company says it has the largest deposit share in West Virginia, more than $800M ahead of the number two bank in the 2025 summary of deposits.⁴ Its biggest single market is the Washington, D.C. area, with $11.0B of deposits in 57 branches, where it ranks sixth; it is first in Morgantown and Parkersburg, second in Charleston, and 21st in Atlanta with $1.2B.⁶ At 30 September 24.6% of deposits paid no interest, 24.5% were interest-bearing transaction accounts, 28.7% money market, 4.5% regular savings and 17.6% time deposits.⁶ Time deposits have an average maturity of about five months, and about 14% of all deposits are tied to a floating index.⁶ An estimated 32% of deposits were uninsured and uncollateralized, and the bank counts $22.4B of available liquidity.⁶

This bank paid 2.17% on average in 2025, up from 0.19% in 2021, and a lower 1.88% in the third quarter of 2026 on my arithmetic C (calculated).⁵² The share of deposits that cost nothing has slipped from 36% in 2020 to 25%, and that is the part of the margin I would watch if rates fall slowly.⁵⁶ The company’s own sensitivity says a 100 basis point rise in rates adds about 1.9% to net interest income in the first year, and a 100 basis point fall takes 0.2% away in the first year and 2.2% in the second.⁶ About 48% of loans carry fixed rates, and $1.8B of them mature within a year at 5.1%.⁶

Credit costs

Over ten years the bank charged off an average of 0.13% of its loans a year, and provisioned 0.20% C (calculated).⁵ That is a clean record. In 2025 the charge-off rate was 0.20%, against 0.00% to 0.06% in 2021 to 2024, and $21.8M of it was two loans to the same sponsor, secured by office buildings in Northern Virginia, with a full guarantee that the sponsor could no longer support.³ In the third quarter charge-offs were $11.6M, 0.18% of loans, and $6.3M of that came from one commercial relationship that was fully reserved the quarter before.² Non-performing loans were $109.9M, 0.43% of loans, and non-performing assets 0.35% of assets; the allowance was $294.1M, or 1.16% of loans and 268% of non-performing loans C (calculated).² Loans 30 to 89 days past due rose to 0.38% of loans from 0.28% in June, and the allowance, counting the reserve for unfunded commitments, fell to 1.31% of loans from 1.36% a year earlier.⁶² They are the two numbers I would watch.

The balance sheet

This is where a regional bank differs from the big ones on this site. Securities are only $3.84B, 11% of assets, and almost all of them, $3.50B, are available for sale, which means they are carried at market value and the loss is already in book value.² Gross unrealized losses on those bonds were $218.5M at 30 June 2026, and the average duration of the portfolio is 4.0 years.¹⁰⁶ The company opted out of counting that unrealized loss in regulatory capital, so its capital ratios do not move with bond prices; its tangible book does.³ Bank of America carries $109B of unrealized bond losses, 54% of its tangible equity, mostly on bonds held to maturity and not in book value.⁷ United’s mark, after tax, was $143.4M at 30 June, about 4% of tangible equity.¹⁰² I have to estimate the 30 September figure, because the 10-Q was not yet filed when I wrote this. Equity fell $96.6M in the quarter although the bank earned $128.7M, paid $52.0M of dividends and spent $98.7M on buybacks; the gap is about $75M, and I assume it is mostly the bond mark, which would take it to about $220M, or 6.6% of tangible equity E (estimate).¹¹²

The AOCI chart is the securities-loss story in one picture. It was a $333M loss in 2022, when rates rose, and had shrunk at every year end since as the bonds pulled toward par, though it ticked back up to $143M at June 2026.⁵¹⁰ If rates do not rise further, the loss turns back into book value over several years with no help from earnings. I put $25M a year of that into my base case for six years, nothing in the bear case and $37M a year in the bull.

A bank’s operating cash flow moves with the size of its balance sheet, so I do not use a free-cash-flow model here. Section VI explains the method I use.

IIThe moat

A durable advantage is a cost or a habit that a rival with money and talent cannot copy in a reasonable time. A bank’s edge is often narrower than it looks, because loans are commodities. Where United is hard to copy is in deposit share in small markets, a low cost base, and a record of buying banks and keeping their deposits. I can test each against numbers.

The test of an advantage is whether it has widened. Here is the evidence both ways.

Evidence it is widening

• The efficiency ratio fell from 53% on my arithmetic in 2024 to 48.5% in 2025 and was 48.2% in the third quarter of 2026 C (calculated).⁵²
• Net interest margin rose from 3.77% in the first nine months of 2025 to 3.82%, and to 3.86% in the third quarter.²
• Return on tangible equity rose from 13.63% in the first nine months of 2025 to 14.74%.²

Evidence it is narrowing

• The cost of money. Non-interest-bearing deposits fell from 36% of deposits in 2020 to 24% in 2025 C (calculated), and the total cost of deposits was 2.17% in 2025 against 0.97% in 2019 C (calculated).⁵
• Expenses grew faster than revenue in the third quarter: up 7% from a year earlier while revenue rose less than 1% C (calculated), with a $2.8M reserve build for unfunded commitments against a $3.2M release a year earlier.²

On balance the advantage is real in deposits and cost, absent in lending, and not widening fast.

IIIThe competitive advantage

The last section was about why a rival cannot copy the bank. The advantage is what the bank earns because of it. For a bank the measure that matters most is how much profit it makes on the tangible equity it holds. The cleanest test is to put it beside the two banks on this site that are the largest and the best known, Bank of America and JPMorgan Chase.

Here is what that table says. On return, United sits level with Bank of America at 14.7% against 14.9% before its settlement charge, and well below JPMorgan at 20%.²⁷⁸ On price it is the most expensive of the first two: it trades at 1.97 times tangible book where Bank of America trades at 1.70, for about the same return and a smaller, simpler business.¹⁷ JPMorgan earns about five points more and trades at 3.0 times, which is what the gap in return predicts C (calculated). I cannot call United cheap because its book multiple of 1.23 looks low. The multiple that matters is the one on tangible book, and on that measure the market pays more for United than for the better-known bank.²

Why it earns what it earns

Return on tangible equity is the product of return on assets and leverage. United’s return on assets was 1.50% in the third quarter, and its average assets were 9.8 times its average tangible equity, which gives 14.7% C (calculated).² The 1.50% is high for a bank. The company’s proxy says its 2025 return on assets of 1.41% ranked in the 94th percentile of its own peer group, whose average was 0.99% and median 1.15%, and its net interest margin of 3.78% compared with 3.50% for the group.⁴ The group is the company’s choice and I have not checked who is in it.

First, cost. Noninterest expense was $156.7M in the third quarter against revenue of $325.1M, the 48% efficiency ratio.² The ratio is low partly because there is no trading floor and no card processing, and partly because the bank pays its staff and rents its branches in small towns. Second, the deposit mix, a quarter of it free, in markets where nobody is advertising for money.⁶ Third, the capital: the bank holds a CET1 ratio of 12.9% against a 7.0% minimum with the buffer, and it earns 14.7% anyway. That is the unusual part: a high return on a thick cushion.²³

The ROTCE chart is the picture of the advantage, and of its limits. The bank earned more than my cost of equity in every one of the eight years I can compute, by 3.3 to 5.6 points, which is rare for a bank this size.⁵ The range is narrow: 12.5% to 14.8%, and the average is 13.8%. The nine months of 2026 at 14.7% are in the top of that range, not above it.² So this is a steady bank that earns about four and a half points more than it costs. The price asks for 14.3%, which is where the bank has been for two years. It does not need the bank to get better; it needs it to stay at the top of its range.

The advantage is therefore real, modest, and in the price. The valuation asks which return to expect: the 14.5% of the last two years, the 16% to 17% that the best case needs, or the 12.5% of 2024 and less in a hard cycle.

IVManagement

Management matters more at a bank than at most companies, because the people who make the loans decide how many to make. I read management for three things: how they are paid, how much of their own money is in, and what they did with capital over a full cycle.

Who runs it

The chief executive, Richard M. Adams, Jr., took the job in 2022. The executive chairman, Richard M. Adams, who is 79, had been chairman and chief executive before that and still takes part in the pay process: the proxy says that before the February 2025 committee meeting he and the committee chairman reviewed the company’s results, the chief executive’s performance and the chief executive’s recommendations for the other officers.⁴ In 2022 the title of chief executive passed to the then president, Mr. Adams, Jr., and Mr. Adams stayed on as executive chairman. That is continuity, which suits a bank that sticks to what it knows. It is a risk if the next acquisition is bigger than the last.

Pay, and what it pays them to do

The chief executive’s 2025 pay was $6.14M: a salary of $1.13M, stock awards of $2.64M, an annual cash incentive of $2.20M and $0.17M of pension change and other pay.⁴ The median employee earned $72,502, a ratio of 85 to 1.⁴ Stock was 43% of the chief executive’s pay C (calculated), a good deal less than the 96% at Bank of America, though what matters is what it is paid for.⁴⁷

The annual incentive has three parts. Forty percent is earnings per share, with a target of $2.93 and a maximum of $3.22; the bank earned $3.27, so that part paid at the maximum. Another 40% is return on average assets against a peer group, ranked at the 94th percentile. The last 20% is the ratio of non-performing assets to total assets against the same peers, ranked at the 59th. The payout was 152% of target.⁴ What does that pay them to do? Earn more per share, and keep return on assets high, and keep bad loans low at a point in time. It does not pay for growth in tangible book value, and it does not pay for return on tangible equity. Return on assets does not charge them for the capital they hold, so holding more equity than needed costs them nothing. Earnings per share does charge them for shares issued in a deal, which is some check on paying with stock.

Stock pay is better designed. Sixty percent of the long-term award is performance stock units, half measured on three-year total return to shareholders against peers and half on three-year average return on adjusted tangible equity against peers, and 40% is time-based stock that vests in thirds over three years.⁴ The 2023 awards, measured over 2023 to 2025, paid 70% of target, because the stock ranked 33rd among its peers and the return on tangible equity ranked 37th.⁴ I count that as the habit working: pay fell short when the bank trailed.

Their own money

The five officers hold 1.86M shares between them, worth about $92M at the close, and the 27 directors and officers together hold 4.43M shares, 3.16% of the company, about $218M C (calculated).⁴¹ Mr. Adams, the chairman, holds shares worth about $59M, eleven times his pay; the chief executive’s $11.0M is about 1.8 times his. Executives other than the chief executive must hold three times salary in stock and options, and hedging is banned; I did not find the chief executive’s own guideline in the passage I read.⁴ Two directors have pledged shares as collateral, 100,000 and 112,412, together about $10.5M.⁴¹ A pledge is a loan against the stock, small here against $218M, and it is the one place where an insider’s money works against the shareholders in a fall.

What they did with the money, over a full cycle

The record is of a company that buys other banks, pays out a steady dividend and keeps the rest. From 2016 to 2025 the bank earned $3.05B, paid $1.64B in dividends and bought back $0.38B of stock, so 66% of earnings went to owners and a third was kept C (calculated).⁵³ The buybacks came in clumps: $101M in 2018, $79M in 2022 and $127M in 2025, $11M in 2021, and almost none in 2023 or 2024.⁵ In the first nine months of 2026 the bank bought 5.4 million shares for $236M at an average of $43.77, and in the third quarter 2.1 million at $46.84, which is 1.9 times the tangible book of $25.29 at the start of the quarter C (calculated).² That is a lot to pay for your own stock when my base value is $45: it adds to earnings per share, takes from tangible book per share, and at a price above value moves value from holders to sellers.

Acquisitions were paid in stock, and the stock count shows it. Shares outstanding went from 81.0M at the end of 2016 to 139.9M at the end of 2025.⁵ Net income rose 216% and earnings per share 64% over those years C (calculated). Tangible book per share rose 5.2% a year over the eight years to 2025, from $16.36 to $24.63, and the dividend per share rose from $1.32 to $1.49.⁵ Piedmont is the latest example: $281M of stock for $202M of equity, and $130M of goodwill.³ The deals added to profit, but per share the owners’ gain was a fraction of the headline. That is how a bank grows by purchase, and it is why I do not take net income growth at face value and value everything per share on today’s count.

The dividend is the part that management does well and talks about most. The proxy says 2025 was the 52nd consecutive year of increases, “a record only one other major banking company in the United States has achieved,” which is the company’s claim, and I have not checked it.⁴ I checked the years in the filings: every one of the ten went up, from $1.32 to $1.49, and 2026 is on track for a 53rd, with $1.14 declared in nine months from $1.11 and a quarterly rate of $0.38.⁵²¹² The streak is real. What is easy to miss is how small the raises are: 1.4% a year over nine years, 1 to 3 cents. In the first nine months of 2026 dividends took 41% of earnings and buybacks 62% C (calculated), so the dividend is covered and the buyback is what would give way first.² A one-cent raise keeps a streak alive in any year. It does not tell me the board would raise it by more if the bank could afford to.

VFinancials

Here are ten years in one table, as reported; ratios are my arithmetic. Tangible book and return on tangible equity start in 2017 and 2018, because the other-intangibles figure I need is not in the data set for 2016.

The pattern is a bank that doubled in size by buying other banks and kept its returns roughly level. Revenue rose 2.5 times, net income 3.2 times and shares 1.7 times, so earnings per share rose 1.64 times C (calculated).⁵ Credit was clean: charge-offs under 0.30% of loans in every year.⁵

Tangible book per share, the number a bank is valued on, grew 5.2% a year from 2017 to 2025, from $16.36 to $24.63, with one dip, in 2022, when rates rose and the bond mark fell by $328M.⁵ At 30 September 2026 it was $24.97, up 1.4% in nine months, and down from $25.29 at the end of June. The price lagged the book until this year: $34.75 at the end of 2017, $38.40 at the end of 2025, $49.23 now.¹ Over the ten years to the close in Q4 2026 the price rose 1.33 times, from $37.15, which is 2.9% a year before the dividend.¹ The highest close in that time is $49.37 on 4 August 2026, so the stock is at the top of its ten-year range.¹

Reading 2026

The quarter was good and not as good as it looks. Net interest income reached a record, but return on tangible equity fell from 15.15% to 14.68%, noninterest income fell by $7.7M, and the efficiency ratio went from 47.8% to 48.2%.² Tangible book per share fell 32 cents, from $25.29 to $24.97, in a quarter in which the bank earned 94 cents, paid 38 cents in dividends and spent $98.7M on stock at 1.9 times book.²

From reported 2026 to my starting point

Credit costs this year are unusually low, so the reported year is the wrong place to start. The provision is running at 0.11% of loans, against 0.20% over ten years and my 0.25% for the model. I start from a year with normal credit costs and no securities gains.

My normalized 2026 net income is $483M, 6% below the $514M I expect the bank to report E (estimate). On that figure the price is 13.8 times earnings, against 13.1 times my estimate of reported earnings of $3.74 a share and 13.4 times the last twelve months of $3.68.²

Capital

The capital is high, and the bank has been spending it: CET1 fell from 13.4% at the end of 2025 to 12.9%, because it returned 103% of earnings in nine months and loans grew.²³ A bank this size faces no stress test, so no loss estimate is published for it. Keeping the bond mark out of regulatory capital flatters the ratio by about 0.8 points on my arithmetic E (estimate).³ The company’s new repurchase plan covers 6.8 million shares, about 5% of the stock, and 5.1 million were left on 21 October.¹²⁶

Reported earnings and owner cash

What can an owner take out? A bank has to keep its capital growing as its loans grow. I treat the first 3% a year of growth in tangible equity as the cost of standing still, since that is about what risk-weighted assets grow with inflation alone, and count the rest of the year’s earnings as cash the owner could take. Stock pay is already in expenses, $13M in 2025, so it is deducted.⁵

So the owner’s yield is 5.7% at this price, 0.8 points above the 4.90% the ten-year Treasury paid on 4 November.¹³ Add growth of 3% to 4% in what the bank can pay out and the total is about 9%, close to my 9.2% hurdle. Dividends at $1.52 are 54% of that owner cash, which leaves the rest for buybacks.

VIValuation

Start with what would go wrong

Before any numbers, I ask what would most likely make this a poor investment from here, and work back from that. These are the three ways it goes badly wrong, in the order I think most likely.

The price falls back to the multiple the bank has usually had. The stock trades at 1.97 times tangible book. At the end of each of the last nine years it traded at between 1.56 and 2.12 times, 1.84 on average, and at 1.56 a year ago C (calculated).¹⁵ If the bank earns what it earns now and the market pays 1.5 times tangible book at the end of 2031, the return from $49.23 is about 4.7% a year with the dividends and buybacks paid on the way, which is less than the 4.90% the Treasury pays.¹³ Nothing in the business has to break for that to happen.

A real estate cycle arrives while the capital is being paid out. The bank has $9.0B of non-owner-occupied commercial real estate and $3.4B of construction and land loans, together $12.4B, which is 3.7 times its tangible equity.⁶² The company says its non-owner-occupied commercial real estate is about 301% of total risk-based capital, which matches my arithmetic only if the construction loans are in it, so I treat the $12.4B as the exposure. The agencies’ 2006 guidance uses 300% of total capital, together with 50% growth in three years, as one screen for closer review.⁶¹⁴ A loss of 5% on that $12.4B is $622M, 18% of tangible equity and 1.3 years of my normalized earnings C (calculated). The bank would survive it, with CET1 near 11%, but the shares would not trade at 2 times book while it happened. My bear case has credit costs of 1.2% and 0.9% of loans in 2028 and 2029, about $0.55B in all, earnings falling to $227M in 2028 and a value of $30.

The company pays too much, in stock or in cash. It has made 34 acquisitions, with the share count up 73% in nine years, and it is buying back stock at 1.9 times tangible book, above my base value of $45.⁵⁴² Both cost the owner who stays if the price paid is above the value received. I cannot see future deals, so I do not model them.

The first I show below. The second is my bear case. The third I flag and do not model.

What the price implies

The stock is within 0.3% of its highest close in ten years and has gained 7% since the end of September.¹² The market is paying a clear premium to the bank’s own history, in a year in which the bank did everything right. That is the debate in the price.

How I value a bank

For most businesses on this site I discount free cash flow. A bank cannot be valued that way. Loans and deposits are its raw material, so the cash flow statement moves with the size of the balance sheet. The right question is how much a shareholder can take out and still leave the bank as strong. So I work on tangible equity, as I did for JPMorgan Chase and Bank of America: an excess-return model, which values the shares as tangible book plus the present value of what the bank earns above its cost of equity.⁸⁷

A dollar of tangible equity that earns exactly the cost of equity is worth a dollar; one that earns more is worth more, for as long as it can. That gives the same number as discounting the cash the owner can take out each year, after keeping the capital ratio at a target. I do both, and they agree to the cent.

Where the base value comes from

About 51% of the value sits beyond 2036; I show that sensitivity below. Note how little of the value is above tangible book: of $45 a share, $25 is the book itself and $19 is the premium for earning about five points over the cost of equity for a decade and a smaller premium for ever.

Reverse the question: what return does the price need?

I take my base growth and capital ratios and ask a different question. What return on tangible equity, held for ten years and for ever after, makes the shares worth $49.23? I solve it.

The answer is a return on tangible equity of 14.3% every year, for ever. With the faster growth of my bull case it is 13.9%, and with the slower growth of my bear case, 15.2%. There is a shorter route. If the bank holds a return R for ever and grows at g, the multiple of tangible book is (R minus g) over (cost of equity minus g). At today’s 1.97 times, R has to be 14.7% if growth is 3.5%, 15.2% if it is 3% and 13.3% if it is 5%. Either way the price needs the bank to keep earning about what it earns today for a decade or more; the eight-year average is 13.8%.

That is the honest headline for this page. The price pays for the bank as it is today, at the top of its range, and for holding it there: the third-quarter return of 14.7% is the top of my eight-year range and the average is 13.8%. Bank of America’s price needs about 14.6% against a recent 14.9%, and JPMorgan’s needs about 19% against 20%.⁷⁸

The discount rate

I use the same method as on my JPMorgan Chase and Bank of America pages. The capital asset pricing model with the 4.90% ten-year Treasury and Damodaran’s 4.20% implied equity risk premium gives 8.3% on United’s beta against the market over 36 months of 0.80, 8.0% over 60 months at 0.74 and 9.2% over 119 months at 1.03.¹³¹⁵¹⁶ The rule on those pages is to take the highest of the three, because a bank’s equity is the thin top layer of a very leveraged balance sheet and its beta tends to read low in calm years. That gives 9.2%. JPMorgan’s three betas on the same method were 0.97, 1.01 and 1.08, for 9.5%, and Bank of America’s 1.19, 1.21 and 1.37, for 10.6%.⁸⁷

I use 9.2%, the same rule as on the other two pages. A beta computed on a regional bank over the recent window does not see three things I can see: $12.4B of commercial real estate and construction loans against $3.4B of tangible equity, 32% of deposits uninsured, and a bank outside the stress-test regime, so that nobody has published a loss estimate for it.⁶² I do not pad the rate for them. They are a loss that may or may not arrive, and I put it in the bear case, where it costs $15 a share, rather than charge for it every year in every case. For reference, at 10.0% the base would be $40 and the price 23% above it.

So the price is reached by a terminal return of 14.0% instead of my 12.5%, or by a required rate below 9.0% with my 12.5%. Either way it pays for a bank that keeps its present return after 2036, which is a longer stretch of today than I am willing to assume, because credit and funding are hard to see from outside.

Three cases

My base case is the bank keeping what it has and giving a little back. Revenue grows 4.5% in 2027 and eases to 3.5% by 2031, about 3.8% a year on average, well under the 10.7% of the last decade, which was mostly acquisitions, because the margin is already at 3.86% and I do not assume it widens. The efficiency ratio drifts up from 48.2% to 51%, about where it was in 2018 and 2019, since costs are already rising faster than revenue.² Credit costs run at 0.25% of loans, above the 0.11% of 2026, and above the 0.13% charge-off average and the 0.20% provision average of ten years. The result is a return on tangible equity of 14.4% in 2027, 14.6% in 2031 and 14.2% in 2036, an average of 14.4%, above the eight-year average of 13.8%.

The bull case has the margin and the cost advantage persist and widen: revenue growth of 5.5% easing to 4%, an efficiency ratio falling to 47%, credit costs of 0.20% and tangible equity run down to 10.0% of risk-weighted assets. That gives an average return of 16.6%, above anything in my eight-year table, and a value of $57. The bear case has a commercial real estate cycle in 2028 and 2029, an efficiency ratio of 53.5% afterwards, higher capital, and a return that settles near 12% and then 9.5% for ever. Even the bear is a profitable bank, worth about 1.2 times tangible book.

What I can’t judge

Some inputs I can’t know. The first is credit, and it is the main one. I can see that $12.4B of the loans are commercial real estate and construction, that the company says the non-owner-occupied part is about 301% of total risk-based capital, that office is $0.7B, 49% of it around Washington and none in the central business district, with $54.8M set aside, and that the 60 largest office loans, three quarters of the office total, had a weighted loan-to-value of 58% when they were made.⁶ What I cannot see is the current value of the buildings behind $9.0B of loans, the debt-service cover on them, or when they mature. I cannot see how many of the multifamily loans, 29% of the book, are still in lease-up, or how the $1.6B of hotel loans are doing.⁶³ I cannot tell how good the sponsors’ guarantees are, and the 2025 losses came from a loan with a full guarantee that failed.³ I cannot say whether an allowance of 1.31% of loans is enough, with 30-to-89-day delinquencies up from 0.28% to 0.38% in a quarter, and I cannot judge what a cut in federal spending would do to the Washington and Baltimore market, where 34% of the loans sit.⁶

The second is the funding. A quarter of deposits cost nothing, and 32% are uninsured; the bank has $22.4B of liquidity. What I cannot see is how many of the free deposits would move if rates stayed high, or if the Fed cut and the bank’s own sensitivity of −2.2% in year two turned out low.⁶ The third is the bond mark at 30 September, which I estimate and will check against the 10-Q. The fourth is what it does next with its capital, since a large deal changes everything on this page.

What the price returns at different multiples of tangible book

On my base case, with the cash paid out on the way, from $49.23. The exit date is the end of 2031 or the end of 2036. Tangible book per share in the base is $28.62 at the end of 2031 and $34.83 at the end of 2036.

The multiple does the work. Held at 2.0 times, the return is 9.5% to 2036, just above my hurdle. At the multiple the stock had a year ago, 1.56 times, it is about 7.5% to 2036; at 1.25 times, 6.0%. At 1.0 times, the book alone, 4.5%.

Multiply it out

If the case needs several things to go right at once, the odds multiply. For the price to be fair on my 9.2%, three things have to be true together for a decade. The return has to average about 14%, against an eight-year best of 14.8%. Credit has to stay at or below 0.25% of loans, with no real estate cycle in ten years. And the market has to keep paying about 2 times tangible book. My guesses are 5 chances in 10, 6 in 10 and 5 in 10. Multiplied, that is 15%, about 1 in 7. They are not independent, so the true figure is higher. I do not use the number in any value; I use it to see that the price is a bet on three things at once.

Quality and price are two questions

How good the bank is and how good the price is are not the same question. On quality, it looks like a good business: a return on assets of 1.50%, a cost base that spends 48 cents on the dollar, a charge-off average of 0.13% of loans, a deposit lead in its home markets and a dividend raised for 52 years.²⁵⁴ On price, this looks like a good bank at a full price. At 1.97 times tangible book the earnings yield is 7.5% and the owner cash yield 5.7%, the market is paying more for it than for Bank of America, and the stock sits at its ten-year high.¹⁷ A good bank at a full price earns the market’s return, not the bank’s. The price pays for the quality: my base case earns 8.6% a year, a little under the 9.2% hurdle, and only the bull case earns clearly more.

The alternative

The next best use of the same dollar is not only the Treasury at 4.90%. On my own pages, the yearly return at the price in the base case is 8.6% here, against 8.7% at JPMorgan Chase and 10.1% at Bank of America, each at its own hurdle of 9.2%, 9.5% and 10.6%, so that the shortfall to the hurdle is 0.6, 0.8 and 0.5 points.⁸⁷ On price against value, United is where Bank of America is: the price is 9% above my base value here and 8% above there, against 15% at JPMorgan Chase. Bank of America also carries a bond problem I cannot value precisely. United’s advantage is simplicity and a thick capital cushion. It is not price.

The margin of safety

I ask for a price at which my bear case still earns about my required return. At $29, 36% under my base value, the bear case returns about 9.4% a year, close to my 9.2% hurdle, and my base case returns 12.9%. The price is 70% above that. The stock last closed below $29 in November 2023.¹ I do not expect the price to go there, and I do not say it will. It is the price at which I would not need the credit cycle to stay away.

Where this could be in 5 and 10 years

These are my estimates of what a share is worth at later dates, from the model above. The value at a date is the present value, at that date, of the owner cash that comes after it, on today’s 134.9M shares and in that year’s dollars. It includes the value of the bank beyond 2036 and leaves out cash paid before the date, which the fourth column adds up from now to the end of 2036, including the fourth quarter of 2026.

In the base case a share is worth about $48 at the end of 2031 and $55 at the end of 2036, with $34 of cash paid out on the way, which is a return of 7.7% a year from today’s price. The bull case earns 11.2%, above my hurdle, and the bear case 2.1%.

VIIThe case for the price

What most people miss

First, the bond loss is small, and it is already counted. Securities are $3.84B, 11% of assets, almost all of them available for sale, so the mark is in book value; the company opted out of counting it in regulatory capital, and the after-tax loss was $143M at 30 June, about 4% of tangible equity C (calculated).²¹⁰³ Bank of America’s is 54% of its tangible equity.⁷ Second, the office loans that scare people in this sector are 8% of the non-owner-occupied book: $0.7B, 2.7% of all loans, with none in the central business district of Washington and $54.8M reserved against them, 7.9%; the largest category is multifamily at 29%.⁶ Third, deposit costs fell while a quarter of the deposits cost nothing: the cost was 1.88% in the third quarter against 2.14% a year earlier on my arithmetic, net interest income rose 5%, and the company’s own sensitivity has a 100 basis point rise adding 1.9% to net interest income in a year C (calculated).²⁶

The bull case

The case is a simple, low-cost bank with a thick capital cushion that keeps earning in the mid-teens on tangible equity, buys back about 5% of its shares a year, and raises a dividend that has gone up for 52 years. The clearest version is the company’s own. On 22 October 2026 the chief executive, Richard M. Adams, Jr., said in the third-quarter release that record net interest income, a wider margin and organic growth drove earnings, and credited expense control and credit underwriting.² The 2026 proxy statement, filed on 1 April, says the dividend increase for 2025 was the 52nd in a row.⁴ On 24 August the board approved a new plan to repurchase up to 6.8 million shares, about 5% of the stock.¹² Of thirty managers whose 13F filings I searched, none held the stock at 30 June 2026.¹⁷ I found two broker targets on undated aggregator pages and could not date or confirm them from a primary source, so I leave them out.

Against reality

Much of the bull case is true today: net interest income is at a record, the margin is widening, credit is quiet, the buyback is running, and the capital is thick.² If the margin holds near 3.9%, the efficiency ratio falls to 47%, credit costs stay at 0.20% of loans and the bank keeps buying back shares, my bull case gives a return averaging 16.6% and is worth $57 E (estimate). If the bank does that and hits its targets, $49.23 is 14% below that value. The price needs 14.3% held for ever, against 14.7% today and a best year of 14.8% in the eight I can compute. That is a return the bank has earned, held for a decade. The bull case needs what the bank has not done: a return above anything in its record, with a cost ratio that is already the best in ten years.²⁵

What a 10x would take. Ten times $49.23 is $492 a share, about $66.4B of market value, against $6.64B today C (calculated). At 13.4 times earnings, the current multiple, that needs about $5.0B of net income, 10.3 times my normalized 2026 figure of $483M. At today’s 1.97 times tangible book it needs about $34B of tangible equity, ten times the $3.37B the bank has. At a net margin near 37% of revenue, $5.0B of profit means about $13.3B of revenue, 10.3 times today’s, which takes about 30 years at 8% a year and about 23 years at the 10.7% the company managed over the last decade, nearly all of it by acquisition C (calculated).²⁵ The stock rose 1.33 times from $37.15 in November 2016, or about 2.9% a year before dividends.¹ A tenfold move is not plausible for a bank of this size, and I do not think it is a useful way to think about this stock.

Where I come down

What moved me: a bank with a clean ten-year credit record, a cost base that is the best of the decade, a capital cushion of about $1.6B over its minimum, and bond losses that are small and already in book value. What did not: the price is 9% above my base value, it needs 14.3% held for ever, close to its best year of 14.8%, and a third of the loans are real estate that can only be judged after a cycle. What would raise my values: two more years of a margin above 3.9% with an efficiency ratio under 48.5% and credit costs under 0.20%, or a price near $29, where the bear case still earns my rate.

VIIIRisks, and what would change my view

The three ways this goes badly wrong are at the start of Valuation. These are the rest.

• Real estate and geography. Non-owner-occupied commercial real estate and construction loans are $12.4B, 3.7 times tangible equity, and about 86% of them are in the Southeast and the Washington and Baltimore area C (calculated).⁶² A federal spending cut would hit Washington, and storms, which the 10-K lists among its risks, would hit the coast.³
• Funding and rates. A quarter of deposits pay nothing and 32% are uninsured. Time deposits are $4.9B with a five-month average maturity.⁶ The company’s own sensitivity is positive when rates rise and slightly negative when they fall: −0.2% of net interest income in year one and −2.2% in year two for a 100 basis point cut.⁶
• Capital return. The bank paid out 103% of earnings in nine months and has a new plan for 6.8 million shares. At 1.9 times tangible book each dollar of buyback lowers tangible book per share, and the cushion over the minimum shrinks.²¹²
• Acquisitions and goodwill. There have been 34 deals, with $2.0B of goodwill, 37% of equity, and the share count up 73% in nine years.³⁴⁵ A bigger deal changes the bank, and the 2025 Piedmont deal added $2.1B of loans in a new city.³
• The bond mark. Available-for-sale bonds carry $218.5M of gross unrealized losses at an average duration of 4.0 years.¹⁰⁶ If rates rise it grows and cuts tangible book. The opt-out that keeps it out of regulatory capital is an election the rules allow, and if that changed, the CET1 ratio would be about 0.8 points lower on my arithmetic E (estimate).³
• Leadership. The executive chairman is 79 and the chief executive, who shares his name, is 57.⁴ The proxy says the committee reviewed the management succession and development plan in November 2025.⁴ I cannot judge it from outside.
• Competition and technology. Online banks and money funds compete for deposits, and the 10-K lists cyber incidents among its risks.³
• What I could not see. The September 10-Q was not yet filed, so I used the earnings release and slides; the AOCI figure, risk-weighted assets and capital ratios for 30 September are estimates, and the 10-Q could move them.¹¹²