How to Value Bank Stocks: P/B, ROE, and the Justified Price-to-Book Formula

Run a bank through a standard stock screener and it will look like either a screaming bargain or a broken business — often both at once. Single-digit P/E, enormous debt, negative free cash flow, an EV/EBITDA figure that means nothing at all. Banks are the clearest case in investing where the valuation toolkit itself has to change: the numbers that matter are price-to-book and return on equity, and the connection between them is one of the most elegant formulas in valuation.

This guide is part of our series on sector-specific valuation — adapting the general playbook of valuing a stock to industries where the default tools mislead. Here we cover banks: why the standard models fail, how a bank actually earns money, why book value is the anchor, and how to tell a genuinely cheap bank from a value trap.

Why Standard Valuation Tools Break for Banks

For a normal company, debt is financing: the business makes cars or software, and borrowed money is one way to fund the machines that make it. For a bank, debt is the raw material itself. Deposits and other borrowings are the inputs the bank buys at one price (the interest it pays savers) and sells at a higher price (the interest it charges borrowers). Asking a bank to operate without debt is like asking a steel mill to operate without iron ore.

That single fact wrecks the standard toolkit. Enterprise value multiples like EV/EBITDA assume you can split a firm into an operating business plus a financing decision. For a bank the split is impossible: the 'debt' is the operating business, and interest expense is not a financing cost but the cost of goods sold. EBITDA, which adds interest back, literally deletes the bank's core expense line and produces a number with no economic meaning.

Free-cash-flow models fail for a subtler reason. A classic DCF starts from operating cash flow and subtracts capital expenditure — but a bank's capex is trivial (branches and software, not factories), and its operating cash flow swings wildly with deposit inflows and loan originations. A bank that gathers $5 billion of new deposits shows a huge cash inflow while becoming riskier; one that grows its loan book shows an outflow while building its earnings engine. 'Free cash flow' is simply ill-defined for a bank — what it can hand to shareholders is set by earnings and regulatory capital, not by the cash flow statement. That is why bank valuation shifts to equity-based methods: dividends, earnings, and above all book value.

How a Bank Actually Makes Money

A bank has three profit levers. The first and largest is the interest spread, measured by net interest margin (NIM): the difference between what the bank earns on its loans and securities and what it pays for deposits and other funding, expressed as a percentage of its interest-earning assets. A NIM of 3% means that for every $100 of loans and securities, the bank clears $3 a year of net interest income before costs. The second lever is fee income — payments, wealth management, investment banking — which uses little balance sheet and steadies earnings when rates squeeze the spread. The third works in reverse: credit losses. Some borrowers always default, so a slice of the spread is consumed every year by provisions set aside for bad loans.

That last line is the essence of banking: small margins multiplied by large leverage — and the source of the sector's fragility, since at 10× leverage, losses equal to 10% of assets wipe out all equity. Business models differ in mix: a giant like JPMorgan layers large fee businesses on top of the spread, while a more deposit-driven franchise like Bank of America is more sensitive to the rate cycle. But the arithmetic of spread, fees, losses, and leverage is universal.

Why Book Value Is the Anchor

For most companies, book value is an accounting relic. Coca-Cola's brand, Microsoft's installed base, a railroad's century-old land — none of it sits on the balance sheet at anything close to real worth, so price-to-book ratios of 5× or 10× tell you little. Banks are the exception. A bank's assets are financial — loans, bonds, and cash — carried at values close to what they could actually fetch: securities largely marked to market, loans at cost minus a reserve for expected losses, deposits at face value. When the accounting is honest, a bank's book value is a reasonable estimate of what shareholders would recover if the balance sheet were unwound — which is exactly what book value is supposed to mean.

This is why P/B is the sector's headline multiple, and why reading the balance sheet matters more for banks than for any other industry. One refinement: strip out goodwill. When a bank overpays to acquire another, the premium sits on the balance sheet as goodwill — an asset that can never absorb a loan loss or be paid out to shareholders. Tangible book value removes it, and price-to-tangible-book (P/TBV) is the stricter, more comparable metric, especially for serial acquirers whose stated book value is padded with deal premiums.

The Justified P/B Formula: Where P/B Meets ROE

A bank's fair multiple of book value is not a matter of taste — it follows from three numbers: its return on equity, its cost of equity r (the return investors require for holding a risky bank), and its long-run growth rate g. Derived from the dividend discount model — earnings are ROE times book value, dividends are the earnings not retained for growth — the discounted stream collapses into one line.

The intuition works without a calculator: a bank earning more on its equity than investors require creates value with every dollar it retains, so each dollar of book value is worth more than a dollar — P/B above 1 is justified. A bank earning less than its cost of equity destroys value as it grows — P/B below 1 is not a market error but a fair verdict. Growth only amplifies the effect in whichever direction the ROE points, which is why 'cheap' and 'expensive' can never be read off the P/B ratio alone.

Two Banks, One Question: Which Is Cheaper?

This is the single most common mistake in bank investing: sorting the sector by P/B and buying the bottom of the list. The market usually prices low-ROE banks below book for good reason, and the apparent discount is often more than offset by weak profitability. The real question when judging whether a bank is overvalued or undervalued is always the pair: what P/B am I paying, and what sustainable ROE am I getting for it?

Asset Quality: When Cheap Is Cheap for a Reason

The justified P/B math assumes the 'B' is real. Sometimes it is not: loans are carried at cost minus a management-estimated loss reserve, so stated book value is only as honest as the credit assumptions behind it. Three numbers test them. Non-performing loans (NPLs) are loans where the borrower has stopped paying — typically 90+ days overdue — as a percentage of total loans; for a healthy bank in normal times, around 1% or less. The loan-loss allowance is the reserve already deducted from book value to absorb expected defaults. And coverage — allowance divided by NPLs — tests adequacy: coverage well below 100% means known problem loans exceed what has been set aside, and future losses will eat directly into book value.

Capital Requirements: CET1 in Plain Language

Because leverage is both the engine and the danger, regulators cap it. The key gauge is the CET1 ratio — Common Equity Tier 1. In plain language: the bank's highest-quality capital (common equity minus goodwill and other items that could not absorb a loss) divided by risk-weighted assets, where each asset counts according to its riskiness — government bonds near zero, mortgages partially, unsecured consumer loans heavily. A CET1 ratio of 12% means $12 of true loss-absorbing equity per $100 of risk-adjusted exposure. Regulators set a minimum — typically around 10-11% with buffers for large banks — and breaching it means forced dividend suspension, blocked buybacks, and dilutive emergency capital raises.

For a valuation, CET1 matters twice. First, it caps growth: a bank can only expand its loan book as fast as it can generate or raise capital to stand behind it. Second, it governs payouts — dividends and buybacks can only be funded from capital above the regulatory minimum, so a bank sitting barely over its requirement has little distributable profit regardless of what its earnings per share suggest. Between two banks with similar ROE, the one with the bigger CET1 cushion deserves the higher multiple: its earnings are more distributable and its equity less likely to be diluted in a downturn.

Earnings Cross-Checks: Normalized P/E and Dividend Capacity

P/B and ROE are the anchor, but two earnings-based cross-checks keep them honest. The first is a P/E ratio on normalized provisions. Bank earnings are cyclical mainly because credit losses are: in a benign year provisions fall near zero and flatter earnings; in a recession they erase them. Rather than capitalizing a peak or trough year, re-estimate earnings at a mid-cycle provision rate (for many diversified banks, roughly 0.3-0.5% of loans annually) and put the multiple on that. A bank on 7× boom-year earnings may be on 11× normalized earnings — still reasonable, but a different proposition.

The second is dividend capacity. Mature banks are natural income stocks, and the payout ratio plus the CET1 surplus define what is sustainably distributable: a bank earning a 12% ROE that must retain 3 points of it to fund growth can pay out roughly three-quarters of earnings across dividends and buybacks. Valuing that stream directly is a useful triangulation — banks are one of the few sectors where the dividend discount model genuinely applies, precisely because free cash flow is undefined and dividends are the cleanest measurable cash flow to owners.

Red Flags When Valuing Bank Stocks

Rapid loan growth is the classic warning. A bank growing its loan book 25% a year in a market growing 5% is almost certainly winning business by underpricing risk, and the losses arrive with a two-to-four-year lag — long after the growth has been applauded. Be equally wary of reserve releases inflating earnings: when a bank reverses previously booked provisions, the release flows through as profit, so a chunk of reported earnings can come from an accounting judgment rather than the business — and it is non-repeatable by construction.

Concentration is the third flag. A bank with half its loan book in one segment — commercial real estate, construction, energy, a single region — is a leveraged bet on that segment, and heavy reliance on large uninsured deposits or wholesale funding can turn a solvency question into a liquidity run within days. Fourth, a P/B far below peers without a visible cause is itself the red flag: a bank at 0.5× book while similar peers fetch 1.0× usually has a problem — embedded securities losses, a deteriorating loan segment, or a looming capital raise — that has not yet surfaced in the reported numbers. Treat an unexplained discount as a research assignment, not a buy signal.

Putting It Into Practice

A practical sequence for any bank stock: start with the sustainable ROE — averaged over a full cycle, haircut for reserve releases and one-off gains. Estimate a cost of equity (10% is a common baseline; more for smaller or concentrated lenders) and a modest growth rate, then compute the justified P/B and compare it with the market's, using tangible book for both. Stress the 'B': check NPLs, coverage, and CET1 to make sure the book value would survive a recession. Cross-check with a normalized P/E and dividend capacity. This mirrors how FairPriceIndex treats financial-sector stocks: because a standard free-cash-flow DCF is unreliable for banks, the FPI valuation model leans on the relative-valuation pillar — book-value and earnings multiples benchmarked against financial-sector peers rather than EV/EBITDA — blended with analyst consensus to produce a fair value.

Banks are the deepest example of a broader rule: the valuation tool has to fit the business model. The same logic, adapted differently, applies to REITs (where depreciation distorts earnings) and cyclical stocks (where trailing multiples invert). To put the framework to work, browse fair values across 37,000+ stocks and see how the banks you follow stack up on P/B, ROE, and fair value.