Case: 2008 Bank Balance Sheets

A bank is a levered portfolio, not a business with a multiple

The ordinary valuation framework has to be inverted for a bank. Earnings multiples are unhelpful because the earnings are a spread between what the assets yield and what the funding costs, scaled by leverage, and a small change in credit losses swamps the operating result. Book value is the anchor, and the question is not what the bank will earn next year but whether the book value is real: a bank trading below book is the market saying it doubts the assets, and a bank trading far above book is the market saying the franchise — deposits, relationships, fee income — is worth more than the assets it holds. The balance sheet is short-term-funded and long-dated in its assets, which is the definition of a liquidity mismatch and the source of the business model’s profit. Deposits are insured up to a limit and are therefore the stickiest funding; wholesale borrowings, commercial paper and repo are not insured and are withdrawn at the first sign of doubt. The share of funding that is uninsured and short is the number that decides whether a loss becomes a failure, because a portfolio of good assets funded with money that runs is a portfolio that must be sold. Regulatory capital is defined for the same reason: common equity against risk-weighted assets, with a minimum and a buffer, is a rule about how much loss the institution can absorb before the question of its survival is asked. The rule is coarse and the risk weights can be gamed, which is why the market looks at the leverage ratio and at tangible equity as well — and why the disclosed detail that matters most in a stress episode is the asset composition and the funding mix rather than the headline capital number. The loss, against equity and against assets — Assets $1bn, equity $40m: a 4% asset loss is the entire equity · Assets $1bn, equity $100m: a 4% asset loss is 40% of equity ← · The same 6% loss: 1.5× equity at 25× leverage, 0.6× equity at 10× · What decides survival: the equity layer and the funding mix, not the loss itself Book value and market value diverge exactly when they matter most. In a panic, the assets are marked at prices that only exist while nobody else is selling, so a "cheap" bank can be correctly priced — and a solvent one can be illiquid at the same moment.

How a loss becomes a failure

The sequence is mechanical once the funding structure is known. Losses on the asset side reduce equity; reduced equity makes lenders and counterparties demand more collateral or refuse to roll; the refusal forces sales; the sales push prices down, which marks the remaining assets lower and reduces equity again. Each step reinforces the last, and the amplification is a feature of the system rather than an accident, because the same security is collateral for many borrowings and every participant is measuring its own exposure against the market price of that collateral. Two features made 2008 worse than the arithmetic alone implies. The first was the funding: a large share of the system’s assets were financed with overnight and short-dated repo, which is functionally a deposit that can be withdrawn without notice and, unlike a deposit, is not insured and not limited by a branch network. A run on repo does not look like a queue outside a building; it looks like a collateral schedule that nobody will renew. The second was the accounting and the structure: off-balance-sheet vehicles held assets the sponsor was implicitly obliged to support, so the reported leverage understated the economic leverage, and the losses arrived in a place the analysis had not looked. The resolution is the part that matters for a valuation, because it is where the value either survives or is transferred. Recapitalisation dilutes existing shareholders; a forced merger transfers the franchise to a competitor; nationalisation transfers it to the state; and failure of the firm with a rescue of the creditors leaves the equity worthless. In every case the operating business continued — branches opened, deposits were insured, customers were served — while the equity was either wiped out or diluted, which is the same conclusion as F19 by a different route: the loss accrues to whoever holds the residual claim, not to the enterprise. • Losses reduce equity; reduced equity stops the funding; no funding forces sales; sales create more losses. • Overnight repo is an uninsurable deposit, and a run on it is invisible until collateral stops rolling. • Off-balance-sheet vehicles understated the leverage that the losses actually reached. • In every resolution the business continued and the equity was wiped out or diluted. The policy response matters to the analysis too: deposit insurance removes the run risk on insured deposits, central-bank facilities supply liquidity against collateral, and capital rules force the equity layer to be thicker. Each changes the probability of the sequence, not its existence.

The test applied to a bank today

Take the disclosed capital, strip out the goodwill and intangibles to get tangible common equity, and divide it by tangible assets to get a leverage ratio you can compare across banks. Then ask what loss that equity absorbs: at 6% tangible equity against assets, a 6% loss on the portfolio is total, so the question becomes what a plausible credit shock would do to the loan book — a loss rate, applied honestly, rather than a scenario label. Then the funding: what share of the balance sheet is funded by uninsured, short-dated borrowings, and how much of the asset side is marked to market? A bank funded by insured deposits and holding loans at amortised cost is a different proposition from one funded by wholesale money and holding securities at fair value. Then the concentration: a bank with most of its assets in one sector, one geography or one asset class is a portfolio of the same risk, and the diversification claim in the annual report can be tested against the segment disclosure. And finally the off-balance-sheet and contingent items, because guarantees, unfunded commitments and the assets of sponsored vehicles are where the leverage hid last time — the disclosure exists, in the notes, and the question is whether the obligation is contingent in form or in fact. The result is not a multiple. It is a statement of the form: this bank can absorb a loss of about X% of its portfolio, its funding is Y% uninsured and short, its marks are Z% of the balance sheet, and here is the loss rate at which it needs capital. That is a valuation of the equity as an option on a levered portfolio, and it is a better description of a bank than an earnings forecast, because the earnings are a derivative of the same credit cycle that decides whether the equity survives it. A bank, read as a levered portfolio — Tangible common equity ÷ tangible assets: the loss the equity absorbs before capital is needed · Uninsured short-dated funding as a share: the size of the run risk · Share of assets marked to market: how quickly a panic reaches the equity ← · Concentration by sector and geography: whether the loan book is one risk or many · Contingent and off-balance-sheet obligations: the leverage the headline ratio does not show A low price to book is not automatically a bargain and a high one is not automatically expensive. Low means the market doubts the assets; high means it values the franchise. The analysis is deciding which of the two you believe, and what the loss rate would have to be to change your mind.

Reading a bank in its own language

A bank valued as a levered portfolio has to be read through its own reporting, and the terms that matter are not the ones a generalist analyst uses. Four sets of numbers carry nearly all the information, and they are found in a different order from the one a conventional read would follow. The first is **funding**. A bank’s business is the spread between what it pays for liabilities and what it earns on assets, so the deposit franchise is not a background detail — it is the product. The questions are how much of the funding is deposits, how sticky they are, what they cost, and how much comes from wholesale markets at a market rate. A bank funded with cheap, stable deposits earns a wider margin when rates rise; one funded with wholesale money reprices immediately and does not. The loan-to-deposit ratio and the reliance on borrowings summarise it. The second is the **loan book**. Loans are carried net of an allowance for expected credit losses, and the two numbers to read against each other are non-performing loans and that allowance. Their ratio is the **coverage ratio**, and it is the single most informative line about whether the reserve is adequate: non-performers rising while coverage falls is the earliest visible sign of a book deteriorating, and it appears before it reaches the income statement. Loans are mostly held at amortised cost, which means the market value of a fixed-rate book stays invisible until something forces a sale — the exposure that a sharp rate move reveals. The third is **capital**, and it comes in two measures that can disagree. Regulatory capital is expressed against risk-weighted assets, which scale the book by perceived risk rather than by size, so the ratio depends on the risk weights themselves — a matter of policy as much as of accounting. The leverage ratio measures the same equity against total assets, without the weighting, and it is the backstop. A balance sheet can be comfortable on one and tight on the other, and the two together describe how much room there is to lend and to return capital. The fourth is **liquidity**, and it is the one that history says to read first. Ratios of liquid assets to short-term obligations, and of stable funding to the assets it supports, describe whether the bank can survive a week in which depositors leave. Capital is a buffer measured over months; liquidity is survival measured over days, and a bank fails on the second before it fails on the first. The practical order is therefore funding, then the loan book, then capital measured both ways, then liquidity, then the securities portfolio and its unrealised marks — and only then the multiple. The valuation ratio is a summary of a judgement about all of that, and it means nothing before the judgement. • The deposit franchise is the business: stickiness and cost decide how a rate move lands. • Non-performers against the allowance — the coverage ratio — is the earliest read on the book. • Risk-weighted capital and the leverage ratio can disagree, because one depends on risk weights. • Read liquidity before capital: a bank runs out of days before it runs out of months. The unrealised mark on a held-to-maturity securities portfolio is the quietest line in a bank’s accounts and the one that can turn a comfortable capital ratio into a fundraising within a quarter. Find it in the notes, not on the face of the balance sheet.

Sources

• Leverage, capital ratios and the arithmetic of a bank failureStandard bank analysis; Basel capital-framework definitions of common equity tier 1
• Funding runs, repo and the shadow banking system in 2007–2009Gorton & Metrick (2012), “Securitized Banking and the Run on Repo”; Financial Crisis Inquiry Report (2011)
• Mark-to-market accounting and fire-sale externalitiesPlantin, Sapra & Shin (2008); standard accounting literature on fair value in illiquid markets
• Deposit insurance, liquidity facilities and the policy responseFederal Reserve and FDIC documentation of the 2008–2009 facilities