Valuing Banks, REITs and Cyclicals

Banks: a spread business with a regulator

A bank takes deposits and makes loans, and earns the difference. That makes its debt the input rather than the financing, and it breaks the enterprise-value machinery entirely: adding deposits to a market capitalisation to compute an EV would be like adding the inventory of a retailer. Which is why banks are valued on the things that survive that structure — price to book, and return on tangible equity against the cost of equity. A bank earning a return above its cost of equity should trade above book; one earning below should trade below it, and that single relationship explains most bank valuations. The second half is the regulator, and it is not decoration. Capital requirements set a floor on how much of the balance sheet must be funded by equity, which caps the leverage and therefore the return on equity a bank can produce. When a regulator raises the requirement, the bank must either raise equity — diluting the return — or shrink the balance sheet, which reduces the earnings that the return is computed on. So the model is not a forecast of profits but a forecast of spread, credit losses, and the capital the regulator will let the bank distribute. Credit losses are where the analysis lives, and they are cyclical by construction. In a benign year provisions are small and the return on equity is flattering; the useful question is what the loan book would have to lose for the capital position to become uncomfortable, and whether the reserve policy anticipates that or lags it. Which is why bank disclosures are read backwards: the allowance for credit losses against the non-performing loans, the composition of the book by collateral type, and how much of the loan book is concentrated in the one area that would suffer in a recession. Two bank-specific warnings. Reported book value can be optimistic when the loan book has not been marked down, and tangible book value — which strips out goodwill — is the number to use. And diluted share counts matter more than usual here, because a capital raise is the bank’s standard response to a problem.

REITs and cyclicals: two different repairs

A REIT owns cash-producing property, and its earnings are distorted in a specific direction: depreciation is charged against income even though well-maintained buildings often hold or gain value, so reported earnings understate the cash available to distribute. The industry’s own measure, funds from operations, adds depreciation back and removes gains on property sales, which makes it closer to an operating measure. Adjusted funds from operations then subtracts the recurring capital expenditure needed to keep the buildings competitive — the same distinction as free cash flow against operating cash flow, applied to property. The metrics that follow are cash-flow metrics rather than earnings metrics: the payout ratio measured against AFFO rather than earnings, the cap rate as the property yield, and net asset value as the appraisal value of the portfolio against the price of the shares. Which brings in the essential caveat: a REIT’s value is very sensitive to interest rates, because the properties are long-duration assets financed with debt, and because the dividend is a large share of the total return. Rising rates hit the discount rate and the financing cost at the same time. A cyclical breaks a different assumption. Its earnings are not a description of the business at all; they are a description of where the cycle is, and a trailing multiple computed in a boom year looks absurdly low while the same company at 8× in a bust looks expensive. The repair is normalization: value the business on mid-cycle earnings, on replacement cost of the assets, or on a multiple of peak earnings applied with the knowledge that you are buying the peak. Which is why the useful question about a cyclical is not what it earned last year but what it would earn at a normal level of utilization, and whether it can survive the trough that has not arrived yet. Three industries, three tools — A bank: Price to book against return on tangible equity; credit losses and capital are the analysis · A REIT: Price to AFFO, cap rate, NAV; interest rates are the dominant variable ← · A cyclical: Mid-cycle earnings, replacement cost, or a trough analysis of survival · What none of them want: An unrestricted DCF on this year’s cash flow and a sector P/E comparison In a boom, a cyclical’s P/E is low and its earnings are above sustainable; in a bust the P/E is meaningless because the denominator collapsed. A screen that ranks by earnings multiple will systematically buy cyclicals at the top and avoid them at the bottom.

Two more models: insurers and asset managers

The two financial businesses left out of the bank lesson are worth their own paragraph, because each one hides its economics behind a statement that reads like an ordinary company’s. An **insurer** collects premiums today and pays claims later, so the accounting is dominated by reserves that are estimates about the future. Two numbers do the work. **Book value**, because the assets are financial and marked close to reality, which is why price-to-book is the starting multiple. And **float** — the money held against future claims — because a policyholder’s premium is money the insurer gets to invest before it pays out. An insurer earning a healthy investment return on a large float is a different business from one whose underwriting loses money on every policy and hopes the portfolio covers it. Reading the combined ratio separates them: below 100 means the underwriting itself is profitable. An **asset manager** is simpler and stranger: its revenue is a percentage of assets it does not own, so the whole valuation turns on flows and on the mix of the assets. Fee rates fall as money migrates to index products, which means asset growth does not translate into revenue growth at the old rate. And because the costs are largely people, a fee-compression year shows up as margin decline rather than as a loss. The right multiple here is on fee-related earnings, not on headline profit, and the right question is whether the assets are sticky — which is a question about the client base, not the market. Which metric, which sector — Bank: price to tangible book, plus the capital ratio · Insurer: price to book, plus the combined ratio and float growth ← · Asset manager: fee-related earnings, plus net flows and fee rate ← · REIT: FFO and the discount to net asset value · Cyclical: normalised earnings, never a peak-year DCF The pattern under all of these: find the number the industry itself manages to, then check whether the market is pricing something that number cannot see. For insurers that is reserve adequacy; for asset managers it is the fee rate two years from now.

Regulated returns and the bond proxy

Utilities, pipelines, water and many telecoms earn returns that are set, in whole or in part, by a regulator. The regulator allows a return on a rate base — the capital invested in the network — and the utility earns roughly that return on that base. This changes which variable drives the value. It is not the growth of demand, which is low and stable; it is the **rate base** and the **allowed return**, both administrative facts rather than competitive outcomes. A utility that invests in its grid grows its rate base and therefore its earnings, provided the regulator keeps approving the spending and the return. The consequence is that these shares behave like long-dated bonds with an equity option attached. Their earnings are predictable and their cash flows long, so their value is unusually sensitive to the discount rate — which is why they fall hard when yields rise even though their businesses are recession-proof. The equity part is regulatory risk: a stricter allowed return, a disallowed expense, or a political intervention can cut the cash flows the maths depends on. A valuation of a regulated business therefore rests on two judgments, the trajectory of interest rates and the posture of the regulator, and only one of them is in the filings. Two metrics keep the analysis honest here. **Regulated asset base per share** and its growth describe the engine; **the spread between the allowed return on equity and the cost of equity** describes whether the business is creating value as it invests. A utility growing its rate base while earning below its cost of capital is destroying value with every dollar of capex, which is invisible in an earnings chart. For a regulated business, ask what the regulator allows, not what the market grows. The regulatory compact is the business model.

Oil, gas and miners: the value is a price the company does not set

The sectors in the previous reads all set their own prices in some measure — a bank its spread, a REIT its rents, a utility its regulated return. Commodity producers do not: an oil company sells a barrel at whatever the world price is on the day, so its revenue is the product of a price it cannot influence and volumes that decline unless it spends money to replace them. That single fact rewrites which metrics are useful. An earnings multiple on a commodity producer is close to meaningless on its own, because the earnings figure is a function of the price that happened to prevail that year: the same company earned multiples of the same profit at $110 crude and a fraction of it at $50. The question is never “how cheap is it on earnings”, it is “what does the business earn across the cycle, and what does that require the commodity to do”. The industry’s own metrics exist for that reason. **Reserve life** — proved reserves divided by annual production — describes the asset’s duration before it needs replacing, and a producer with a five-year reserve life is buying inventory continuously while one with twenty years is harvesting. **P/NAV** discounts the reserves at a stated price deck and compares the result with the market value of the equity, which forces the analyst to name the price assumption instead of hiding it in earnings. **EV/EBITDAX** adds exploration expense back, because successful exploration creates the reserves that the depreciation line then consumes. And the number the market watches most is the **breakeven**: the commodity price at which the business funds its capital programme and its dividend out of cash flow. That breakeven is comparable across companies in a way that earnings are not, because it is stated in the units the risk actually arrives in. The second structural fact is the **cost curve**. Commodities are priced at the margin, so the marginal producer earns approximately nothing and sets the price; companies further down the curve earn a margin through the cycle, and companies above it earn a margin only near the top. That is why “low-cost producer” is a genuine moat in this sector in a way that “good management” is not, and why the analysis starts with where the company sits on the curve rather than with its recent results. Miners add a layer: grade and jurisdiction determine the cost position, and a mine’s capital cannot be moved once it is sunk, so a high-cost mine stays in production at a loss long after the price falls. The consequence for an investor is the one F20 makes for banks — the leverage is operational rather than financial, and a commodity producer at the bottom of the cost curve with debt is a business whose survival is a price forecast. The third fact is the one that ties back to the moat lesson: commodity industries are where the **capital cycle** is most visible, because capacity is expensive, slow to build and impossible to relocate. High prices today authorise the projects that will create the surplus in four years, and the surplus is what breaks the price. A producer’s own capital programme is therefore information about the forward curve, and the sector’s collective capex — not its collective earnings — is the better predictor of where the price is going. That is the same mechanism F7 describes for competitive advantage in general, running in the one industry where the lag is long enough that almost everyone can see it and almost nobody acts on it. The sector map, completed — Energy producers: P/NAV of reserves at a named price deck, reserve life, breakeven, position on the cost curve · Miners: P/NAV, grade and jurisdiction, cash cost decile — and sunk capital that stays in production ← · Utilities and pipelines: Regulated asset base per share and the allowed return against the cost of equity · Exchanges and data: Volume-driven revenue with high operating leverage — P/E on a mid-cycle volume base ← The discipline across every one of these sectors is the same sentence: find the number the industry itself manages to, then ask whether the price is assuming something that number cannot see. For a producer the number is breakeven and the invisible assumption is the commodity price; for a utility it is the allowed return and the invisible assumption is the regulator.

Sources

• Why banks are valued on book value and returns on tangible equityStandard bank valuation practice; Koller, Goedhart & Wessels, "Valuation", ch. on financial institutions
• FFO and AFFO as the earnings measures for real-estate investment trustsNAREIT definitions of FFO and AFFO; standard REIT analysis
• Normalized earnings and mid-cycle valuation for cyclical businessesStandard cyclical valuation practice; Damodaran, "Investment Valuation"
• Regulatory capital as a constraint on bank distributionsBasel III framework; Federal Reserve stress-test disclosures