Why Sector Determines How You Should Value a Stock

Two stocks trade at 12 times earnings. One is an automaker coming off the best year in its history; the other is a software company growing revenue 20% a year with 85% gross margins. Same multiple, opposite verdicts: the automaker is probably expensive, because those record earnings will shrink when the cycle turns, while the software firm is probably a bargain, because its earnings are durable, capital-light, and still compounding. A single number told you nothing until you knew what kind of business stood behind it. That is the central fact of valuation that whole-market stock screens quietly ignore — and the reason this article exists.

This piece opens our series on sector-specific valuation. It maps which metric matters in each major sector and why, building on the general framework in our guide to how to value a stock and our explainer on the P/E ratio. The deep dives that follow cover the three sectors where standard tools fail hardest: banks, REITs, and cyclicals.

The Problem: One Yardstick, Many Kinds of Business

Every valuation multiple is a compressed model of a business. A P/E ratio assumes that one dollar of this year's earnings is a reasonable proxy for the stream of future earnings you are actually buying. That assumption holds within a group of similar companies and collapses across dissimilar ones. A dollar earned by a subscription software firm will likely recur next year, needs almost no new capital to produce, and may grow 15%. A dollar earned by an airline at the top of a travel boom may vanish entirely in the next recession. Paying the same multiple for both is not neutral — it systematically overpays for fragile earnings and underpays for durable ones.

This is why the market's sector-level multiples differ so persistently. Software, medical devices, and luxury brands have traded at premium multiples for decades; autos, airlines, and steel have traded at single-digit P/Es for just as long. Those gaps are not inefficiency waiting to be arbitraged. They are the market correctly pricing differences in earnings quality. The investor who screens the entire market for the lowest P/Es does not find bargains — he finds a portfolio of automakers, banks at cycle peaks, and shrinking legacy businesses, and wonders why cheap keeps getting cheaper.

Why Sectors Differ: Five Structural Forces

Five forces explain nearly all of the variation in what a fair multiple looks like. First, capital intensity: a railroad must pour billions into track and locomotives to grow, while a software firm grows on marginal server costs, so a dollar of railroad earnings converts to far less free cash flow than a dollar of software earnings. Second, leverage: for most companies debt is a financing choice, but for banks and insurers leverage is the raw material of the business itself — a bank borrows deposits at one rate and lends at a higher one, so metrics that try to strip out debt destroy the very thing you are measuring. Third, accounting distortions: depreciation rules force real estate companies to book large non-cash charges against buildings that often appreciate, making reported earnings nearly meaningless. Fourth, cyclicality: commodity producers, semiconductor makers, and automakers see earnings swing 50% or more through a cycle, so any single year's E is a snapshot of a moving target. Fifth, growth durability: a multiple is an implicit forecast of how long above-average returns last, and that horizon differs enormously between a niche software monopoly and a regional airline.

Banks and Insurers: Price-to-Book and ROE

A bank's balance sheet is its business. Its assets are loans and securities carried at values close to reality, its liabilities are deposits, and its earnings are the spread between them, multiplied by leverage that routinely runs ten to one. That makes price-to-book the natural primary metric, paired with return on equity as the quality gauge. The logic is tight: a bank that earns a 15% ROE deserves to trade above book value, because each dollar of equity is producing more than investors could get elsewhere at similar risk; a bank earning 6% deserves a discount to book. High-quality franchises like JPMorgan have historically commanded persistent premiums to book for exactly this reason, while weaker lenders languish below it. Earnings-based multiples still have a supporting role, but P/E alone misleads because loan-loss provisions make bank earnings lurch with the credit cycle. The full mechanics — justified P/B, tangible book value, and the credit-cycle traps — are in our deep dive on how to value bank stocks.

REITs: FFO and AFFO, Not Earnings Per Share

Real estate investment trusts are the clearest case of accounting rules breaking a metric. GAAP requires a REIT to depreciate its buildings over decades, producing a large non-cash expense every year — even though well-maintained commercial property often holds or gains value. The result is that a REIT's reported net income drastically understates the cash its properties actually generate, and its P/E ratio looks alarmingly high to anyone screening on earnings. The industry's answer is funds from operations, which adds depreciation back and strips out one-time property-sale gains.

Serious REIT analysis goes one step further to adjusted FFO, which also subtracts the recurring capital spending needed to keep properties competitive, giving the truest picture of distributable cash. Dividend durability, cap rates, and net asset value all hang off these numbers — a blue-chip landlord like Realty Income is analyzed almost entirely in FFO and AFFO terms. Our guide to how to value REIT stocks works through the full toolkit.

Cyclicals: Normalized Earnings and the Peak-P/E Trap

Autos, semiconductors, airlines, chemicals, miners, and homebuilders share a defining feature: their earnings oscillate violently around a long-run average. That creates the most expensive optical illusion in investing. At the top of the cycle, earnings are inflated and the P/E looks tantalizingly low; at the bottom, earnings collapse and the P/E looks terrifyingly high or turns negative. The naive read is exactly backwards — cyclicals often deserve to be bought at high P/Es on depressed earnings and sold at low P/Es on peak earnings.

The fix is to value the cycle, not the year: average earnings or margins across a full cycle, apply a mid-cycle multiple, and check the balance sheet for the strength to survive the trough. Book value also regains usefulness here as a floor for asset-heavy cyclicals. We cover normalization techniques, where semis differ from commodities, and when to buy in our deep dive on how to value cyclical stocks.

Software and Asset-Light Compounders: Free Cash Flow and Growth

At the opposite pole sit software, internet platforms, and other asset-light businesses. Their economics — high gross margins, negligible capital spending, recurring revenue — mean reported earnings often understate cash generation early on (heavy growth investment runs through the income statement) and track it closely at maturity. The primary lens is free cash flow: the FCF yield tells you what cash return you are buying today, and EV/EBITDA offers a capital-structure-neutral cross-check. Because growth does most of the valuation work in this sector, raw multiples must be growth-adjusted — a PEG-style comparison or, better, a full DCF model that makes the growth assumptions explicit. The danger runs opposite to cyclicals: here the trap is not overpaying for peak earnings but overpaying for growth durability that competition eventually erodes.

Utilities, Staples, and Energy: Yield, Stability, and the Cycle

Regulated utilities and consumer staples are the closest thing equities offer to bonds. Demand is steady, growth is low but reliable, and a large share of earnings is paid out as dividends. The right tools are therefore income-based: dividend yield versus history and versus interest rates, dividend growth rates, and stable-assumption DCF or dividend discount models that low earnings volatility makes unusually trustworthy. For utilities, the regulated allowed return on equity caps the upside and anchors the valuation. The main risk is rate sensitivity: when bond yields rise, these bond-like stocks reprice downward.

Energy producers look superficially similar to utilities — big, dividend-paying, capital-intensive — but they are really cyclicals wearing a different uniform, with one addition: a depleting asset base. An oil producer must replace every barrel it pumps, so valuation rests on two questions. What are the reserves worth — measured by reserve life, replacement cost, and net asset value per share? And what cash does the business generate through the commodity cycle, not at today's spot price? A producer that gushes free cash flow at $90 oil and burns it at $55 must be valued on mid-cycle price assumptions, exactly like the normalized-earnings approach for other cyclicals. Single-year P/Es on energy stocks are close to noise.

The Golden Rule — and the Mistakes It Prevents

Everything above compresses into one rule: compare a company against its sector peers, never against the whole market. Relative valuation only works when the comparison set shares the same economics, because only then does a discount to peers signal mispricing rather than a structural difference. This is why the relative valuation component of the Fair Price Index valuation model — 30% of the blended fair value, alongside 50% DCF and 20% analyst consensus — benchmarks every company against its own sector's multiples, not a market-wide average. A bank is judged against banks on P/B; a software firm against software on cash flow multiples.

Three cross-sector mistakes account for most self-inflicted damage. First, screening the whole market with a single P/E cutoff, which mechanically loads the results with peak-cycle cyclicals and cheap-for-a-reason financials while excluding every quality compounder. Second, comparing a REIT's payout ratio to a tech company's: a REIT is legally required to distribute at least 90% of taxable income, so its high payout is a structural feature, not a red flag — and measured against FFO rather than EPS, it is usually far more comfortable than it looks. Third, applying EV/EBITDA to banks, which is not merely imprecise but meaningless: a bank's interest expense is its cost of goods sold, its debt is inseparable from its operations, so 'enterprise value' and 'earnings before interest' are undefined concepts for a business whose entire product is interest.

How to Apply Sector-Specific Valuation

A practical workflow: before quoting any multiple, name the sector and its primary metric — P/B and ROE for banks and insurers, price-to-FFO for REITs, normalized earnings for cyclicals, FCF yield and growth-adjusted multiples for software, dividend-anchored models for utilities and staples, mid-cycle cash flow and reserve value for energy. Build your comparison set strictly from sector peers of similar size and quality. Ask whether current earnings are peak, trough, or normal before trusting any single-year ratio. And treat a discount to peers as the start of the investigation, not the conclusion — sometimes the cheapest bank in the group is cheap because its loan book is about to blow up.

This sector logic is built into every number on Fair Price Index. Our fair value estimates for more than 37,000 stocks blend a DCF, sector-relative multiples, and analyst consensus, so a bank is never scored like a software company and a REIT's depreciation never poisons its valuation. Continue with the deep dives on banks, REITs, and cyclicals linked above, or browse fair values across every sector at our stock screener to see the same company-versus-peers comparison done for you.