Direct Answer
Valuation multiples are shortcuts for estimating intrinsic value. They work by comparing a price metric (stock price, enterprise value) to a financial metric (earnings, EBITDA, revenue, book value). The problem is that the relationship between price and each financial metric varies dramatically by sector, because different sectors have different growth rates, capital intensity, accounting treatments, leverage norms, and cash flow conversion rates. Using the same multiple across all sectors produces comparisons that are meaningless at best and misleading at worst.
The correct approach is to use the multiple that best captures the underlying economics of each sector: price-to-book for banks and insurers (where asset value is the primary driver of intrinsic value); EV/EBITDA for capital-intensive industrials, utilities, and materials companies (where depreciation and leverage distort earnings); EV/Revenue or EV/ARR for high-growth software (where earnings are intentionally minimal due to growth investment); P/E for mature consumer businesses with stable earnings and modest capex; and risk-adjusted NPV for pharmaceutical pipelines where the value is in future drug revenues not yet earned.
Key Takeaways
- Financials (banks, insurers): Price-to-Book (P/B) and Price-to-Tangible Book (P/TBV). Bank assets and liabilities are marked to market or near-market values; book value is a meaningful anchor. ROE vs. cost of equity determines the appropriate P/B premium or discount.
- Utilities: EV/EBITDA and dividend yield. High depreciation on long-lived assets and regulated capital structures make EBITDA more stable than EPS; dividend yield reflects the income character of utility investment returns.
- Energy: EV/EBITDA and EV/DACF (debt-adjusted cash flow). Oil and gas companies have volatile earnings from commodity prices; EBITDA and cash flow metrics smooth these, and debt-adjusted cash flow removes financing structure differences.
- Materials: EV/EBITDA. Capital-intensive, cyclical businesses where D&A is large and leverage varies; EBITDA removes these distortions for peer comparison.
- Industrials: EV/EBIT and EV/EBITDA. Capital intensity varies widely by sub-industry; EBIT is appropriate when depreciation is real economic cost; EBITDA for highly capital-intensive sub-sectors.
- Consumer Staples and Discretionary: P/E (forward). Mature businesses with stable earnings and modest capex; forward P/E is the standard institutional comparison metric.
- Health Care: P/E, EV/EBITDA for large pharma; EV/Revenue and rNPV for biotech. Pipeline value requires probability-weighted NPV analysis, not multiples of current earnings.
- Technology: EV/Revenue or EV/ARR for growth software; P/E and EV/FCF for mature tech. Loss-making growth companies cannot be valued on earnings; revenue multiples capture scale and market position.
Core Concepts
Financials: Price-to-Book and ROE Framework
Banks and insurance companies hold financial assets and liabilities that are marked to market or near-market values on their balance sheets. This makes book value (shareholders' equity = assets minus liabilities) a meaningful economic anchor in a way that it is not for, say, a manufacturing company whose plant and equipment may be recorded at historical cost far below or above current market value.
The theoretical relationship between P/B and ROE is well-established: P/B = ROE ÷ Cost of Equity (COE), when earnings are in steady state and grow at the cost of equity. A bank earning 15% ROE with a 10% COE should trade at 1.5x book. A bank earning 8% ROE with a 10% COE should trade below book — at 0.8x. This formula is a first approximation, not a precise target, but it provides a principled basis for comparing bank valuations.
Price-to-tangible-book (P/TBV) strips out intangible assets and goodwill from the book value calculation, which is important because goodwill from acquisitions does not generate returns like earning assets do. P/TBV is a more conservative and common institutional metric for comparing bank valuations, particularly after periods of acquisition activity.
Insurance companies use a similar P/B framework but with additional focus on the combined ratio (losses plus expenses as a percentage of premiums) and embedded value (a discounted cash flow approximation of the present value of future insurance profits), which is particularly relevant for life insurance companies whose profits emerge over decades.
Utilities and Real Estate: Yield-Based and Cash Flow Multiples
Utilities are valued primarily on EV/EBITDA and dividend yield. The regulated nature of their earnings means that EBITDA is highly predictable — regulators set allowed returns on the rate base (the total invested capital in the utility's infrastructure) rather than allowing market competition to determine prices. This regulatory framework creates stable, bond-like cash flows that investors primarily value on an income basis.
Dividend yield is particularly relevant for utilities because they are mandated to distribute a high proportion of earnings (low capex relative to earnings for many mature utilities, regulatory incentives to return capital). A utility yielding 4.5% versus 10-year Treasury yields of 4.0% reflects a modest yield premium for the equity risk of the utility versus the guaranteed return of the Treasury. When Treasury yields rise significantly, utility stocks typically fall because their relative yield advantage narrows.
REITs (Real Estate Investment Trusts) in the Real Estate sector are valued on AFFO (Adjusted Funds From Operations) yield, FFO/share, and NAV (Net Asset Value) — a property-by-property valuation of the real estate portfolio. GAAP EPS is distorted by depreciation (real estate often appreciates in value rather than deprecating in economic terms), making FFO-based metrics far more relevant. A REIT trading at a 10% discount to NAV is cheaper than one trading at NAV, all else equal.
Energy and Materials: Commodity-Adjusted Cash Flow Multiples
Energy companies (particularly exploration and production companies) have earnings that are highly volatile due to oil and natural gas price fluctuations. GAAP earnings can swing from large profits in high-price years to large losses in low-price years. To compare companies across commodity price environments, analysts use EV/EBITDA or EV/DACF (debt-adjusted cash flow).
DACF adjusts cash flow from operations for the after-tax cost of interest expense, removing the effect of capital structure differences between highly leveraged and conservatively capitalized companies. When comparing Chevron (conservative leverage) to a smaller E&P with significant debt, DACF allows an apples-to-apples comparison of the fundamental cash-generating ability of the oil and gas assets.
Analysts also commonly use price-to-NAV (P/NAV) for E&P companies, estimating the present value of all proved and probable reserves at a specific commodity price assumption (typically using strip pricing — the futures market's implied future prices) and comparing that estimated value to the enterprise value. A company trading at 0.8x NAV is theoretically cheaper than one at 1.0x NAV, assuming the reserve estimates and price assumptions are comparable.
Technology: Revenue Multiples for Growth, FCF for Maturity
High-growth software companies invest heavily in sales, marketing, and R&D, intentionally producing minimal or negative GAAP earnings. For these companies, P/E or EV/EBITDA multiples are either undefined (negative denominator) or produce misleadingly high numbers that make all growth-stage companies look equally "expensive" regardless of actual differences in growth rate, market position, or path to profitability.
EV/Revenue (enterprise value divided by annual recurring revenue for SaaS companies, or total revenue for broader tech) is the primary metric for growth software. The market pays 8-15x EV/ARR for high-growth (40%+ ARR growth) SaaS with net revenue retention above 120% and improving gross margins. It pays 3-6x ARR for slower-growth (15-25%) SaaS approaching profitability. The multiple reflects the market's assessment of whether the company can convert revenue growth into durable earnings at scale.
Mature technology companies with stable earnings and high free cash flow conversion — Microsoft, Apple, Cisco at various points in their histories — are appropriately valued on P/E (forward) and EV/FCF (free cash flow). These companies are essentially high-quality businesses with durable competitive advantages, and the income statement and cash flow statement give a reliable picture of their earnings power.
Worked Scenario: Choosing the Right Multiple for a Healthcare Comparison
- Identify the sub-industry: You are comparing two companies — Company A is a large-cap, diversified pharmaceutical company with $25B in revenue and several approved drugs. Company B is a clinical-stage biotech with $0 revenue but a Phase 3 oncology drug that could generate $3B in peak sales if approved.
- Select the appropriate multiple for Company A: Use forward P/E and EV/EBITDA. At $25B revenue and assuming 30% EBITDA margin ($7.5B EBITDA), if the enterprise value is $90B, EV/EBITDA = 12x — within the normal historical range for large pharma (10-15x). Forward P/E would also be meaningful given stable earnings.
- Select the appropriate metric for Company B: EV/EBITDA and P/E are meaningless — Company B has no revenue and no earnings. The correct approach is risk-adjusted NPV (rNPV). Estimate peak sales of $3B, apply a 35% operating margin at peak, project the revenue ramp over 5-10 years, discount at an appropriate rate (12-15% for biotech given failure risk), and multiply by the probability of regulatory approval (Phase 3 oncology approval rates are approximately 50-60%). If the rNPV calculation yields $2.5B, compare that to the company's enterprise value to assess relative value.
- Avoid cross-metric comparisons: Do not compare Company A's P/E of 18x to Company B's lack of a P/E and conclude they are "not comparable." They are genuinely not comparable — they are at different stages with different value drivers. The correct comparison within biotech is rNPV/share vs. current price; the correct comparison within large pharma is P/E vs. large pharma peers.
- Check against historical multiple ranges: A large pharma P/E of 18x is modestly above the historical range (12-17x over 10 years). This could reflect higher pipeline value being priced in, or multiple expansion from rate environment changes. Understanding whether the current multiple is high or low relative to its own history is an essential second step after establishing which multiple to use.
Measurement Framework
| Sector | Primary Multiple | Secondary Multiple | Why |
|---|---|---|---|
| Financials (Banks) | P/TBV | P/B, ROE vs COE | Financial assets at market value; book is economic anchor |
| Utilities | EV/EBITDA | Dividend yield | Regulated returns; high D&A; income-investor base |
| Real Estate (REITs) | P/AFFO | P/NAV, FFO yield | GAAP depreciation distorts EPS; FFO captures economic earnings |
| Energy (E&P) | EV/DACF | P/NAV, EV/EBITDA | Commodity price volatility; capital structure varies widely |
| Materials | EV/EBITDA | EV/EBIT | Capital-intensive, cyclical; D&A and leverage vary |
| Industrials | EV/EBIT | EV/EBITDA, P/E | D&A often real economic cost; capital intensity varies by sub-industry |
| Consumer Staples | P/E (forward) | EV/EBITDA | Stable earnings, modest capex; earnings-based valuation is reliable |
| Consumer Discretionary | P/E (forward) | EV/EBITDA, EV/Sales | Mix of stable and cyclical; use trailing P/E cautiously at cycle peaks |
| Health Care (large pharma) | P/E (forward) | EV/EBITDA | Stable marketed product revenues; earnings meaningful |
| Health Care (biotech) | rNPV | EV/Revenue | Pipeline value; no or minimal earnings; clinical-stage companies |
| Tech (growth software) | EV/ARR | EV/Revenue | Intentionally low earnings; revenue scale and NRR are value drivers |
| Tech (mature) | P/E (forward) | EV/FCF | Stable earnings; FCF conversion is high; income-like characteristics |
| Communication Services | Varies by sub-industry | EV/EBITDA (telecom), P/E (internet) | Heterogeneous sector; apply sub-industry appropriate multiple |
Common Failure Modes
Using P/E to compare a technology company to a utility
A utility with a P/E of 15x and a high-growth software company with a P/E of 45x are not "the same" versus "3x as expensive." The utility earns stable regulated returns with minimal growth; the software company may be growing revenue 40% per year and will earn dramatically more in five years than it does today. The P/E of the software company reflects that future earnings power, not the current year's earnings. The P/E of the utility reflects the modest growth rate and income character of its business. Neither is cheap or expensive in comparison to the other; they are different asset classes that should not be compared on a single metric.
Applying EV/EBITDA to banks
EV/EBITDA is not appropriate for financial companies. Banks' interest expense is not a financing cost to be added back — it is the primary cost of goods sold (the cost of borrowing deposits and wholesale funding to lend out at a spread). EBITDA adds back interest expense, which would remove the most important cost driver in a bank's economics. P/B and ROE-based metrics are the correct framework for banks; EBITDA-based metrics actively obscure the business model.
Using trailing P/E at a cycle peak for cyclical sectors
At a cyclical peak, energy companies, materials companies, and some industrials earn very high earnings because commodity prices or order volumes are at cycle highs. The trailing P/E at a peak looks deceptively low — Energy might trade at 8x trailing EPS because earnings have surged — but the correct forward-looking question is what earnings look like at a normalized commodity price. Normalized P/E (using through-the-cycle average earnings rather than peak earnings) avoids the value trap of buying "cheap" cyclical stocks at the top of their earnings cycle.
Ignoring capital structure when comparing EV multiples across companies
EV/EBITDA correctly includes debt in the enterprise value, which is why it is better than P/E for cross-company comparisons. However, some analysts make a subtler error: they compare EV/EBITDA across companies without noting that EBITDA itself can vary based on whether the company uses operating leases (capitalized on the balance sheet under IFRS 16/ASC 842) versus owned assets. A company that leases all its facilities has lower EBITDA (lease costs flow through the income statement) under one accounting treatment than under another. Adjust for lease capitalization when comparing across companies with different asset ownership structures.
Treating any multiple as definitive without checking the range
Every multiple is only meaningful in context — compared to the company's own history, compared to industry peers, and compared to the relevant benchmark's current multiple. An industrial company at 12x EV/EBITDA is not expensive or cheap in isolation; it is potentially expensive if its historical range has been 7-10x, or cheap if it has historically traded at 14-16x. Always situate a current multiple in its historical and peer-comparison context before drawing a valuation conclusion.
FAQ
Why can't you compare P/E ratios across different sectors?
P/E reflects growth expectations, capital structure, accounting treatments, and business model characteristics that vary systematically across sectors. A utility's P/E of 15x reflects modest regulated growth; a software company's P/E of 40x reflects future earnings power from compounding growth. These are not comparable multiples — they measure different things about different business economics. Cross-sector P/E comparisons produce misleading conclusions about relative cheapness or expensiveness.
What is the best valuation multiple for banks?
Banks are most commonly valued on price-to-book (P/B) and price-to-tangible-book (P/TBV). Because banks hold financial assets and liabilities close to market value, book value represents a meaningful economic anchor. The ROE vs. cost-of-equity framework determines whether a bank deserves a premium or discount to book: P/B ≈ ROE ÷ COE in steady state. EV/EBITDA is inappropriate for banks because interest expense is a core business cost, not a financing add-back.
What multiple is used for utilities?
Utilities are primarily valued on EV/EBITDA and dividend yield. Regulated earnings structures make EBITDA highly predictable. Utilities also pay large dividends, so dividend yield is a primary return metric for income-oriented investors. When comparing utilities to bonds, the yield spread (utility yield minus Treasury yield) is a useful relative-value indicator — it reflects the equity risk premium investors demand to own the utility versus a risk-free instrument.
Why do high-growth tech companies trade on EV/Revenue instead of P/E?
High-growth software companies intentionally produce near-zero or negative GAAP earnings by reinvesting all gross profit into sales, marketing, and R&D to maximize revenue growth. P/E is undefined when earnings are negative. EV/Revenue values the company on its revenue base, which grows steadily even during the investment phase, reflecting the market's belief that the company will achieve high operating margins once it reaches scale and slows growth investment.
What is a normal EV/EBITDA range for industrial companies?
Industrials have historically traded at EV/EBITDA of 8-14x over full cycles. Defense/aerospace commands higher multiples (12-16x) due to long-term government contract visibility. Cyclical machinery and equipment manufacturers trade at the lower end (7-10x) given earnings volatility. Specialty industrials with recurring revenue streams (industrial automation, testing equipment) often trade at the higher end. Always compare to the specific sub-industry's historical range rather than a single sector-wide number.
How should you value a pharmaceutical company?
Large pharma with marketed products: forward P/E and EV/EBITDA, compared against large-cap pharma peers. For companies with significant pipelines, add a sum-of-the-parts analysis that separately values each pipeline asset using risk-adjusted NPV (rNPV) — probability-weighted expected cash flows discounted at a biotech-appropriate rate. Clinical-stage biotech with no revenue: rNPV is the primary valuation method; EV/Revenue only applies once commercial revenue begins.
What is EV/EBITDA and why is it better than P/E for capital-intensive businesses?
EV/EBITDA divides enterprise value (market cap plus net debt minus cash) by EBITDA. It is better than P/E for capital-intensive businesses because it removes depreciation (which varies by asset age and mix, creating incomparability between companies with different asset bases) and removes interest expense (allowing comparison regardless of leverage choice). Two identical industrial companies with different debt levels and asset ages will show very different P/Es but similar EV/EBITDAs.
Where can I find historical sector P/E and EV/EBITDA data?
Damodaran Online (pages.stern.nyu.edu/~adamodar/) publishes free annual industry-level valuation multiples including P/E, EV/EBITDA, P/S, and P/B with multi-year history. Yardeni Research publishes regularly updated S&P 500 sector P/E data. The SPDR sector ETF site shows weighted-average P/E and P/B for each sector ETF. Bloomberg and FactSet are the professional-grade sources with real-time and deeper historical data.
Sources
Disclaimer
This article is for educational and informational purposes only and does not constitute personalized investment, financial, or legal advice. Valuation multiples and their appropriate ranges change as market conditions, interest rates, and industry structures evolve. Always verify current valuation data from primary sources. Trading involves risk, including the possible loss of principal.