Valuation multiples and when each fails

In this article

1. Before you start
2. Why interviewers ask this
3. Valuation multiples from first principles
4. Worked example
5. Trade-offs and edge cases
6. How to answer it in an interview
7. Common mistakes
8. Follow-up questions
9. Sources

Before you start

This page is one step of the investment banking interview questions learning path.

Why interviewers ask this

A wrong multiple produces a confident wrong price, and a multiples question tests whether you know that. Candidates report questions of this kind in investment banking and finance technical rounds, usually as "which multiple would you use for this company, and why?" The investment banking technicals cover the neighboring formulas.

A weak answer recites formulas. A strong answer says what the denominator leaves out, matches the numerator to it, and picks the multiple that fits the business. It also says what a multiple cannot tell you. Practice is concentrated: in a survey of nearly 2,000 Chartered Financial Analyst (CFA) Institute members, 88 percent of analysts said they used P/E, and 77 percent used a multiple based on enterprise value (EV), most often EV/EBITDA. Both fail in known ways, which is what this page maps.

Valuation multiples from first principles

Valuation multiple

A valuation multiple is a ratio that divides a price, either a share price or the value of the whole business, by a standardizing metric such as earnings, cash flow, book value or revenue, so that companies of different sizes can be compared on one scale.

EBITDA

EBITDA is earnings before interest, taxes, depreciation and amortization: a profit figure measured before financing costs, taxes and the non-cash charges that spread the cost of long-lived assets over time. It is a rough proxy for operating cash flow, and companies report it as a measure not defined by generally accepted accounting principles (GAAP).

In plain words

Think of pricing a house. You take a neighbor's sale price, divide it by the floor area, and multiply the result by the area of your own house. A multiple does the same for companies: divide a peer's price by its earnings, then apply that ratio to your company's earnings. The analogy breaks in one place. Floor area is measured the same way for every house. Earnings are not. Debt, tax and accounting choices all change them.

In one sentence: relative valuation is valuing a business by how similar businesses are priced, and a multiple is the tool that turns those prices into a ratio.

Step 1: A multiple is a price divided by a standardizing metric

A multiple is a price divided by a standardizing metric. Price per share divided by earnings per share is P/E; enterprise value divided by EBITDA is EV/EBITDA; price divided by book equity or by revenue gives the other common forms. The metric exists to remove size, so a large and a small company can sit on one scale.

Mauboussin and Callahan put it this way: "A multiple in any form is a shorthand for the process of valuation." The shorthand hides the inputs that a full valuation would show. Earnings can be trailing, meaning the past year, or forward, meaning an estimate for the coming year.

Input: a price and one metric. Output: one ratio. You now know what a multiple is; the next step covers the rule that stops the ratio from being nonsense.

Step 2: Match the numerator to the denominator

Enterprise value pairs with metrics before interest; equity value pairs with metrics after interest. P/E is an equity multiple, because the price and earnings per share both belong to shareholders after lenders are paid. EV/EBITDA is an enterprise multiple, because EV covers every investor and EBITDA does not subtract financing costs or taxes.

The same logic covers cash. If you subtract cash from the numerator, the income that cash earns must be left out of the denominator. A numerator that includes a claim (debt, preferred stock, minority interest) while the denominator ignores that claim's income is a mismatch.

A clean bridge, with illustrative inputs: equity of 500, plus debt of 150, minus cash of 50, gives an EV of 600, and with EBITDA of 100 the multiple is 6.0 times.¹ If cash exceeds market capitalization plus debt, EV turns negative (illustrative: equity 100, debt 20, cash 150 gives −30), and an EV multiple is then not meaningful.¹

Step 3: Why P/E and EV/EBITDA give different answers

P/E and EV/EBITDA diverge because of depreciation, debt and taxes. EBITDA sits above net income, since it ignores those three, and the P/E is almost always higher than the EV/EBITDA for a profitable company.

The Walmart and Apple pair shows the size of the gap. On next-twelve-month (NTM) estimates in early March 2024, Walmart traded at a P/E of 25.5 and EV/EBITDA of 13.3, Apple at 25.4 and 20.1. Dividing P/E by EV/EBITDA gives 1.9 times for Walmart and 1.3 times for Apple.¹ The source attributes the difference to Walmart's lower profitability, its net debt against Apple's net cash, and a tax rate about 10 points higher.

Input: one business with three changes. Output: P/E moves, EV/EBITDA does not. The worked example below isolates each cause.

Step 4: A multiple compresses a discounted cash flow into one number

A multiple compresses growth, return on invested capital (ROIC) and risk into one number. Mauboussin and Callahan note that "Multiples provide no direct insight into the magnitude of a firm's investments or whether they will generate a sufficient return." A multiple is only as good as the cash-flow story hidden inside it.

The Gordon growth formula is a perpetuity formula: value equals next cash flow divided by the discount rate minus the growth rate. Use the inputs of the source's appendix: sales of 500, earnings before interest and taxes (EBIT) of 100, depreciation of 25 (so EBITDA of 125), a tax rate of 15 percent, capital spending of 31.25, no change in working capital, a weighted average cost of capital (WACC) of 8.5 percent and growth of 4 percent. Free cash flow is 100 × (1 − 15 percent) − (31.25 − 25) = 78.75, and EV is 78.75 divided by 4.5 percent, which is 1,750.¹

That implies EV/EBITDA of 14.0, P/E of 20.6 with no debt, and EV/revenue of 3.5.¹ Change growth or WACC and every multiple moves.

Step 5: Pick the multiple that fits the business, then read it against the median

The right multiple depends on the sector and is read against the median. Damodaran's table of conventions pairs each sector with the multiple that survives its accounting. Young loss-making growth firms use revenue multiples, because earnings are negative. Infrastructure uses EV/EBITDA because of early losses and big depreciation. Real estate investment trusts (REITs) use price to cash flow from earnings (CFE), meaning net income plus depreciation, because real estate carries big depreciation charges. Banks use price to book equity, with the open question of whether assets are marked to market. Growth firms use the price/earnings to growth (PEG) ratio, and cyclical manufacturers often use P/E on normalized earnings.

Then compare against peers. Multiples have skewed distributions, so averages are seldom good indicators of typical multiples. Use the median, or the multiple with the highest R-squared against fundamentals in that sector. Deal-based peers for the mergers side of the desk are covered under precedent transactions in mergers and acquisitions questions.

1 · Price over a metric2 · Match the claims3 · Ratios diverge4 · Hidden DCF5 · Fit the sectorReplay

Step-by-step description

1. A multiple divides a price by one metric.
2. Enterprise value pairs with pre-interest metrics.
3. Depreciation, debt and tax move P/E only.
4. Growth, ROIC and risk sit inside the ratio.
5. Choose by sector and compare to the median.

• A multiple is a price divided by a standardizing metric.
• Enterprise value pairs with metrics before interest; equity value pairs with metrics after interest.
• P/E and EV/EBITDA diverge because of depreciation, debt and taxes.
• A multiple compresses growth, return on invested capital and risk into one number.
• The right multiple depends on the sector and is read against the median.

Text description of the diagram

• Young loss-making growth firms use revenue multiples.
• Infrastructure uses EV/EBITDA because of early losses and big depreciation.
• REITs use price to net income plus depreciation.
• Banks use price to book equity.

Worked example

The table rebuilds the logic of Exhibit 4 in Mauboussin and Callahan (2024). Sales are 1,000, cost of goods sold is 800 and selling and administrative cost before depreciation is 100, so EBITDA is 100 in every case, and EV is 1,500 in every case.¹ Case A is the starting point, case B adds depreciation to it, and cases C, D and E all start from B and add debt, tax, or both. For a model-building view of the same mechanics, see the financial modeling questions.

You can redo any row with a calculator. In case B, net income is 100 − 40 = 60, so P/E is 1,500 / 60 = 25.0.¹ In case C, debt of 300 leaves equity of 1,200, interest is 18, net income is 42 and P/E is 1,200 / 42 = 28.6.¹ In case E, tax is 7 on profit of 42, so net income is 35 and P/E is 34.3.¹ P/E rises from 15.0 to 34.3 across the five cases while EV/EBITDA stays at 15.0. Depreciation, debt and taxes change net income but not EBITDA.

Text description of the diagram

• P/E rises from 15.0 to 34.3 across the five cases while EV/EBITDA stays at 15.0.
• Depreciation, debt and taxes change net income but not EBITDA.

The same effect appears in real accounts. For fiscal 2023, Microsoft reported net income of 72.4 billion US dollars; after the source's adjustment for capitalizing intangible spending, the trailing P/E went from 34.9 as reported to 30.5 as adjusted, and EV/EBITDA went from 24.2 to 16.9. Adjusted net income was 83.0 billion US dollars and adjusted EBITDA 147.0 billion, against 102.4 billion reported.

Trade-offs and edge cases

No multiple works everywhere. The table says where each one stops working; the sections after it explain the main failures.

When does P/E fail?

P/E fails when earnings are negative, because a negative ratio carries no meaning. The source warns that the usual link between the two ratios does not hold for a company with negative net income. Illustrative case: equity of 800, debt of 250 and cash of 50 give an EV of 1,000. With EBITDA of 60 the EV/EBITDA is 16.7, but net income of −20 gives a P/E of −40.0.¹ An analyst then moves to EV/EBITDA if EBITDA is positive, to revenue, or to forward multiples. If EBITDA is also negative, EV/EBITDA fails too. P/E also misleads for cyclical firms at the peak or trough of a cycle, which is why Damodaran suggests normalized earnings there.

When does EV/EBITDA fail?

EV/EBITDA fails when firms need very different amounts of capital spending. Depreciation and amortization is a proxy for the maintenance capital expenditures a company requires to sustain its business, and EBITDA ignores them. Take two firms with EBIT of 125, a 20 percent tax rate and a WACC of 7.6 percent. Net operating profit after tax (NOPAT) is 100 and, assuming zero growth and capital spending equal to depreciation, both are worth 100 / 7.6 percent = 1,315.8.¹

If EBITDA is 1.2 times EBIT (150) the multiple is 8.8; if it is 1.6 times (200) the multiple is 6.6.¹ Same value, different multiples. The fix is to charge for the asset wear: use EV/EBIT, or EV divided by EBITDA minus capital spending. In this case EV/EBIT is 1,315.8 / 125 = 10.5 for both firms, so the gap disappears.¹ Median EBIT as a share of EBITDA in 2023 was 70 percent across the universe, 56 percent for utilities and 75 percent for consumer discretionary.

Why do EBITDA numbers differ between companies?

EBITDA is not one agreed figure, because companies define it themselves. Under Regulation G, a company that publishes a non-GAAP measure must give the most directly comparable GAAP measure and a reconciliation, as the Securities and Exchange Commission (SEC) manual explains. A real filing shows the problem: Dycom's exhibit says that its non-GAAP measures may not be comparable with similarly titled measures at other companies. It also defines Adjusted EBITDA as excluding gains on sales of fixed assets, stock-based compensation and other adjusting items. Expensed intangible spending distorts both multiples, as the Microsoft figures above show. Always check which definition the peer set uses.

How can the choice of multiple or peer group be biased?

The choice of multiple can steer the answer. One slide in Damodaran's course makes the warning explicit: "pick the multiple that yields the highest value" if you want a company to look undervalued. Mauboussin and Callahan add that analysts often select a peer group that leads to the desired answer. Averages add a second problem. For an illustrative peer set of 8, 9, 10, 11, 12, 14 and 60 times, the mean is 17.7 but the median is 11.0.¹ One outlier moves the mean and barely touches the median. State the peer criteria and show the median.

How to answer it in an interview

The spoken structure: define the multiple, name the matching rule, give one failure for P/E and one for EV/EBITDA, then say how you check the peer set. For the full valuation, use the DCF walkthrough and, for the discount rate, the cost of capital in corporate finance questions.

Common mistakes

• "A lower multiple means it is cheaper." (wrong) Growth, ROIC, risk and the depreciation factor all differ. (right) "A lower multiple is cheaper only if those drivers are similar."
• "EBITDA is cash flow." (wrong) It is a non-GAAP proxy that ignores capital spending. (right) "EBITDA approximates operating cash flow before capital spending, tax and working capital."
• "Negative earnings make multiples useless." (wrong) Earnings-based multiples are unusable; others may not be. (right) "I switch to EV/EBITDA if EBITDA is positive, otherwise revenue or forward multiples."
• "I take the average of the comparables." (wrong) Multiples are skewed. (right) "I use the median and show the peer criteria."

Follow-up questions

What is the difference between trailing and forward P/E?

Trailing P/E divides the price by the earnings of the past year, and forward P/E divides it by an estimate of the coming year's earnings. Forward P/E helps when earnings are changing fast, but it depends on the quality of the estimate.

Why use EV/EBITDA instead of P/E?

EV/EBITDA sits before interest and tax, so companies with different debt levels and tax rates become comparable. It still ignores capital spending, so it suits peers with similar investment needs.

Which multiple do you use for a company with negative earnings?

Use EV/EBITDA if EBITDA is positive, an EV/revenue multiple for young loss-making firms, or a forward multiple if earnings turn positive. Say which one you picked and why.

Why can two companies with the same value trade at different EV/EBITDA?

EBITDA ignores depreciation, which proxies maintenance capital spending, so firms with equal value can have different EBITDA. In the worked case, the same value gave multiples of 8.8 and 6.6.¹ The DCF walkthrough shows where the value comes from.

Should you use the mean or the median of comparable companies?

Use the median. Multiples have skewed distributions, so one outlier pulls the mean away from the typical company.

Is EBITDA the same as cash flow?

No. EBITDA is a non-GAAP measure that leaves out capital spending, tax and working capital, and companies define adjusted versions in their own ways.

Sources

1. Valuation Multiples: What They Miss, Why They Differ, and the Link to Fundamentals — Morgan Stanley Investment Management, Counterpoint Global (Mauboussin and Callahan), 23 Apr 2024
2. Choosing Between the Multiples (session 21 slides, spring 2017 course) — NYU Stern (Damodaran), 1 Jan 2017
3. Financial Reporting Manual, Topic 8: Non-GAAP Measures — U.S. Securities and Exchange Commission, Division of Corporation Finance, 31 Dec 2011
4. Dycom Industries Form 8-K, Exhibit 99.1 (EBITDA and Adjusted EBITDA disclosure) — SEC EDGAR, 11 Sep 2012