EBITDA Is Perfectly Imperfect

EBITDA (earnings before interest, taxes, depreciation, and amortization) is not a normal financial metric. It isn’t a GAAP figure, so you won’t find it on the financial statements. To calculate it, you mix items from the income statement and the cash flow statement, so it’s neither a pure measure of profitability nor a measure of cash flow. Charlie Munger, one of the greatest investors ever, called it “bullshit earnings.”

But EBITDA is used too widely to be discarded as a useless tool. It’s imperfect, but so is every other financial metric. EBITDA is quick. It’s easy to understand. It allows comparison between otherwise hard-to-compare peers. And any analysis that requires depth will not look at EBITDA in isolation, mitigating its flaws.

Calculating EBITDA

There are two ways to calculate EBITDA:

• Top-down: EBITDA = EBIT + D&A. EBIT comes from the income statement. D&A comes from the cash flow statement (not the income statement) because on the income statement, D&A is often split up and buried inside COGS and operating expenses rather than shown on its own line.
• Bottom-up: EBITDA = Net Income + Taxes + Interest + D&A. Conceptually, you’re undoing everything that separates net income from EBITDA.

A variant of the bottom-up build shows up in interviews: instead of adding taxes back, you divide net income by (1 − Tax Rate), which returns you to pretax income; then add interest to get EBIT, and D&A to get EBITDA.

Here’s a simplified income statement for an imaginary cloud-computing company:

Total D&A is $15 ($10 in COGS, $5 in SG&A and R&D). Top-down: EBIT of $30 + D&A of $15 = EBITDA of $45. Bottom-up: net income of $20 + taxes of $5 + interest of $5 + D&A of $15 = $45. Note that you could not have computed this from the income statement alone — nothing on it tells you the $15 of D&A hiding inside COGS and operating expenses. That’s why the cash flow statement supplies D&A.

Why Everyone Uses It

The main benefits of EBITDA are that it is not affected by differences in D&A accounting, capital structure, or tax rates. You cannot say the same about EBIT, pretax income, or net income. Even gross profit is not agnostic to D&A accounting: some management teams include depreciation in COGS and some don’t — exactly as in the toy example above.

It’s easy to argue that any one of the items EBITDA ignores is important. That’s true. But that’s the tradeoff we make with EBITDA to get its main benefit: comparability of financials across companies and across time.

Suppose you’re screening companies and all you have in front of you is net income. One US company’s net income jumps in 2018 and stays at the new level. From net income alone, you can’t tell whether that was a big customer win, a margin improvement, lower interest expense from paying down debt, or the 2017 corporate tax cut (35% top rate to 21%), which hit every US company’s 2018 results. Let’s say the last option was the true reason. In that case, EBITDA would have shown you immediately that nothing about the business changed. In general, EBITDA reduces the effort we have to spend making things comparable.

EBITDA in Valuation

EV/EBITDA is the most commonly used valuation multiple in practice, and the pairing is intentional. EBITDA is earnings before any capital provider gets paid (before creditors receive interest and before shareholders receive anything) so it “belongs” to both groups. Enterprise value is the value of the whole business, also attributable to both groups. Numerator and denominator match.

That matching rule is why Equity Value / EBITDA is wrong: equity value belongs only to shareholders, while EBITDA belongs to all capital providers — apples to oranges. If this consistency logic feels familiar, it’s the same rule that pairs UFCF with WACC and LFCF with the cost of equity in a DCF (see A Guide to the DCF).

The same convenience makes EBITDA the standard yardstick in credit. Leverage is quoted as Debt/EBITDA or Net Debt/EBITDA, coverage as EBITDA/Interest, and debt capacity as “turns” of EBITDA. When a lender asks how much debt a business can support, the answer almost always comes back as a multiple of EBITDA.

A Proxy for Cash Flow

Beyond comparison, EBITDA is used as a rough proxy for cash flow, which is what makes it load-bearing in both valuation and leverage analysis. This is despite EBITDA ignoring three genuine uses of cash: capital expenditures, changes in net working capital, and taxes. A company with aging factories that have been neglected for a decade could finally be embarking on a three-year modernization plan, with capex reaching 3–5x normal levels over that stretch. EBITDA would completely ignore that massive cash drain.

EBITDA − Capex can sound like the fix: if EBITDA fails as a cash flow proxy because of capex, just subtract capex. Sometimes this works well. Lenders even use (EBITDA − Capex)/Interest as a covenant metric. But capital expenditures are smoothed into depreciation for a reason. Capex is lumpy: companies often invest heavily in some years and barely at all in others. For many businesses, EBITDA − Capex swings so much year to year that using it for valuation or leverage is unhelpful. Both maintenance capex (e.g., aging factories example) and growth capex (e.g., hyperscalers investing in AI) can swing aggressively.

When to Be Skeptical

EBITDA is most misleading for capital-intensive businesses, where the gap between EBITDA and EBIT is enormous. The clearest example right now is the hyperscalers. In 2026, Amazon, Microsoft, Alphabet, and Meta collectively guided to roughly $700 billion of capital expenditures for AI data centers, with Amazon alone having guided to about $200 billion. For companies spending on that scale, a metric that ignores capex is ignoring the single largest use of cash. In capex-heavy settings, EV/EBIT (which at least charges the company for depreciation) or free-cash-flow-based measures do a better job.

A classic interview question shows how D&A accounting can distort EBITDA comparisons: two identical vending-machine businesses, one owning its machines and one leasing them, with the depreciation and lease payments equal in dollars. The owner’s cost shows up as depreciation (below EBITDA) while the lessee’s rent historically sat in operating expenses (inside EBITDA). These companies have the same economics but different EBITDAs.

EBITDA is also not helpful to analyze financial institutions, where interest is the core business rather than a financing choice, and unprofitable companies with near-zero or negative EBITDA, where the multiple math breaks and valuation should fall back to revenue multiples.

Adjusted EBITDA

Because EBITDA isn’t defined by GAAP, every company gets to define it for itself, and most go a step further and report “Adjusted EBITDA,” adding back stock-based compensation, restructuring charges, transaction costs, and other items management deems non-recurring or non-cash. Each add-back can be defensible on its own. Together, they can drift a long way from economic reality. Never take an adjusted figure at face value. Read the reconciliation from net income and judge each add-back yourself.

Takeaway

EBITDA wears a lot of hats. It’s so widely used because of its convenience as a tool for comparison, as the standard valuation and leverage yardstick, and as a proxy for cash flow. But the sources of its usefulness (ignoring capex and D&A, capital structure, and taxes) are also its limitations. It’s a tool to be used with caution.

A good next step is to look at how real companies define it. We recommend Roper Technologies and TransDigm: in their earnings press releases, these companies publish bridges from net income to reported EBITDA to adjusted EBITDA. You should walk through each add-back and decide for yourself whether it could make sense.