EBITDA: What It Shows, What It Hides

What EBITDA is and why it exists

EBITDA stands for Earnings Before Interest, Taxes, Depreciation, and Amortization. Strip away the acronym and the question becomes: why would anyone want to measure earnings before those four things? The answer has genuine logic to it, even if companies later abused it.

Start with interest. Whether a company paid 8% on $500 million in debt or raised the same capital by selling equity depends on how the founders or private equity owners decided to finance the business — not on how well the underlying operations run. Two identical restaurant chains, one carrying heavy acquisition debt and one debt-free, will show very different interest expense. Strip interest out and you can compare the actual restaurant economics.

Taxes have a similar logic. A company operating primarily in Ireland pays a different effective tax rate than one headquartered in the United States. If you're comparing two industrial businesses competing in the same market, tax jurisdictions are noise. Stripping taxes out lets you get closer to a like-for-like operating comparison.

Depreciation and amortization are where it gets more complicated. D&A is a non-cash accounting charge that spreads the cost of an asset over its useful life. A factory bought for $100 million might be depreciated over 20 years, showing $5 million in annual D&A on the income statement even though no cash leaves the door in year 10. For a purely accounting reason, two companies with identical real-world operations can show very different net income depending on when they bought their fixed assets and how aggressively they depreciate them. Add D&A back and you've removed that distortion.

Here's a concrete example. Two retailers operate identical store formats with identical economics. Retailer A bought its store locations in 2005 and carries high depreciation. Retailer B signed long-term leases on the same footprint and carries low depreciation but higher rent expense. Their net income looks different on paper. Their EBITDA looks nearly identical. That convergence is exactly what EBITDA is designed to produce. The problem is what happens next.

The legitimate use case

EBITDA earns its keep in three specific situations. The first is comparing businesses with different capital structures — which happens constantly in private equity, leveraged buyouts, and M&A. When a private equity firm buys a company, it plans to restructure the debt. The target's current interest expense is irrelevant to what the buyer will pay; what matters is the operating cash generation before financing decisions. EV/EBITDA became the standard M&A metric precisely because it values the business independent of whoever currently owns it.

The second situation is cross-border comparisons. Corporate tax rates vary significantly across jurisdictions — a US company might pay an effective rate of 21–25% while a comparable European business pays 15%. If you're benchmarking an industry globally, after-tax earnings comparisons introduce tax-regime noise. EBITDA removes it.

The third, and probably the most legitimate for stock investors, is analyzing asset-light businesses — particularly software and technology companies that grew through acquisition. When a company buys another business, it often records significant intangible assets on the balance sheet (customer relationships, developed technology, trade names) and then amortizes them over several years. That amortization shows up as a D&A charge on the income statement even though it doesn't represent anything physically wearing out. Salesforce, Oracle, and dozens of other enterprise software companies carry billions in acquired-intangible amortization annually. Stripping it out does give a cleaner picture of ongoing operating profitability. The free cash flow number tells a similar story but with a different set of trade-offs.

The key phrase in all three cases is "legitimate." EBITDA is a reasonable lens when the items being stripped out are genuinely not reflective of the ongoing operating business. The question that follows every EBITDA number is the same: what exactly did we take out, and does it actually go away?

Where it breaks down

Warren Buffett and Charlie Munger spent years criticizing EBITDA, and their sharpest point was about depreciation. Buffett's formulation was something like: management teams that talk about EBITDA seem to believe the tooth fairy pays for capital expenditures. The critique isn't rhetorical. It's mechanically accurate. If a piece of equipment depreciates by $10 million per year on the income statement, that's because in about ten years you'll need to spend real cash to replace it. The depreciation charge is the accounting system's attempt to spread that future cash cost over the asset's useful life. For a capital-intensive business, stripping D&A out of earnings doesn't remove a harmless accounting artifact — it removes the financial system's best estimate of real ongoing reinvestment costs.

Telecom is the textbook case. AT&T has historically reported tens of billions in EBITDA. But telecom requires continuous, massive capital investment to maintain and upgrade networks — fiber buildouts, spectrum purchases, equipment refresh cycles. AT&T's capex routinely runs $20 billion or more per year. When you subtract that from EBITDA alongside interest on its substantial debt load, the number that remains looks very different from the headline EBITDA figure. The company isn't hiding this — the cash flow statement shows it plainly — but EBITDA alone creates a flattering picture that disappears the moment you look one page further.

Airlines face the same problem. A narrow-body aircraft costs $100–150 million. It needs heavy maintenance every few years and eventually full replacement. An airline reporting $2 billion in EBITDA while owning or leasing several hundred planes is not a business where $2 billion flows freely to shareholders. The planes wear out. That wear-out is economically real whether you call it depreciation or not.

Interest expense is the other failure mode. If a company carries $1 billion in debt at 7%, that's $70 million per year walking out the door to creditors. EBITDA pretends those creditors don't exist. For a business with modest debt, that's a small distortion. For a highly leveraged company — think post-LBO businesses, cable operators during build-out phases, or any company that borrowed aggressively when rates were low — ignoring interest means ignoring the largest single line on the income statement. The business might show strong EBITDA while being technically insolvent on a free cash flow basis once debt service is accounted for.

The pattern to watch: the larger the gap between a company's EBITDA and its free cash flow, the more skeptically you should treat the EBITDA headline. Capital-intensive businesses, highly leveraged balance sheets, and companies with large working capital swings all produce wide gaps. The gap is not necessarily a fraud signal — it's often just the cost of operating that type of business — but it means EBITDA is doing very little analytical work for you.

EBITDA vs. free cash flow: the test that matters

Free cash flow is operating cash flow minus capital expenditures. It measures what the business actually generates in cash after keeping itself running and investing in growth. The comparison between EBITDA and FCF for the same company is one of the fastest diagnostic tests available in a financial statement. They start from different places — EBITDA begins on the income statement, FCF begins on the cash flow statement — but for an asset-light business they should end up reasonably close. When they don't, the gap is the story.

Consider a fictional industrial manufacturer (the numbers below are illustrative, not from any specific company). Revenue of $1 billion, EBITDA of $200 million — a 20% margin that looks solid. Now follow the cash. Capex runs $140 million per year, split between maintenance capex (replacing aging equipment) and modest growth investment. Operating cash flow, after changes in working capital that consume another $15 million, comes in around $175 million. Subtract capex and FCF is $35 million. After interest on $700 million in debt at 6% — $42 million per year — the company is effectively FCF-negative before taxes. EBITDA said 20% margin. Free cash flow said the business is barely treading water.

This isn't exotic. Businesses in capital-intensive industries regularly show this pattern. The issue is that an investor who stops at EBITDA never sees it. The reconciliation between the two numbers breaks down into three components: capex intensity (how much reinvestment the business requires), working capital dynamics (does the business tie up cash as it grows?), and debt service (how much goes to creditors before equity holders see anything). A software company with low capex, negative working capital (customers pay before costs are incurred), and no debt might convert 85–90 cents of every EBITDA dollar into free cash flow. The industrial manufacturer above converts less than 20 cents.

The practical habit: whenever you see an EBITDA number in a press release or investor presentation, pull the cash flow statement from the most recent 10-K or 10-Q and calculate FCF. Divide FCF by EBITDA. If that ratio is consistently above 70%, the business probably doesn't require enormous reinvestment and EBITDA is a reasonable approximation. Below 50%, EBITDA is papering over something real. Below 30%, you're looking at a business where EBITDA is close to useless as a valuation anchor.

EBITDA margins and EV/EBITDA multiples

EBITDA margin — EBITDA divided by revenue — is one of the most common profitability benchmarks in equity research. Software companies with strong pricing power often run 25–40% EBITDA margins; healthcare services might run 10–15%; grocery chains operate in the 3–6% range. The only valid use of the margin is as a within-industry comparison. A 12% EBITDA margin is strong for a distributor and mediocre for a SaaS business.

EV/EBITDA — enterprise value divided by EBITDA — is the valuation multiple built on top of it. Enterprise value is market cap plus net debt (total debt minus cash), which makes EV/EBITDA capital-structure neutral in the same way EBITDA itself is. An acquirer paying 12x EBITDA for a business is agreeing to pay 12 times the operating earnings before financing and accounting adjustments. The multiple embeds assumptions about growth, capex requirements, and industry durability. Software often trades at 20–30x or higher. Industrials and consumer businesses typically trade at 8–13x. When a company trades at a discount to its sector average on EV/EBITDA, the question is whether that discount reflects a real problem — higher capex needs, weaker growth, more leverage — or a genuine mispricing worth investigating. The multiple alone doesn't tell you which. Enterprise value has its own nuances worth understanding before leaning heavily on EV/EBITDA as a screening tool.

How to read it in an earnings release

EBITDA doesn't appear in GAAP financial statements. You won't find it on the income statement or in the audited financials. Companies report it in press releases, earnings call supplements, investor presentations, and 8-K filings — and they are required by SEC rules to reconcile it back to the nearest GAAP figure (typically net income) whenever they present it. That reconciliation table is the most important piece of the document.

A standard reconciliation runs: GAAP net income → add income tax expense → add interest expense → add depreciation and amortization → EBITDA. That's the clean version. Many companies then keep going to reach "adjusted EBITDA" by adding back additional items they've designated as non-recurring or non-cash. This is where the number can drift significantly from economic reality.

Common adjusted EBITDA add-backs to scrutinize:

• Stock-based compensation. A real cost to shareholders — it dilutes existing holders — but consistently excluded by technology companies. If SBC runs $150 million annually on $500 million of EBITDA, the adjusted number is 30% higher than what equity holders actually capture.
• Restructuring charges. Legitimate when a company genuinely reorganizes once. A red flag when restructuring charges appear in six of the last eight years.
• Integration and transaction costs. Serial acquirers sometimes spend continuously on integrating past deals while actively pursuing new ones. These costs are structural, not episodic.
• Litigation settlements and regulatory fines. Businesses in regulated industries — pharmaceuticals, financial services, energy — often face recurring legal costs. Calling each one non-recurring is technically accurate per settlement; calling the pattern non-recurring is not.

The test is simple: go back four to six years and count how often each excluded category appeared. If "non-recurring" charges recur every year, they are operating costs. Add them back into your own calculation and see what you're left with. That number is closer to the truth than the headline adjusted EBITDA. For more context on where these line items originate, see the guide on reading an income statement and the companion piece on reading an earnings release.