A company can go bankrupt while its EBITDA grows. That isn’t a textbook paradox or a laboratory case: it’s the ordinary outcome whenever someone mistakes an accrual metric for the ability to actually pay wages, interest and suppliers. EBITDA does not measure cash. It measures an edited version of operating profit in which four things have been deliberately switched off: capital structure, taxation, accounting policy on long-lived assets and — this is the part almost nobody says out loud — the moment money actually moves.
That last point is the one that matters. EBITDA inherits accrual accounting’s founding rule: revenue is recognised when earned and expenses when incurred, not when collected or paid. A sale on 180-day terms hits EBITDA the day the invoice is issued. Inventory bought and not yet sold doesn’t hit it at all. A customer prepayment doesn’t either. EBITDA is, in essence, a photograph of the promise; operating cash flow (CFO) is a photograph of the bank statement.
The bridge between the two is the tool that reconciles those photographs. And it isn’t an academic exercise: it’s the most honest operational question you can ask a set of financial statements. Of these €26 million of EBITDA, how many have I actually seen?
What EBITDA Measures, and What It Chose Not to Look At
EBITDA was built for a specific purpose: comparing the operating profitability of businesses with different debt structures, tax regimes and depreciation policies. In capital-intensive sectors — infrastructure, energy, telecoms, cable — it was a reasonable answer to a real problem. The trouble is that a metric designed to neutralise accounting differences ended up being used as a proxy for cash generation, which is exactly what it isn’t.
It helps to lay out the exclusions and ask, one by one, whether there’s real money behind them:
| Excluded from EBITDA | Is there a cash outflow? | When it happens |
|---|---|---|
| Interest | Yes | Contractual, recurring, senior |
| Taxes | Yes | Instalments and annual settlement |
| Depreciation (fixed assets) | Not this period | Cash left earlier (capex) and will leave again |
| Amortisation (intangibles) | Not this period | Cash left at acquisition; sometimes never again |
| Change in working capital | Yes | Continuous, invisible in the P&L |
| Capex | Yes | Neither in EBITDA nor in CFO — it sits in investing |
Three of the six lines are pure cash. And the two that aren’t — D&A — aren’t free either, as we’ll see.
The Structure of the Bridge
The full bridge, in canonical form:
EBITDA
± Non-cash items still embedded in EBITDA
(provisions, share-based comp, inventory write-downs)
± Change in operating working capital
(receivables, inventory, payables, other)
− Interest paid
− Taxes paid
= CASH FLOW FROM OPERATIONS (CFO)
− Capex
= FREE CASH FLOW (FCF)
Before we go line by line, a warning about the order of operations. The classic indirect method starts from net income and adds D&A back. When your starting point is EBITDA, that add-back has already happened: D&A doesn’t appear in the bridge because it’s baked into the starting point. This causes constant confusion in interviews and in investment committees — someone asks «where do you add back depreciation?» and the correct answer is «it’s already added back, and that’s precisely the problem.»
1. Interest and Taxes: Cash Feels Them Even If EBITDA Doesn’t
Interest
EBITDA ignores interest because it wants to be neutral about how the business is financed. That’s a useful abstraction when comparing two identical plants, one levered at 70% and one debt-free. It’s a lethal abstraction when the question is whether the company survives the next quarter, because lenders don’t get paid in EBITDA.
The arithmetic is brutal and rarely made explicit. At net debt of 4.0x EBITDA and a 7% average cost, interest consumes 28% of EBITDA before a single euro of tax, capex or working capital. At 5.0x and 9% — entirely plausible in the recent rate environment — 45 points of EBITDA are gone. That’s the real reason Net Debt / EBITDA is an incomplete indicator: it measures the stock, not the service.
You also need to separate two numbers that almost never match:
- Interest expense (accrual): what shows up in the P&L. It includes PIK interest that capitalises and never leaves the bank account, amortisation of debt issuance costs, and the unwinding of discounts on provisions.
- Interest paid (cash): what leaves the bank account. This is the number that belongs in the bridge, and it’s disclosed in the cash flow statement.
A PIK tranche inflates interest expense without touching cash today — and punishes you with compounding tomorrow. Capitalised interest on assets under construction does the opposite: cash leaves but never passes through the P&L, because it’s capitalised on the balance sheet.
IFRS 16 deserves its own paragraph. Since it took effect, operating lease rent is no longer an operating expense: it’s split into depreciation of the right-of-use asset and interest on the lease liability. Both sit below EBITDA. The result: EBITDA for retailers, airlines and any business with meaningful leases jumped overnight without cash improving by a cent. Comparing EV/EBITDA multiples pre- and post-IFRS 16 without adjusting is comparing different things, and most banking covenants were renegotiated precisely because of this.
One technical note worth carrying around: under IAS 7, interest paid can be classified in either operating or financing activities. Two identical companies can report different CFO purely through presentation choice. Normalise before you compare.
Taxes
Here the mismatch is a different animal. Corporate income tax isn’t computed on EBITDA but on taxable income, which starts from accounting profit and applies fiscal adjustments. And the tax expense in the P&L also isn’t the tax paid, for three recurring reasons:
- Deferred taxes: temporary differences between accounting and tax treatment (accelerated depreciation, provisions non-deductible until paid, limits on interest deductibility).
- Loss carryforwards: a company with accumulated tax credits can post profit and pay close to zero cash tax for years. That’s a genuine advantage — and a finite one.
- Payment timing: instalments based on estimates, later settlement, refunds. The lag can run several quarters.
What belongs in the bridge is tax paid, not tax accrued. If the gap between the two is persistent and large, it isn’t noise: it’s a signal that accounting profit and fiscal reality are telling different stories, and it’s worth understanding why.
2. D&A: Not Cash Today, but Maintenance Capex Proves It Still Matters
Depreciation and amortisation are non-cash charges. This is true and it is boring. What’s interesting is why they’re non-cash: because the money already left. Depreciation is the accounting allocation, across an asset’s useful life, of a cash outflow that happened in the past. It isn’t a fictional expense; it’s an expense deferred in its recognition and prepaid in its settlement.
Hence the conceptual error that props up half the market: treating EBITDA as if capex were optional. In a steady-state business — no growth, assets that wear out and get replaced — maintenance capex is as recurring and as compulsory as payroll. The only difference is that it’s booked in investing rather than operating.
Two concepts worth separating, which companies rarely disclose:
| Type of capex | Nature | Can it be deferred? |
|---|---|---|
| Maintenance | Replace existing capacity | Only temporarily, at a growing deferred cost |
| Growth | Add new capacity | Yes, it’s discretionary |
The usual shortcut — maintenance capex ≈ D&A — works reasonably well in steady state with low inflation. It breaks in three very common situations:
- Replacement cost inflation: depreciation is calculated on historical cost. Replacing a machine bought ten years ago costs materially more today. Depreciation understates true maintenance capex.
- Stretched useful lives: extending the accounting life of assets lowers depreciation, lifts EBIT and changes nothing in cash. It’s one of the cleanest cosmetic levers available, and it’s entirely legal.
- PPA intangible amortisation: after an acquisition, purchase price allocation creates intangibles (customer relationships, brands, technology) that amortise for years with no associated replacement capex. That amortisation genuinely is paper — but the cash left in full on closing day.
There’s an additional effect that runs the other way and needs watching: expense capitalisation. When a company capitalises internal development, software or customer acquisition costs that previously went through the P&L, it moves operating expense into capex. EBITDA rises. Cash doesn’t move an inch. A jump in EBITDA margin accompanied by a jump in capex and in balance sheet intangibles isn’t an operational improvement — it’s a relocation.
And one final structural observation: capex doesn’t appear in CFO. It sits in investing activities. That’s why CFO also systematically flatters capital-intensive businesses, and why the bridge doesn’t end at CFO but at free cash flow.
3. Working Capital: Where Growth Turns Into a Hole
This is the line that kills the most companies and shows up least often in investor presentations. The sign convention is elementary and needs to be internalised until it’s automatic:
An operating asset that increases consumes cash. An operating liability that increases releases cash.
Growing receivables or inventory means value has been recognised without being collected, or goods have been paid for without being sold. Growing payables means someone is financing your operations without charging explicit interest for it.
Accounts Receivable
EBITDA recognises the sale when it’s invoiced. Cash arrives when it’s collected. The distance between those two moments is DSO (Days Sales Outstanding):
DSO = (Receivables ÷ Revenue) × 365
In a growing company, that distance is a structural bleed. With revenue of €100m and DSO of 90 days, receivables sit at roughly €24.7m. If revenue grows to €130m while collection policy stays exactly the same, the balance moves to €32.1m: €7.4m of cash trapped without anything having deteriorated. Growth alone funds receivables.
What matters isn’t the level but the derivative: if receivables grow faster than revenue, DSO is deteriorating and you need to ask why. The answers are usually three, and none of them is good — you’re selling to worse customers, you’re stretching terms to close the quarter, or there’s billing the customer disputes and has no intention of paying.
And a measurement trap: the receivables balance is a snapshot at the reporting date. A factoring facility signed in December cleans up the balance sheet at 31/12 and produces an immaculate DSO that reflects nothing about the year. Go find the receivables-sold disclosure in the notes.
Inventory
Inventory is EBITDA’s most elegant blind spot. When goods are purchased, cash leaves but there is no expense: the cost is capitalised on the balance sheet and only reaches the P&L as cost of goods sold when the product is sold. Which means a company can build stock aggressively, drain its treasury and report EBITDA that is intact or even improved.
DIO = (Inventory ÷ COGS) × 365
Two nuances that get overlooked:
- Fixed cost absorption through producing for stock improves the period’s margin even when demand doesn’t exist. Cash finds out before the P&L does.
- An inventory write-down is a non-cash charge in the period it’s recognised — but the cash was spent when the goods were bought. Adding the write-down back in the bridge is technically correct and economically misleading: it doesn’t mean the inventory cost nothing, it means it cost money earlier.
Accounts Payable
Stretching supplier payments generates cash. It is, literally, short-term financing, and it appears in the bridge as an inflow. The problem is that it’s booked as a trade liability rather than financial debt, which makes it invisible in the leverage ratio.
DPO = (Payables ÷ COGS) × 365
A DPO that expands persistently is one of the most reliable warning signs there is, for two reasons. First, because it’s finite: at some point the supplier demands payment upfront or stops shipping, and the reversal is abrupt and happens at the worst possible moment. Second, because it usually travels with reverse factoring (supply chain finance): a bank pays the supplier immediately and the company pays the bank later. Economically that’s financial debt; for a long time, in accounting terms, it stayed within trade payables. When an auditor, a regulator or a buyer forces reclassification into debt, apparent leverage and apparent cash generation both get restated at once. The Carillion collapse in the UK and Abengoa in Spain are widely documented examples of how far this distortion can go.
The Cash Conversion Cycle
The three fold into a single metric:
CCC = DSO + DIO − DPO
The CCC is the number of days between paying and collecting, and it translates directly into euros. Each day of DSO is worth roughly one day of revenue; each day of DIO or DPO is worth one day of COGS. For the company in our example (€130m of revenue), cutting 10 days of DSO releases about €3.6m of cash in one go — without selling a single euro more or gaining a point of margin. This is why the first hundred value-creation levers in any operational improvement plan sit in working capital, not in the P&L.
4. Other Adjustments: Provisions, Prepayments and Deferred Revenue
That leaves the items that fit none of the above categories and typically hide inside a three-line «other».
Provisions. A restructuring, litigation or warranty provision reduces profit in the year it’s recognised, with no cash out. In the year it’s settled, cash leaves with no P&L impact. This is the accrual mismatch in its purest form, and it works in both directions:
- The year of recognition: if the provision was booked above EBITDA, it gets added back in the bridge (it’s non-cash).
- The year of payment: it gets subtracted, because cash is leaving that no EBITDA figure captures.
This is where the most common vice of adjusted EBITDA shows up: excluding «non-recurring» restructuring charges that were paid in cash and that, suspiciously, reappear every single year. A non-recurring expense that occurs four years running is a recurring expense with better PR.
Prepaid expenses. Annual insurance, software licences, prepaid rent, contracted campaigns. Cash leaves in full today; the expense is recognised over the following months. A rising prepaid balance consumes cash while EBITDA doesn’t flinch.
Customer prepayments and deferred revenue. The mirror image, and from a treasury standpoint the best business model there is: you collect before you deliver. SaaS with annual upfront billing, subscriptions, construction contracts with favourable milestones, or any negative-working-capital business generates cash by growing, not despite growing. It’s exactly the inverse of the distributor on 90-day terms. Worth remembering in the other direction too: when that business stops growing, the effect stops dead.
Share-based compensation. It isn’t cash, so it’s added back in CFO. But it is a real cost: it transfers value from existing shareholders to employees through dilution. Treating it as free because it doesn’t touch the bank account is the symmetrical error to treating capex as optional.
The Full Bridge, With Numbers
An industrial distributor. Revenue from €100m to €130m (+30%), stable 20% EBITDA margin, net debt of €60m at 7%.
| Bridge line | €m | Comment |
|---|---|---|
| EBITDA | 26.0 | 20% on €130m of revenue |
| Δ Receivables (DSO 90d) | (7.4) | Growth funds the customers |
| Δ Inventory (DIO 100d) | (5.3) | Stock to support higher volume |
| Δ Payables (DPO 60d) | +3.2 | Trade financing, limited |
| Δ Working capital | (9.5) | 37% of EBITDA trapped |
| Restructuring cash payments | (1.5) | Excluded from «adjusted EBITDA» |
| Interest paid | (4.2) | €60m × 7% |
| Taxes paid | (3.5) | On taxable income, not on EBITDA |
| Cash flow from operations (CFO) | 7.3 | Conversion: 28% of EBITDA |
| Maintenance capex | (6.0) | Non-discretionary |
| Growth capex | (4.0) | Discretionary |
| Free cash flow (FCF) | (2.7) | Negative |
The press release headline will read «EBITDA of €26m, +30% year on year, leverage of 2.3x». Every word of that is literally true. And the company burned €2.7m growing at a 20% margin. The faster it grows, the more cash it needs. If credit dries up, the business stops — not for lack of profitability, but through the arithmetic of working capital.
How to Read a Bridge in Ten Minutes
A protocol that works for any company in any sector:
- Compute cash conversion: CFO / EBITDA. Sustained above 80% is healthy. Sustained below 50% demands a specific explanation, not a narrative.
- Aggregate five years. Σ CFO against Σ EBITDA. Timing differences wash out over a full cycle; a gap that persists for five years is structural, not calendar.
- Compare receivables growth to revenue growth. If the former outruns the latter, DSO is deteriorating and you need to ask about billing quality.
- Watch DPO and search the notes for supply chain finance. A sustained extension of payment terms is financial debt dressed as a trade balance.
- Contrast D&A with capex. If capex sits systematically below depreciation, the business is disinvesting. If it sits well above, split maintenance from growth.
- Audit every adjusted-EBITDA add-back. For each excluded item: did cash leave? Does it repeat? If both answers are yes, it wasn’t an adjustment — it was an expense.
- Close with the number that matters: EBITDA − interest paid − taxes paid − Δ working capital − maintenance capex. That’s what’s left to repay debt, pay dividends or reinvest.
So Is EBITDA Good for Anything?
Yes — and it’s worth avoiding the opposite dogma. EBITDA answers one specific question well: what is the operating profitability of this business before financing, tax and depreciation-policy decisions? For comparing two similar assets with different structures, for defining covenants, for building valuation multiples with a reasonably homogeneous denominator, it’s a legitimate and useful tool.
The error isn’t using it. The error is using it to answer a question that isn’t its own. EBITDA doesn’t tell you whether the company can service its debt, doesn’t tell you whether growth is self-funding, and doesn’t tell you how much money is left at year end. That’s what the bridge is for.
The discipline, in the end, reduces to a habit: every time an EBITDA figure appears, ask what happens between that figure and the bank balance. There are almost always five or six lines in between. And almost always, those lines contain the company’s real story.
Deja una respuesta