Maintenance vs. Growth Capex: What Belongs in Free Cash Flow

The Purpose Decides the Treatment

Free cash flow gets used for three different jobs, and each one implies a different treatment of capital expenditure. Deciding which job is in front of you settles most of the debate before it starts.

In a discounted cash flow model, all capex goes in. The forecast already contains the revenue that growth spending is supposed to produce, so dropping the spending while keeping the revenue counts the benefit without the cost. There is no real debate here, only the occasional model that has quietly made this mistake in row 40.

In credit and liquidity work, all capex goes in as well. A lender cares about cash that actually leaves the building. The only distinction worth drawing is between capex the company could defer in a downturn and capex it cannot, because that gap is the cushion available when the cycle turns.

The third job is the interesting one: using free cash flow as a measure of current earning power, which is what an FCF yield or a price-to-FCF multiple implies. The question there is what the business as it stands today could distribute to owners without shrinking. That calls for maintenance capex only. Charging a company for investing in attractive growth makes it screen as expensive precisely when it is doing the right thing.

Most published FCF yields are built on the first treatment while being interpreted as the third. That mismatch is the single most common error in this area, and it systematically penalises companies during their heaviest investment years.

Maintenance Capex Always Belongs In

Maintenance capex is the spending needed to hold unit volumes and competitive position where they already are. Accounting spreads it over several years through depreciation, but economically it is a recurring cost of operating, in the same sense as wages. Any free cash flow measure that excludes it describes a company that does not exist. That is the core objection to valuing a business on EBITDA, and it applies just as forcefully to a homemade FCF figure that quietly drops the replacement spending.

The difficulty is that almost no company discloses maintenance capex cleanly, so it has to be estimated. Depreciation is the usual proxy and it is a poor one in both directions. For asset-heavy businesses it understates the true figure, because depreciation is struck on historical cost and replacing a fifteen-year-old plant costs more than the plant did. For businesses whose assets outlive their accounting lives, it overstates.

Bruce Greenwald's approach is more defensible: take the long-run ratio of property, plant and equipment to sales, multiply it by the year's increase in sales to estimate growth capex, and treat whatever remains as maintenance. Worked through on an illustrative industrial business, the method looks like this.

The figures above are illustrative rather than drawn from a specific filing, but the shape is typical: depreciation understated the replacement burden by a fifth. Two cross-checks are worth running on any estimate of this kind. Compare the implied maintenance intensity against the capex-to-sales ratio of mature, non-growing peers, and compare it against whatever management says in the capital markets day deck, with appropriate scepticism about the incentive behind that number. The logic here is close to Buffett's owner earnings, covered separately in Owner Earnings vs. Free Cash Flow.

Growth Capex Comes Out Only With Discipline

Growth capex can legitimately be excluded from a steady-state measure, on one condition. The business then has to be valued as if the growth does not happen, or the growth has to be valued separately and added back. The mistake to avoid is stripping out growth capex to produce a flattering FCF figure and then applying a multiple that already assumes strong growth. It is the DCF error in different clothing, and it appears in investor presentations constantly.

The second discipline is to distrust the label itself. Management has every incentive to classify spending as growth, because doing so makes underlying cash generation look better without changing anything real. A useful test: ask what happens to revenue if the spending stops. If revenue holds, the spending was growth capex. If revenue erodes, it was maintenance regardless of what the slide said.

Retail makes the point concretely. A chain opening 50 stores while closing 40 is adding ten stores of net capacity, so roughly 80% of that store capex is replacing selling space the business is losing. Presenting the full amount as growth investment overstates both the expansion and the underlying free cash flow. Technology has its own version. A company spending heavily just to keep its product at parity with rivals is doing defensive capex, which earns no incremental return and belongs in the maintenance bucket. In genuinely competitive industries, a large share of what gets labelled growth capex is simply the price of staying in the game.

The Grey Zones Outside the Capex Line

Several categories never touch the capex line at all, and they are where the larger errors tend to live. The table below sets out how conventional free cash flow handles each one and what a steady-state measure should do instead.

Acquisitions deserve the most attention. Excluding them is reasonable for a company that does one deal a decade. For a serial acquirer, where buying businesses is the growth strategy and a substitute for internal research or capex, acquisitions are capex under another name. Take a business generating $1.0B of operating cash flow against $100M of capex: reported free cash flow of $900M looks superb until you notice average annual deal spending of $600M, which leaves $300M that shareholders could actually keep.

Capitalised software creates a subtler problem, a comparability trap. A company that capitalises development shows higher operating cash flow than an otherwise identical company that expenses it, purely because of the accounting choice. Deducting the capitalised amount puts the two back on the same footing before any multiple is applied.

Leases need care under IFRS 16, because the principal portion of lease payments now sits in financing cash flow. Operating cash flow minus capex therefore flatters every lease-heavy business, including retailers, airlines and shipping companies. A retailer with $900M of operating cash flow and $300M of capex reports $600M of free cash flow; deduct $250M of lease principal and the honest figure is $350M, some 42% lower. Finance leases have the mirror-image problem. The assets never appear in the capex line because no cash moved on day one, yet they are economically capex, so the non-cash investing disclosures are worth reading.

The remaining categories are less contentious but easy to get wrong. Purchased intangibles are capex for the business models that depend on them, and a streaming company's content spend is its maintenance capex. Regulatory, safety and environmental spending earns no return and still has to be included in full, because it is the price of the licence to operate; the same logic covers resource companies replacing depleted reserves. Asset sale proceeds are lumpy and finite, which is why netting them against capex flatters the result, with car rental and equipment leasing standing as the fair exceptions where net capex is the honest figure.

Report Two Numbers, and Read the Gap

Rather than defending one definition, report two. The first is free cash flow after all capex, including lease principal, capitalised intangibles and a normalised level of acquisitions where the company is acquisitive. That is the hard number and nobody can argue with it. The second is free cash flow after maintenance capex only, clearly flagged as an estimate, with the method stated so a reader can disagree with the assumption rather than the arithmetic.

The gap is the most informative of the three figures. It quantifies how much management is choosing to reinvest, which can then be judged on its own terms by tracking incremental returns on invested capital over the following three to five years. Where those returns are high, a low headline free cash flow is a feature rather than a defect. Where they are mediocre, the "growth" capex was either maintenance in disguise or outright value destruction, and the lower free cash flow figure was the true one all along.

Running this exercise across a portfolio also changes which screens are worth trusting. A screen built on conventional free cash flow will rank a heavy reinvestor below a harvesting competitor even when the reinvestor is compounding capital at attractive rates, which is why the reinvestment gap belongs next to the yield rather than buried underneath it. The FCF screener and the FCF Yield Calculator both start from the conventional definition, so the adjustments above are applied on top of the screened figure rather than instead of it.

Conclusion

The principle underneath all of this is consistency. Capex can be excluded from free cash flow only when the benefit it buys is also excluded from the valuation. Every error in this area, from EBITDA multiples to the adjusted FCF figures in investor presentations, comes from keeping the benefit and dropping the cost.

Maintenance capex is an estimate and will stay one, so the useful discipline is stating the method rather than pretending to a precision the disclosures do not support. Two numbers, the hard one and the estimated one, with the reasoning for the split visible between them, is a more honest presentation than any single figure. The gap between those two numbers is where the actual analysis lives, because it converts a question about accounting into a question about capital allocation, which is the one that determines returns.

For the mechanics of the underlying calculation, see How to Calculate Free Cash Flow. For why cash-based measures resist manipulation better than earnings, see Free Cash Flow vs. Net Income. And for the tests that separate durable cash generation from a good year, see Assessing FCF Quality.

Disclaimer: This analysis is for educational purposes only and does not constitute investment advice, financial advice, trading advice, or any other type of advice. You should not make any investment decision based solely on this analysis. Always conduct your own due diligence and consult with a licensed financial advisor before making any investment decisions. Past performance does not guarantee future results. All investments carry risk, including the potential loss of principal.