Maintenance CapEx Explained: Three Ways to Estimate the Number No One Reports

Maintenance capital expenditure is the portion of a company's capital spending required just to keep the existing business running at its current level — replacing worn machinery, re-roofing a plant, refreshing a truck fleet. The rest is growth capital expenditure: spending that adds capacity the company did not have before.

No company reports the split. The cash flow statement carries one line — "purchases of property, plant and equipment" — and it lumps the two together. Yet almost every serious cash-flow valuation depends on separating them, because only the growth half is optional. An analyst who treats all capital spending as a permanent cost understates the business; one who treats none of it as permanent overstates it.

This guide covers what the split is for, the three methods practitioners actually use to estimate it, a worked comparison of all three on the same set of numbers, and the situations where every method fails.

Why the Split Matters

Three widely used measures break if maintenance capital expenditure is wrong.

Owner earnings. The measure Warren Buffett described in his 1986 shareholder letter — covered in more depth in the Warren Buffett stock valuation method guide — is defined as reported earnings plus depreciation and amortisation, minus maintenance capital expenditure, minus any incremental working capital the business needs. Maintenance capital expenditure is the only term in that formula that does not appear in a filing.

Free cash flow quality. Free cash flow is normally computed as operating cash flow minus total capital spending. That is conservative for a company in an expansion phase: it charges the current year with spending whose benefit arrives in later years. A company reinvesting heavily can show near-zero free cash flow while the underlying business throws off cash comfortably.

Terminal value. In a discounted cash flow model, the assumption about steady-state reinvestment runs through every projected year and then through the terminal value, which typically accounts for the majority of the total. An error here does not average out — it compounds in the same direction forever. Arithmetically, at a fixed discount rate, a persistent 15% overstatement of steady-state cash flow produces roughly a 15% overstatement of the resulting estimate.

That last point is why the number deserves real work rather than a default.

Method 1: Depreciation as a Proxy

The simplest approach sets maintenance capital expenditure equal to depreciation expense. The reasoning is that depreciation is accounting's own estimate of how fast the asset base wears out, so replacing that much each year should hold capacity constant.

It has the advantage of being free — the number is already on the income statement — and it is roughly right for a mature, stable, capital-intensive business in a low-inflation period.

It has four known failure modes, and all four bias in the same direction.

• Historical cost. Depreciation is booked against what an asset originally cost. A press bought in 2009 is depreciated at 2009 prices and replaced at today's prices. In any inflationary stretch the proxy understates the real replacement bill.
• Growth lag. In a company whose asset base has been expanding, this year's depreciation reflects an average of a smaller historical base, not the larger base now in service.
• Purchase accounting. After acquisitions, the depreciation and amortisation line carries amortisation of acquired intangibles — customer lists, developed technology — that require no cash replacement at all. The proxy then overstates maintenance spending, the one case where it errs the other way.
• Useful-life estimates. Asset lives are a management estimate. Extending the assumed life of a fleet lowers depreciation immediately without changing how often trucks physically need replacing.

The proxy is a reasonable sanity check and a poor primary estimate.

Method 2: The PP&E-to-Sales Ratio

This is the method Bruce Greenwald set out in Value Investing: From Graham to Buffett and Beyond, and it is the one most commonly taught for a reason: it uses only reported figures and it makes its assumption explicit.

The logic is that a business needs a roughly stable amount of gross property, plant and equipment to support each dollar of sales. If that ratio holds, then the capital required to support this year's increase in sales is growth capital expenditure, and whatever remains of total capital spending is maintenance.

The procedure:

1. For each of the last five years, compute gross PP&E divided by sales.
2. Average those five ratios. This is the capital intensity of the business.
3. Multiply the average ratio by the current year's change in sales. That is estimated growth capital expenditure.
4. Subtract it from total capital expenditure. The remainder is estimated maintenance capital expenditure.

Use gross PP&E, before accumulated depreciation. Net PP&E falls as assets age even when nothing about the business changes, which makes the ratio drift for reasons that have nothing to do with capital intensity.

Worked Example

The figures below describe an illustrative manufacturer, Cascade Industrial. They are constructed to demonstrate the arithmetic and do not describe any real company.

Step 1 and 2 — capital intensity

Five-year average ratio: 0.704

Step 3 — growth capital expenditure

Sales rose from $2,200M to $2,400M, an increase of $200M.

Growth CapEx = 0.704 × $200M = $141M

Step 4 — maintenance capital expenditure

Cascade's 2025 cash flow statement reports total capital expenditure of $310M.

Maintenance CapEx = $310M − $141M = $169M

Depreciation for the same year was $145M. The two methods differ by $24M, and the depreciation proxy is the lower of the two — consistent with the historical-cost and growth-lag effects described above.

What the gap does to owner earnings

Cascade reported net income of $210M, depreciation and amortisation of $145M, and a $18M increase in working capital.

A $24M difference in one input moves the output by 14%. Nothing else in the calculation changed, and both columns are defensible from the same filing. That spread is the honest uncertainty in the number, and it is why the estimate belongs in a sensitivity range rather than in a single figure carried to two decimal places.

Method 3: Reading What Management Discloses

A minority of industries disclose the split directly. Railroads, pipelines, utilities, mining companies and many REITs separate replacement or sustaining capital from expansion capital in the management discussion section, in an earnings presentation, or in a regulatory filing.

Where it exists, this disclosure is the best available starting point, because management knows the physical asset base and an outside model does not. Three cautions apply:

• The definitions are not standardised. One company's "sustaining capital" includes regulatory compliance spending; another's does not.
• The line is not separately audited. It is a management characterisation of an audited total, not an audited figure in its own right.
• The classification is discretionary and has a direction of bias — a larger growth bucket flatters every derived cash flow measure. Comparing the disclosed split against the PP&E-to-sales estimate is a cheap consistency check.

Where Every Method Fails

Asset-light businesses. For a software company, capital expenditure is a rounding error, and the spending that actually maintains competitive position — research and development, and the engineering headcount funded partly through stock-based compensation — is expensed as incurred rather than capitalised. A PP&E ratio computed here produces a number with no economic meaning. The maintenance question for these businesses is a different question, asked of the operating expense base.

Serial acquirers. Acquisitions add gross PP&E without any corresponding organic capital expenditure, which breaks the ratio in step 1 and understates capital intensity in the years after a large deal. Adjusting requires stripping acquired assets out of the gross PP&E series, which is often not possible from public disclosure alone.

Leases. The 2019 lease-accounting changes (ASC 842 and IFRS 16) brought right-of-use assets onto the balance sheet. Whether they sit inside the PP&E line depends on presentation choices, and the discontinuity that year can distort a five-year ratio series that spans it. Check that the series is measured consistently before averaging it.

Declining sales. When sales fall, the change in sales is negative and step 3 returns negative growth capital expenditure, implying maintenance spending above total capital expenditure. That is not necessarily an error — a shrinking business may genuinely be under-maintaining its asset base — but the arithmetic should be treated as a flag for further work rather than an output.

Deliberate under-investment. Any method anchored on what a company actually spent measures spending, not need. A firm cutting maintenance to protect near-term reported cash flow will show a falling maintenance capital expenditure estimate at exactly the moment the real requirement is rising. Multi-year trends in the ratio, and in the average age of the asset base (accumulated depreciation divided by gross PP&E), are the usual cross-checks.

How This Shows Up in Equity Rank

Equity Rank's cash-flow models run from reported figures rather than from an estimated maintenance split, which keeps them reproducible from filings and free of an unobservable input that would differ between any two analysts. The one place a maintenance factor appears explicitly is the REIT adjusted-funds-from-operations calculation, which applies a fixed industry-average share of capital expenditure — a convention of that sector's reporting, and a simplification that the model states rather than hides.

For readers doing the work by hand, the free cash flow and normalised earnings guides cover the adjacent adjustments, and the intrinsic value calculator accepts a cash flow figure of the reader's own construction — including one built on a maintenance capital expenditure estimate derived by the method above.

Frequently Asked Questions

Is maintenance CapEx reported anywhere in the financial statements? No. The cash flow statement reports total capital expenditure as a single line. Some companies in capital-intensive industries disclose a maintenance-versus-growth split in the management discussion or an earnings presentation, but it is a management characterisation rather than a separately audited figure.

Can depreciation be used as maintenance CapEx? It can be used as a rough check. It tends to understate the figure for a growing company or during inflationary periods, because depreciation is booked at historical cost against a smaller historical asset base, and it can overstate it for a company carrying large amortisation of acquired intangibles.

Should gross or net PP&E be used in the ratio method? Gross. Net PP&E declines as existing assets age even when the business is unchanged, so a ratio built on it drifts for accounting reasons rather than economic ones.

What happens if the estimate comes out negative? A negative result usually means total capital expenditure was lower than the estimated growth requirement, which happens when the ratio is measured across a period containing an acquisition or a lease-accounting change. Re-check the gross PP&E series for consistency before using the output.

How much does the estimate change a valuation? It scales roughly one-for-one. Because the assumption runs through every projected year and into the terminal value, a persistent 15% error in steady-state cash flow moves the resulting estimate by roughly 15% at a fixed discount rate. That sensitivity is the argument for carrying a range rather than a point estimate.

Does maintenance CapEx apply to software and other asset-light companies? Not usefully. Their capital expenditure is small and the spending that sustains the business is expensed rather than capitalised, so a PP&E-based estimate returns a figure with no economic content. The equivalent question for those businesses concerns the operating expense base, not the capital budget.

This content is for educational and informational purposes only and does not constitute investment advice. Equity Rank is not a registered investment adviser. Cascade Industrial is an illustrative construction and does not describe any real company; the figures are arithmetic demonstrations, not estimates about any security. Accounting treatments differ by jurisdiction, industry and reporting period — financial statement data should be verified against primary filings before use.