Capitalizing vs Expensing a Cost

A cost is a cost. But the moment a company writes a cheque, an accountant has to answer one question about it: did that money buy something that will still be useful next year? If yes, the spend goes onto the balance sheet as an asset and is charged against profit slowly, over the years it helps to produce. If no, it hits the income statement immediately and is gone.

That is the whole distinction. Capitalize means put it on the balance sheet and amortise it. Expense means take the hit now. The decision changes nothing about the cash — the money left the building on the same day either way — but it completely reorders when the profit appears, and it moves the outflow from one section of the cash flow statement to another.

Where the line actually sits

US GAAP is more specific than the general principle suggests, and the boundaries are the part worth memorising.

• Property, plant and equipment. The purchase price plus everything needed to get the asset ready for its intended use — freight, installation, testing, site preparation — is capitalized. Ordinary repairs and maintenance are expensed. A betterment that extends the useful life or increases capacity is capitalized.
• Interest during construction. Interest incurred while a qualifying asset is being built is capitalized into its cost rather than expensed, under ASC 835-20.
• Internal-use software (ASC 350-40). The rule is staged. Costs in the preliminary project stage — evaluating alternatives, scoping — are expensed. Costs in the application development stage — coding, configuring, testing — are capitalized. Costs after go-live, including training and maintenance, are expensed. Two companies building the same system can capitalize very different fractions purely by drawing the stage boundaries differently.
• Research and development. Under US GAAP (ASC 730) R&D is expensed as incurred, full stop. Under IFRS (IAS 38) research is expensed but development costs must be capitalized once six specific criteria are met. This is one of the largest genuine differences between the two regimes and it makes cross-border R&D-heavy comparisons unreliable.
• The capitalization threshold. Every company sets a floor in policy — commonly a few thousand dollars per item — below which everything is expensed regardless of useful life, because tracking a $400 chair for ten years costs more than it is worth.

Worked example: the same $100 million, two ways

A company has revenue of $500 million and cash operating costs of $300 million, before a $100 million project it undertakes this year. Tax is 25 percent, and assume the tax authority allows an immediate deduction either way — so cash taxes are $25 million in both scenarios and any book/tax difference lands in deferred tax. If capitalized, the asset amortises straight-line over five years, $20 million a year.

$ millions	Expensed	Capitalized
Revenue	500	500
Cash operating costs	(300)	(300)
Project cost through P&L	(100)	—
EBITDA	100	200
Amortization	—	(20)
Operating profit	100	180
Tax expense (25%)	(25)	(45)
Net income	75	135
Net income	75	135
+ amortization	—	20
+ deferred tax	—	20
Cash from operations	75	175
Capital expenditure	—	(100)
Free cash flow	75	75

Check the cash directly: receipts $500 million, less $300 million of operating costs, less the $100 million project, less $25 million of cash tax, equals $75 million. That is true in both columns, because nothing about the cash changed.

Now look at what moved. EBITDA doubled. Net income rose 80 percent. Operating cash flow went from $75 million to $175 million. Every metric that stops before capital expenditure is flattered, and only free cash flow — the one that subtracts capex — is honest. This is precisely why EBITDA multiples are dangerous when comparing companies with different capitalization policies.

Worked example: reversing the policy for R&D

Analysts often undo the rule and capitalize R&D themselves, because a research-heavy company's assets are real even if GAAP refuses to record them. A company has spent, in millions, 248, 273, 300, 331, 364 and 400 over the last six years — growth of about 10 percent a year. Amortise each year's spend straight-line over five years, starting the year after it is incurred.

1. This year's amortization: $0.2 \times (364 + 331 + 300 + 273 + 248) = 0.2 \times 1516 = 303.2$.
2. The research asset: the unamortized remainder of each year's spend, so $400 + 0.8(364) + 0.6(331) + 0.4(300) + 0.2(273) = 1064.4$.
3. Adjust operating profit: add back the $400 million expensed, subtract the $303.2 million of amortization, a net $+96.8$.
4. Adjust invested capital: add the $1,064.4 million asset.

If reported operating profit was $600 million on $3,000 million of invested capital, reported return on capital is 20.0 percent. Adjusted, it is $696.8 / 4064.4 = 17.1$ percent. Capitalizing R&D made the company look less profitable, not more, because the balance sheet grew proportionally more than earnings.

Note the condition that drives all of it. If R&D spending were flat, amortization would exactly equal the current year's spend and reported profit would be unchanged. The distortion exists only while spending is growing or shrinking — which is exactly when analysts most want to compare companies.

Key terms

• Capitalize — record a cost as an asset and charge it to profit over its useful life.
• Expense — charge a cost to profit in the period incurred.
• Betterment — a spend that extends life or capacity, and so is capitalized rather than treated as maintenance.
• Application development stage — the middle stage of an internal software project, the only one whose costs are capitalized.
• Research asset — the analyst-constructed balance sheet asset from capitalizing R&D that GAAP expenses.