Two companies, both reported as earning "22% on capital". One of them earns 22% on money that includes a ₹1,400 crore cash pile it has no use for. The other earns 22% on money that is entirely at work.
The second business is better and the ratio doesn't say so.
That gap is what separates return on capital employed from return on invested capital. It's worth a page, because most of the disagreement about which company is a compounder comes down to which denominator somebody used.
The two definitions, side by side
Return on capital employed is operating profit divided by capital employed, where capital employed is total assets minus current liabilities.
Take a manufacturer earning ₹960 crore of operating profit on ₹4,300 crore of capital employed. It returns 22.3%.
Return on invested capital is net operating profit after tax divided by invested capital. Invested capital is the money actually at work in operations: net fixed assets plus operating working capital, or equivalently total funding minus surplus cash and non-operating investments.
Now tax that same company at a 25% marginal rate and it earns ₹720 crore of net operating profit after tax. If ₹700 crore of its capital employed is surplus cash and investments in an associate, its invested capital is ₹3,600 crore and it returns 20.0%.
Both numbers are correct. They measure different things.
Advertisement
What each one is actually asking
Return on capital employed asks: what does the whole long-term funding base earn, before tax?
Return on invested capital asks: what does the operating business earn on the money tied up in it, after tax?
The first is easier to compute and harder to compare across companies, because two firms with different cash balances and different tax positions will differ for reasons that have nothing to do with how well they run their operations. The second requires judgement about what counts as surplus and what counts as operating, and it rewards that judgement with a number you can set directly against the cost of capital.
The tax question, which decides more than people expect
Return on capital employed is conventionally computed before tax. Return on invested capital is computed after it.
That single difference is enough to make the two ratios diverge by five or six percentage points at a normally taxed Indian company. Quote one against the other without saying which is which, and you get arguments that are really about definitions.
Taxing operating profit at the marginal rate, rather than the reported effective rate, is deliberate. For a leveraged company the reported rate is lower because interest is deductible, and building that saving into the numerator while also using an after-tax cost of debt in the discount rate counts the same benefit twice. Tax the operations. Finance the company separately.
The surplus cash question
An Indian company with ₹1,400 crore in liquid mutual funds earning treasury rates has ₹1,400 crore that isn't part of its business.
Leave it in the denominator and the return falls for a reason that says nothing about the operations. Take it out and the ratio measures the business, which is what you wanted.
Every business needs some working balance, though, so a fair version excludes only what is surplus beyond it. A rough and defensible rule is to treat cash above two or three per cent of revenue as surplus for a manufacturer, and to say in the note that you did. Consistency across the peer set matters far more than getting the threshold right, because the comparison is the point.
Goodwill, and the two legitimate questions
Whether to include goodwill in the denominator is the other judgement, and both answers are defensible because they answer different questions.
Include it and you're measuring what the shareholders' money earned, counting what was paid for acquisitions. Exclude it and you're measuring what the operating business earns, which is the right question when asking whether the business itself is any good.
A serial acquirer will look very different under the two treatments. A company earning 25% on capital excluding goodwill and 9% including it has a good business and has paid too much to assemble it, and that sentence is a more useful description than either ratio alone.
Compute both. Say which one you're quoting.
Why the comparison to the cost of capital is the whole point
Neither number means anything on its own.
A company earning 20% on invested capital against a weighted average cost of capital of 11.5% runs an 8.5-point spread, and every rupee it reinvests turns into more than a rupee of value. Reverse the signs and a company earning 9% against the same 11.5% destroys value with every rupee it retains, and one in that position growing quickly is destroying value faster.
Compressed, that comparison is the theory of value creation, and the rest of valuation is commentary on it.
It also explains why growth isn't automatically good. Growth multiplies whatever the spread is. Applied to a positive spread it compounds value. Applied to a negative one it compounds the problem, and the fastest way for such a company to help its owners is to stop growing and return the money.
Return on capital employed is the more useful ratio in one specific case
Indian filings don't print net operating profit after tax and rarely print operating profit either. Building return on invested capital properly means assembling both sides from the notes and deciding what is surplus. You also have to decide about goodwill, and do it all consistently for five years and for six peers.
That's a real afternoon of work. Return on capital employed can be built in ten minutes from the face of the statements.
So which should you use? For most work, screen on return on capital employed, computed the same way every time, and build return on invested capital properly for the handful of companies that survive the screen. Your screen needs to be consistent more than it needs to be precise.
The version that predicts, rather than describes
Both ratios describe the whole capital base, most of which was invested by people who may no longer work there.
The incremental version measures what management is doing now. Take the change in net operating profit after tax and divide it by the change in invested capital over the same period. A company whose after-tax operating profit rose from ₹600 crore to ₹720 crore while invested capital rose from ₹3,050 crore to ₹3,600 crore earned 21.8% on the new money.
That figure is far more informative than the headline ratio. A business with a historic return of 25% whose incremental return is 8% is drifting toward a lower average, and it will get there regardless of what today's number says.
Compute it over three-year windows, since any single year mixes capital that hasn't yet started earning with profit from capital invested years ago. Then look at the direction across a decade. It's the most predictive series available from published accounts.
The traps that produce wrong answers
Leases come first. After the accounting change, a lease creates a right-of-use asset and a lease liability, both of which belong in the capital base. For a retailer or an airline, or for a hotel company, any returns calculation that ignores them is fiction. Comparing a pre-transition year with a post-transition one without adjusting produces a break in the series that looks like a change in performance.
Capital work in progress is the second. Money spent on a plant that isn't yet commissioned is counted in the denominator while earning nothing. A company with ₹2,400 crore of gross block and ₹960 crore under construction has a reported return that is understated relative to what the assets will produce, and a depreciation charge that is already committed. Both facts belong in the note beside the ratio.
Then there's year-end against average capital. A company that made a large acquisition in March has a year-end capital base that never earned a rupee of the year's profit. Use the average of opening and closing where the balance moved a lot, and say that you did.
Screener data is the last trap. Providers differ on all of the above and mix standalone with consolidated. They rarely disclose their definitions. A five-year series built by hand from the filings is worth more than a twenty-year series copied from a website.
What to actually do
For a first pass, compute return on capital employed from the face of the statements. Do it before tax and on average capital employed. Cover five years and five peers, and compute every one identically.
For anything you might buy, build return on invested capital properly. Tax operating profit at the marginal rate and take invested capital net of surplus cash. Run it both with and without goodwill over five years. Then add the incremental figure across three-year windows.
Then set the answer against a weighted average cost of capital you built rather than one you found, and look at the spread.
If the spread is positive and the incremental return is holding, the next question is what stops a competitor from taking it. That's the moat question, and it's the one the ratio can't answer.
Working the two ratios on a real Indian filing
Neither number appears in an Indian annual report. Both have to be assembled, and the assembly is where the answer is decided.
Operating profit isn't printed. Start from profit before tax, add back finance costs, then decide what to do with other income. Interest earned on surplus cash isn't operating income for a manufacturer, and leaving it in flatters both ratios. A company reporting ₹840 crore of pre-tax profit of which ₹210 crore is treasury income has an operating business earning ₹630 crore before interest. That ₹210 crore belongs with the cash that produced it, at a completely different multiple.
Capital employed comes from the balance sheet as total assets less current liabilities, with short-term borrowings added back because they are financing.
Invested capital takes more judgement. It's net fixed assets plus operating working capital, with surplus cash and non-operating investments removed. Then you still have to decide about goodwill.
Do it once by hand for one company and the definitions stop being abstract.
The four traps, in the order they bite
Leases bite first. After the accounting change, a lease creates a right-of-use asset and a matching liability, both of which belong in the capital base. For a retailer or an airline, and for a hotel group too, a returns calculation that ignores them is fiction. Ratios spanning the transition year can't be compared without an adjustment for the same reason. Pull a five-year series from a screener and it often has a break there that looks like a change in performance.
Capital work in progress bites next. Money spent on a plant that hasn't been commissioned stays in the denominator, earning nothing and carrying no depreciation. A company with ₹2,400 crore of gross block and ₹960 crore under construction has a reported return that is understated relative to what the assets will produce, and a depreciation charge that is already committed.
Year-end against average capital comes third. A company that acquired a business in March has a closing capital base that never earned a rupee of the year's profit.
Screener data comes last. Providers differ on every one of the above and mix standalone with consolidated. They rarely publish their definitions.
Five years, built by hand, beats twenty years copied.
The sentence the two ratios exist to produce
At the end of the work you should be able to write one line. This business earns X on the capital in it against a cost of capital of Y, and the reason the gap hasn't closed is Z.
The first number is arithmetic. The second is a model. The third is the investment case, and it's the only part a competitor can't copy from the filings.
The comparison that decides everything else
Set the return against the cost of capital and you have the only statement in valuation that isn't an opinion about the future. A business earning 20% on the capital in it against a cost of 11.5% is creating value at a rate of 8.5 points a year on every rupee it holds. If it can reinvest at the same spread, it compounds. A business earning 9% against the same 11.5% is destroying value on every rupee it retains, and the faster it grows the faster it destroys. Those two sentences cover most of what an investor needs to know about capital allocation, and they are computable from published accounts for any company in about an hour. What you can't compute is how long the spread lasts. That question is the moat question, and the only difficult judgement in the whole exercise.
Competition closes spreads. That's the base rate.
Durability is the claim you have to prove.
Everything else is arithmetic.