Return ratios

You look up a company on a screener and it shows a return on capital employed of 30%. You look up the same company's accounts, do the sum yourself, and get 16%. Neither of you has made an arithmetic mistake. You have simply divided a different profit by a different capital. The screener took a broad profit figure and divided it by a narrow capital base; you took operating profit and divided it by everything the business has employed, including a large pile of idle cash. Same company, same accounts, two returns almost twice apart. illustrative

This is the trouble with return ratios: they all have the same shape — a profit, divided by a capital — and that shared shape hides how much choice sits inside each one. Which profit? Operating, after-tax, net? Which capital? Equity, equity-plus-debt, total assets, only the capital actually working? Every combination has a name — ROE, ROCE, ROIC, ROA — and every combination answers a different question for a different provider of money. Get the pairing wrong, or leave the wrong things in the denominator, and you compute a number that looks authoritative and means nothing. This module is about computing each return ratio so it actually answers the question it claims to, and about the exclusions that separate a real return from a screener's careless one.

A return ratio answers the most basic question an owner can ask: for the money tied up here, how much does the business earn? But "the money tied up here" depends on whose money you mean, and "how much it earns" has to be measured before or after the claims of the people you are excluding. That is why there are four common return ratios, not one, and why they are not interchangeable.

Return on equity (ROE) is the owner's own number: net profit — what is left after everyone else, including lenders and the taxman, has been paid — divided by shareholders' equity. It answers "what did the owners earn on their stake?" Return on capital employed (ROCE) steps back to the whole firm: operating profit (EBIT) divided by capital employed — equity plus debt — answering "what does the business earn on all the long-term capital in it, before deciding how that capital is split between owners and lenders?" Return on invested capital (ROIC) is the sharpest operating version: after-tax operating profit divided by the capital actually invested in operations, deliberately excluding surplus cash and non-operating investments, answering "what does the core business earn on the money genuinely at work in it?" And return on assets (ROA) divides net profit by total assets, a blunt measure used mainly for lenders like banks, where the balance sheet is the business.

The reason to keep them straight is that the numerator and the denominator must describe the same set of capital providers. ROE pairs net profit (owners' profit) with equity (owners' capital) — consistent. ROCE pairs operating profit (everyone's profit, before interest splits it) with equity-plus-debt (everyone's capital) — consistent. The classic error is mixing them: dividing operating profit by equity alone, or net profit by capital employed, which pairs a profit belonging to one group with a capital belonging to another and produces a number that answers no coherent question at all. This module exists because return ratios are the single most-quoted family in all of analysis, and the most casually miscomputed — by screeners, by brokers, and by readers who never learned that the profit and the capital have to match.

Four ratios, and a short set of rules that make each one honest.

Match the numerator to the denominator. This is the first discipline and the one most often broken. If the capital in the denominator includes debt (ROCE, ROIC), the profit on top must be before interest — operating profit — because that profit is what the whole capital base, lenders included, earned. If the denominator is equity alone (ROE), the profit must be after interest and tax — net profit — because that is what is left for the owners. Pair an operating profit with an equity denominator and you flatter the return wildly; the number is meaningless.

Strip the denominator to what is actually working. A capital-employed figure straight off the balance sheet often contains things that are not part of the operating machine: a surplus cash pile, a portfolio of investments, a big new plant still under construction (capital-work-in-progress) earning nothing yet. Each of these swells the denominator without adding to the operating profit on top, so it drags the ratio down and hides how well the real assets are doing. To see the operating return, exclude surplus cash, non-operating investments and CWIP from the capital base — and, on the top line, exclude the non-operating income those assets throw off, so both halves describe the same operations. The gap between the raw ROCE and this cleaned-up operating ROCE is often the most interesting fact about the company.

Use average capital, not year-end. A year's profit is earned across the whole year, but a balance sheet is a snapshot on the last day. If the company raised equity or took on debt mid-year, the year-end capital is bigger than what actually earned the profit, understating the return. Use the average of the opening and closing capital so the denominator matches the period the profit was earned over. Screeners frequently skip this, which is one reason their numbers drift from a careful hand calculation.

Compare the return with the cost of the capital. A return ratio only means something against the cost of the money it used. A business earning 14% on capital that costs it 11% is creating value; a business earning 14% on capital that costs 16% is destroying it, even though 14% sounds healthy. The gap between return and cost of capital — not the return alone — is what tells you whether the business is worth more or less for having grown. A high return that is still below the cost of capital is a value-destroying machine that merely looks impressive.

Return ratios are the family that breaks most dramatically across sectors — some are the wrong tool entirely, and some produce huge numbers that mean far less than they seem. Here are four businesses and which return ratio actually answers a question for each.

The pattern is that a return ratio assumes a clean split between operating capital and financing — and that assumption is exactly what some businesses do not honour. For a lender, borrowing is not financing but stock-in-trade, so ROCE has no meaning and ROA/ROE take over. For an asset-light firm, the denominator is so small that the return balloons into a number that flatters rather than informs. For a holding company, consolidation double-counts, so a group return ratio is not measuring one coherent thing at all. Only the manufacturer honours the ratio's assumptions cleanly, and even there the raw number needs the exclusions from the mechanics before it tells the truth. The lesson carries the previous modules forward: a return ratio is a question about capital productivity, and you must first ask whether this business has the kind of capital the ratio is built to measure.

A composite speciality manufacturer shows a screener ROCE of 15%, and the screen flags it as an ordinary, unexciting business. Recompute it by hand, applying the exclusions, and a different company appears. illustrative

Start with what the screener did: it took a broad profit and divided by the full capital employed off the balance sheet. But that balance sheet carries two things that are not part of the operating machine. First, a cash-and-investments pile equal to nearly a third of the capital base, built up from years of retained profit and earning only a low deposit yield. Second, a large new plant sitting in capital-work-in-progress, not yet commissioned, earning nothing. Both sit in the screener's denominator, dragging the ratio down. Strip them out — surplus cash, investments and CWIP from the denominator, and the small non-operating income they generate from the numerator — and the return on the capital actually operating is not 15% but closer to 26%. The core business is far better than the screen suggested; the raw ROCE was diluted by idle money and a plant mid-build.

Now the value question. The company's cost of capital is around 12%. A 26% operating return against a 12% cost means the core business is creating substantial value on every rupee it employs — and the CWIP that is currently dragging the reported number down is about to add more of that high-returning capacity once it commissions. So the honest read inverts the screener's: this is not an ordinary business but a strong one, temporarily disguised by a fat cash balance and a plant that has not switched on yet. The two facts that mattered — the idle cash and the CWIP — were both in the denominator, and neither was visible in the single number the screener printed. The habit to build is to never trust a screener's return ratio until you have taken its denominator apart.

Do the cleaning on paper, so the jump from a dull screener number to the real operating return is concrete rather than described. Here is the same composite manufacturer, reported the screener's way and then cleaned. illustrative

Work it through. The screener took the reported operating profit of ₹130 crore and divided by the full capital employed of ₹870 crore — 15%, unremarkable. But ₹410 crore of that capital is not operating: ₹290 crore is surplus cash and investments earning a low yield, and ₹120 crore is a plant still in capital-work-in-progress, earning nothing yet. Strip those out and the operating capital is ₹460 crore. On the top line, remove the ₹12 crore of non-operating income the cash throws off, leaving ₹118 crore of genuinely operating profit. The operating return is ₹118 ÷ ₹460 ≈ 26% — not 15%. The business is far better than the screen suggested; the raw number was diluted by idle money and a plant that has not switched on. And against a cost of capital of about 12%, a 26% operating return is creating real value on every rupee employed — a conclusion the screener's 15% completely hid.

A return ratio tells you how well capital is being used, but not how much of it the business can reinvest at that rate — and the second question often matters more for a long-term owner. A company earning 40% on capital but able to redeploy only a trickle of new capital each year compounds slowly; a company earning 18% but able to pour large sums back in at that same 18% compounds far faster. Return times reinvestment is what builds value over time, and the return ratio is only the first half of that product. A superb return with no runway to reinvest it is a good business that will not compound into a great investment.

Nor does a return ratio tell you whether the return will last. A high ROIC attracts competition, and competition is the force that pulls returns back toward the cost of capital. Whether a high return persists depends on the moat around it — a brand, a cost advantage, a network, a regulatory barrier — and none of that is inside the ratio. A high return with no moat is a magnet for the very competition that will erode it.

And a return ratio inherits every distortion in the two numbers that build it. If the operating profit on top is flattered by an accounting choice, or the capital underneath has been shrunk by write-offs or kept off the balance sheet through leases, the ratio computes cleanly and misleads completely. A return that looks high because the denominator was quietly hollowed out is the subject of a later module on how ratios get gamed. The ratio cannot tell you whether its own inputs were honest; that check is yours to make.

How it comes up. When management leads with a headline return, a sharp analyst asks how it was computed and what is in the denominator. The question sounds like this: "You've quoted a 22% ROCE. Is that on average capital or year-end, and does the denominator include the cash pile and the CWIP? What's the return on the operating assets alone, and where does it sit against your cost of capital?" The analyst is refusing the headline until the denominator is clean.

A good answer, verbatim-style.

"Good question. The 22% is on average capital employed including everything. If you strip out the surplus cash and the plant under construction, the return on operating capital is around 31%, well above our cost of capital of about 12%. The CWIP will lift reported ROCE as it commissions and ramps. So the operating business earns in the low thirties; the headline is diluted by cash and the build."

That answer names the basis, gives the cleaned-up operating return, and sets it against the cost of capital — everything you need to judge whether value is being created.

An evasive answer, verbatim-style.

"We're very proud of our best-in-class 22% ROCE, which reflects our capital discipline and operational excellence. We're confident of maintaining industry-leading returns as we continue to invest for growth."

Notice what is missing: the basis of the calculation, what is in the denominator, and any mention of the cost of capital. "Best-in-class" and "capital discipline" are adjectives standing in for the decomposition that would let you check the number.

The follow-up nobody asks, and what its absence means. "Give us ROCE on operating capital only — ex-cash, ex-CWIP, on average balances — and your estimate of the cost of capital it should beat." That forces the return onto a clean, comparable basis and against the hurdle that decides whether growth adds value. If the room lets "22%, best-in-class" stand without the operating figure, either the operating return is lower than the headline (because non-operating income is flattering the top line) or management does not compute it that way and is quoting the screener's number back at you. A return the company will not clean up is a return worth cleaning up yourself.

1. Mismatching the profit and the capital. If debt is in the denominator, the profit must be before interest (ROCE); if only equity is in the denominator, the profit must be after interest and tax (ROE). Pairing an operating profit with an equity base flatters the return into meaninglessness.

2. Leaving idle cash and CWIP in the denominator. Surplus cash, non-operating investments and a plant under construction swell capital employed while earning nothing, dragging the ratio down and hiding the operating machine's real return.

3. Using year-end instead of average capital. A snapshot on the last day misstates the capital that actually earned the year's profit, especially after a mid-year raise. Average the opening and closing balances.

4. Admiring a return without the cost of capital. A 14% return on capital costing 16% destroys value despite sounding healthy. The gap between return and cost, not the return alone, tells you whether growth is worth having.

5. Trusting a huge asset-light ROE. A 35–40% ROE on a near-zero equity base is arithmetic, not proof of excellence. Judge it on margins, growth and cash conversion.

6. Importing ROCE into a lender. For a bank or NBFC, borrowings are raw material and interest is core revenue, so ROCE has no clean base. Use ROA and ROE instead; the manufacturing ratio does not apply.