Returns in Corporate Finance: The Measures and What Each Assumes

A return on capital is a profit figure divided by the capital that produced it, and the answer depends entirely on which profit and which capital are picked up. On Sankalp Industrial Systems Limited, invented, Year 0 gives 15.00 per cent on invested capital, 15.33 on equity, 15.38 on capital employed and 10.14 on the assets the record identifies. Four measures, one company, four answers.

Underneath that sits a rule which sounds obvious and is broken constantly: a numerator and its base have to answer to the same people. If the denominator holds all the capital in the business, the numerator has to be the profit earned before any provider of that capital was paid. If the denominator holds only the owners' capital, the numerator has to be what is left after the lenders and the minority have been paid. Break that pairing and the answer is still a number; it is no longer a return on anything. Invested capital is the denominator that matches the operating profit a business actually earns, and for exactly that reason Koller, Goedhart and Wessels build their whole value driver frame on it.

What is a return on capital actually measuring?

A vegetable cart comes before a company. Somebody puts Rs 20,000/- into stock, a weighing scale and a float of change, works the season, and finishes Rs 3,000/- ahead. The return is 15 per cent, and every part of that sentence is checkable: the Rs 20,000/- can be pointed at, the Rs 3,000/- can be pointed at, and one person stands behind both ends of it. Now suppose the cart was half funded by a cousin who lent the money and takes Rs 600/- of interest a year. Is the return still 15 per cent? The answer depends on which question is being asked. On everything tied up in the cart, the cart earned Rs 3,000/-. On what the vendor personally put in, the vendor kept Rs 2,400/- on Rs 10,000/-, a return of 24 per cent. Neither figure is wrong. Each answers a different question, and the one mistake on offer is to quote one of them while meaning the other.

Every return measure in corporate finance is that same division with different things loaded into the two halves. The structure never changes: some measure of profit for a period, over some measure of capital that was standing there to produce it. Between the four measures below, nothing about the arithmetic changes and everything about the contents does. One counts only the capital in the operating business. One counts only what the owners put in and left in. One counts every rupee funding the company, from owners, minority holders and lenders together. One counts the whole asset base including things that produce no profit at all. Four sets of contents, four answers, and whoever cannot see which contents went in cannot read any of the four.

The two halves are also measured differently in kind, and the difference matters. The numerator is a flow: it happened over a period, twelve months of trading. The denominator is a stock: it is a photograph of the balance sheet on one day. Dividing a flow by a stock is standard and useful. The division leaves one question nobody usually asks out loud: which day the photograph was taken. The choice of day moves the answer by more than a point, and the opening, closing and average bases are set out below.

Whose money is in the denominator, and whose profit is in the numerator?

One discipline decides everything else. Choose the denominator first, and the numerator is not a matter of preference: it is settled by who has a claim on the capital just placed in the base. Work it through on Sankalp Industrial Systems Limited, invented, and it becomes mechanical.

Suppose the denominator is all the capital in the operating business, Rs 12,00,00,00,000. The owners and the lenders funded that base together. So the numerator has to be a profit measured before either of them was paid anything. Earnings before interest and taxThe profit a business makes from trading, struck before any interest is charged and before any tax is taken. It sits above the financing lines in the profit and loss account. of Rs 2,40,00,00,000 is such a figure. So is that same figure taxed, Rs 1,80,00,00,000: tax is a claim of the state rather than of a capital provider. Neither figure has interest deducted from it. The lenders' money is still sitting in the base, so their interest cannot come off a profit measured over it.

Now suppose the denominator is only the owners' book capital, Rs 9,00,00,00,000. The lenders' money is not in that base and neither is the quarter of the subsidiary the group does not hold. So the numerator has to be a profit struck after the lenders have taken their interest and after the minority has taken its share: Rs 1,38,00,00,000. Anything earlier in the account still contains money that belongs to somebody whose capital was just excluded.

The two numerators tie together exactly, and running that line is the cleanest way to prove the pairing is not arbitrary. Start at operating profit after tax of Rs 1,80,00,00,000. Take out the interest the lenders received, after the tax relief on it, Rs 36,00,00,000. Take out the Rs 6,00,00,000 that belongs to the minority in Sankalp Coatings Private Limited, invented. The figure left is Rs 1,38,00,00,000, to the rupee, with nothing over. The value view of this company and the reported profit view are one business looked at from two ends, and the two numerators are two stopping points on a single walk down the account.

What is invested capital, and which profit belongs over it?

Invested capital is the narrowest of the four bases and the most deliberately built. For Sankalp Industrial Systems Limited, invented, it is Rs 12,00,00,00,000 at Year 0, and it is put together from two lines and no others: net working capitalReceivables plus inventory less payables. Money a business leaves standing in stock and in bills customers have not settled, net of bills it has not settled itself. of Rs 1,80,00,00,000 and net fixed assetsLand, buildings, plant and equipment carried at cost less the depreciation charged against them since they were bought. of Rs 10,20,00,00,000. Nothing else is in there. No cash, no surplus land, no holding in another business.

Put the operating profit after tax of Rs 1,80,00,00,000 over it and the answer is exactly 15.00 per cent. The two halves match because both were drawn from the operating business alone: the numerator is struck before interest, so the lenders are still inside it, and the denominator holds the capital the lenders and the owners jointly funded. Nothing in the numerator was earned by an asset that is missing from the denominator, and nothing in the denominator was funded by a claimant whose payment has already been taken out above.

The pairing just described is the whole test. Every return measure, anywhere, runs in both directions. Is there anything in the profit that the base does not contain? Is there anything in the base whose earnings never reach the profit? On this measure, for this company, the answer to both is no, and the return on invested capital comes out to a round 15.00.

What did the owners earn on the capital the owners left in?

Now change the question. Forget the business as a machine and ask what the people who put up the share capital got back on it. The base is the book value of equityThe owners' stake as the balance sheet carries it: money subscribed, plus every rupee of profit retained since, less anything paid back out., Rs 9,00,00,00,000, being Rs 45.00 a share on 20,00,00,000 shares. And the numerator has to be the profit that actually belongs to those owners. On a consolidated company that is not the same as the profit for the year.

Walk it down. Start at Rs 2,40,00,00,000 of earnings before interest and tax. Interest of Rs 48,00,00,000 comes off, leaving profit before tax of Rs 1,92,00,00,000. Tax at the company's own assumed effective rate of 25.0 per cent, invented, is Rs 48,00,00,000, so profit after tax is Rs 1,44,00,00,000. But Sankalp Coatings Private Limited, invented, is 75.0 per cent held and fully consolidated. The whole of its profit came up into that Rs 1,44,00,00,000 even though a quarter of the business behind it belongs to somebody else. The minority's slice of that profit is Rs 6,00,00,000. Take it out and profit attributable to the owners is Rs 1,38,00,00,000.

Put that Rs 1,38,00,00,000 over the Rs 9,00,00,00,000 of book equity and the answer is 15.33 per cent. Use the Rs 1,44,00,00,000 instead and the answer reads 16.00 per cent, crediting the owners with a quarter of a subsidiary they do not hold. The minority interestThe slice of a subsidiary that a group consolidates in full but does not hold. It appears in the accounts because consolidation brings in every rupee of the subsidiary, not just the part the group paid for. line is small, sits low in the account, and is the single easiest thing here to skip. The minority line moves the answer by two thirds of a point, and two thirds of a point is wider than most differences anybody bothers to report.

Why is return on capital employed stated before tax?

The third measure widens the base again. Capital employed is every rupee funding the company, whoever supplied it: the owners' Rs 9,00,00,00,000, the minority's Rs 60,00,00,000 and gross borrowings of Rs 6,00,00,00,000, adding up to Rs 15,60,00,00,000. Interest belongs to one of the three claimants sitting in that base, so the numerator that matches it is earnings before interest and tax, Rs 2,40,00,00,000. Divide the first of those figures by the second and the answer is 15.38 per cent.

Two things about that figure need saying in the same breath as the figure itself. The first is that it is struck before tax on both sides, so it cannot be set beside the two after-tax measures above it without a translation neither of them supplies. A pre-tax return and an after-tax return on the same company are separated by the whole tax charge, and on this company that is a 25.0 per cent bite. The second is arithmetic housekeeping: this 15.38 per cent is derived here from locked components rather than lifted from anywhere, and the components are the three lines just named.

Notice how close 15.38 sits to the 15.00 per cent on invested capital. The gap looks like 38 basis points of something interesting. It is not. Those two figures are separated by a tax charge on one side and by Rs 3,60,00,00,000 of extra denominator on the other, and the two effects happen to push in opposite directions by almost the same amount. They land near each other because of how the invented figures fall, and for no reason connected to how this business performs. Two return measures on a real company sitting a third of a point apart are a prompt to ask what each one contains, not a finding.

What is inside the widest base, and what does it cost the answer?

The fourth measure puts profit after tax of Rs 1,44,00,00,000 over the asset side of the balance sheet. On this company that comes to 10.14 per cent, nearly five points below the return on invested capital, and it is worth being blunt about why: almost none of that gap is about performance. The gap is what happens when a denominator widens without the numerator widening to match.

Two separate things drag it down and they compound. The first is that Rs 2,20,00,00,000 of the base is cash and non-operating assetsSomething a business holds that plays no part in its trading. Whatever it earns turns up below the operating profit line, if it turns up at all., neither of which produced a single rupee of the operating profit. The second is subtler and catches careful people. Look at the numerator: Rs 48,00,00,000 of interest has already come out of it, paid to lenders who funded part of the very asset base sitting underneath. So the numerator has been charged for the lenders while the denominator still counts their money. The crossed pairing is precisely what the matching rule forbids, and it is baked into this measure by construction rather than by anybody's error.

None of which makes the measure useless. Return on assets is the fastest measure to compute from a published balance sheet, it needs no judgement about what counts as operating, and for a business whose balance sheet is almost entirely operating assets it lands close to the invested capital answer. The measure is simply the wrong tool when the question is how hard the trading business works its capital, and on a company sitting on spare land plus a stake in another business it is the wrong tool by nearly five points.

What do the four look like side by side on one company?

Everything above has been one measure at a time. Here they are together, on the same invented company, on the same day, with every input named. Nothing in this table is a different business, a different period or a different accounting policy: the only thing that changes down the rows is what was loaded into the two halves.

The four measures differ in what each half contains, and they do not lie along a continuum. No single dial turns one of them into another, so the only way to read any of the four is to read what was loaded into its two halves.

What does each denominator include, and what does each leave out?

The four bases are not four unrelated ideas. Three of them are slices of the same funding, cut at different depths, and once the slices are drawn it becomes obvious which claimants each measure is answering to. Start from the widest: capital employedAll the capital funding a business at once, whoever supplied it, taken from the funding side of the balance sheet rather than from the asset side. of Rs 15,60,00,00,000 holds the owners, the minority and the lenders. Taking the minority and the lenders out leaves the owners' Rs 9,00,00,00,000 alone. Invested capital of Rs 12,00,00,00,000 is cut differently again: it is not a slice of the funding side at all but a measure of what is standing in the operating business. Cash and the non-operating assets are missing from it for that reason, even though somebody funded those too.

The exclusions are where the real work happens, and they run in both directions. Invested capital leaves out cash because cash is not being worked; it leaves out the surplus land because nobody is making valves on it; it leaves out the associate holding because whatever that earns arrives below the operating profit line, if at all. Capital employed leaves out none of the funding but says nothing about whether that funding is being used. And the asset base leaves out nothing at all, and leaving out nothing is exactly its problem.

Which part of the balance sheet earns none of this profit?

The record identifies Rs 14,20,00,00,000 of assets for this company: the Rs 12,00,00,00,000 of invested capital, plus Rs 1,20,00,00,000 of cash and cash equivalents, plus Rs 1,00,00,00,000 of non-operating assets. That last line is a surplus land parcel carried at Rs 45,00,00,000 and a 26.0 per cent holding in Aruna Tooling Private Limited, invented, carried at Rs 55,00,00,000. Both are real assets. Neither sits inside the trading business.

The Rs 1,20,00,00,000 of cash sits outside the Rs 2,88,00,00,000 of earnings before interest, tax, depreciation and amortisation (EBITDA) entirely, and so do the Rs 1,00,00,00,000 of non-operating assets, so any measure counting either one in its base understates the operating return by construction. The associate is equity accountedA way of carrying a holding in a business a company influences but does not control. One line brings in the share of that business's profit, and it arrives below the operating profit line rather than inside it., so its earnings never enter EBITDA at all. The land earns nothing and was never expected to. The cash may earn a little somewhere below the operating line, and the record does not say it does.

Should the base be the opening, the closing or the average capital?

Here is the convention almost nobody states and everybody uses. The numerator covers a period; the base is a photograph of one day. So which day? Sankalp Industrial Systems Limited, invented, makes this concrete because its capital moves during the year: invested capital opens Year 1 at Rs 12,00,00,00,000 and closes at Rs 13,00,00,00,000, and the Rs 1,00,00,00,000 gap is net new capital going in during the twelve months. The numerator for that year is Rs 1,98,00,00,000 of operating profit after tax.

On the opening base the answer is 16.50 per cent. On the closing base it is 15.23 per cent. On the average base of Rs 12,50,00,00,000 it is 15.84 per cent. Three answers, one year, one company, a spread of 1.27 points, and what separates the three of them is a conventionA choice about how a figure is put together that is settled by habit rather than by arithmetic, and that different houses settle differently. nobody usually writes down.

Each has an argument behind it. The opening base says the capital standing there at the start of the year is what had the chance to earn the profit, and the money put in during the year had barely any time to work. The closing base says the reader wants the most recent picture of what is tied up. The average splits the difference and is the most common, and its own weakness is that a company which invests everything in the last week of the year gets charged for half of it. None of the three is wrong. What is wrong is comparing a figure built one way against a figure built another, which is the failure block further down.

What changes when the denominator is a share price?

Every base so far came off the balance sheet. There is a fourth kind of denominator that comes off the market instead, and it produces a figure people put in the same column as a return on equity without noticing that it answers a completely different question.

Sankalp's book equity is Rs 9,00,00,00,000, or Rs 45.00 a share on 20,00,00,000 shares. The share price is Rs 90.00, so the market capitalisationWhat the whole of the equity is priced at today: every share on issue valued at whatever one of them last traded for. A price, not a contribution. is Rs 18,00,00,00,000 and the price to book ratio is exactly 2.00 times. Now take that same Rs 1,38,00,00,000 belonging to the owners and divide it by the market value rather than the book value. The answer is 7.67 per cent. Read it the other way and the same fact appears: Rs 1,38,00,00,000 on 20,00,00,000 shares is Rs 6.90 a share, and Rs 6.90 against a price of Rs 90.00 is 7.67 per cent.

That figure is not a return on equity, and the two are exactly a factor of two apart here only because the price happens to be twice the book. A return on equity asks what the business earned on the capital the owners actually put in and left in. The market version asks what the earnings are as a share of what somebody would have to pay to buy them today. The first is a fact about the company. The second is a fact about the company and the price together, and it moves every time the price moves without anything happening inside the business at all. It has its own name, an earnings yield, and it belongs in its own column.

What does the record lock here, and what does it not?

One of the four figures here deserves a warning label. The record behind this invented company locks invested capital, book equity, minority interest, gross debt, cash and non-operating assets, and it does not lock a complete balance sheet. No total was ever set, so the locked items do not foot to one.

So when Rs 1,44,00,00,000 is divided by Rs 14,20,00,00,000, that denominator is assembled rather than looked up. It is the Rs 12,00,00,00,000 of invested capital, with the Rs 1,20,00,00,000 of cash added to it and the Rs 1,00,00,00,000 of non-operating assets added after that, being exactly the assets the record identifies. A published return on assets would use the total assets line off the face of the balance sheet. The total assets line picks up whatever else is sitting there, and the answer would land somewhere else. The 10.14 per cent is a fair illustration of what happens when a base widens; it is not a claim about a total nobody stated.

Saying that out loud is not pedantry but the discipline the whole comparison rests on. A reader who does not know how a denominator was put together cannot tell whether two figures are comparable. The number itself never carries that information: 10.14 per cent looks exactly as authoritative when it was assembled from three lines as it would if it came off a filed statement. The only thing that can carry it is a sentence beside the figure, written by whoever did the assembling.

How do a lender, an analyst and an operator each read these four?

The four measures are not in competition for a prize. They survive side by side because different people are asking different questions, and each question makes one of them the natural instrument.

A lender is asking whether the business generates enough on everything standing in it to service the borrowing, and cares about the whole capital base rather than the owners' slice. A credit team reading Sankalp would work near the capital employed end of the range. The money it lent is inside that base, and the profit it is looking at is struck before its own interest was paid. The same credit team would not take a return on equity at face value. A company can lift its return on equity simply by borrowing more, and borrowing more is the opposite of what a lender wants to see rewarded.

An analyst building a view of the business itself lives on the invested capital measure. On that measure alone both halves come out of the trading operation, and neither half is contaminated by how the company happens to be funded or by what it happens to be sitting on. It is also the one that takes the most work: somebody has to decide what is operating and what is not, and that decision is a judgement rather than a lookup.

An owner of shares, or somebody studying one, tends to reach for return on equity because it is the measure written in their own currency. Return on equity answers what the business earned on their money. The measure also has the least stable base of the four. Retained profit, dividends and any buyback all move book equity directly, and return on equity moves with them even in a year when nothing about the operations changed.

And the operator inside the business, the person deciding whether to add the third valve line, is asking a narrower question again: what a specific new rupee would earn, rather than what the whole standing base is earning. That question is a different one, and it is covered separately. What all four readers share is the discipline that comes first: write down the two halves before quoting the answer.

Which measure should be used, and what does the choice depend on?

The choice depends on the question, and it is worth being blunt about which question each measure actually answers. If the question is how hard the trading operation works the capital standing inside it, invested capital is the base, and it is the only one that excludes assets producing none of the profit. If the question is what the owners earned on their own money, equity is the base, and the minority line has to come out of the numerator. If the question is what the whole funding pool produced before the state and the lenders took their shares, capital employed is the base, and the answer is pre-tax and has to be labelled so. If the question is a quick comparable figure off a published balance sheet with no judgement calls, the asset base does it, at the cost of counting things that earn nothing.

Two secondary decisions ride along with the main one and both have to be stated. The first is the base date: opening, closing or average, applied the same way on both sides of any comparison. The second is book against market, and there the answer is not a preference at all: a market denominator produces a different measure with a different name rather than a variant of the same one.

One rule survives every choice, and it is the rule stated at the outset: a numerator has to answer to the same claimants its base does, and a comparison only exists when both figures were assembled the same way. Everything else here is a consequence of that rule applied to a different base. None of these figures is called good, or strong, or disappointing. Judging a return means setting it against what the capital costs, and that rate has not been built yet. The cost of capital is covered separately: what gets measured is built before the yardstick that measures it.