DuPont decomposition

Three companies each report a return on equity of 24%. On the screener they sit in a row, indistinguishable — three good businesses earning the same handsome return for their owners. But one of them earns that 24% by selling a branded product at a fat margin. The second earns exactly the same 24% by running a thin margin at enormous speed, turning its capital over many times a year. And the third earns it by taking a mediocre business and loading it with debt until the leverage magnifies an ordinary return into a striking one. Three identical numbers, three completely different machines — and one of them will fall apart in the first bad year while the other two hold. illustrative

The single ROE could never have told them apart, because a single number cannot show how it was made. That is what DuPont decomposition does. It takes the one return and splits it into the three levers that produce it — how much profit the business makes on each rupee of sales, how many rupees of sales it wrings from each rupee of assets, and how much it has borrowed against its equity. Written out, ROE is exactly net margin × asset turnover × leverage, and reading which of the three is doing the work is the single most revealing exercise in ratio analysis. It tells you not just how good the return is, but what kind of business you are actually holding, and where that return would break.

Return on equity is the number every owner cares about, and precisely because it is a single figure, it flattens away the thing that matters most: the route by which it was earned. Two firms at 24% are not the same investment if one earns it operationally and the other financially, because those two routes carry utterly different risks and utterly different futures. DuPont exists to un-flatten the number.

The decomposition is an identity — it is always true, by construction. Return on equity equals net profit over equity. Multiply that by sales over sales, and by assets over assets, and rearrange, and it becomes three fractions multiplied together: net profit over sales (the net margin, how profitable each sale is), sales over assets (the asset turnover, how hard the assets work), and assets over equity (the equity multiplier, how much leverage sits underneath). Nothing has been added; the same ROE has simply been broken into its parts. But those parts map onto the three ways a business can make money — by charging more per sale, by selling more per rupee of assets, or by borrowing — and each part answers a different question about durability. A return built on margin rests on pricing power and a moat. A return built on turnover rests on operational excellence and volume. A return built on the multiplier rests on debt, and debt magnifies in both directions: it flatters ROE in good times and destroys equity in bad ones. This module exists because the question is never just "how high is the ROE?" It is "which lever is pulling, and can that lever keep pulling through a downturn?" — and only the decomposition answers it.

Split the return, then read which lever earned it.

Read which lever is pulling. The whole value of the split is in seeing which of the three terms is large and which is small. A high margin with modest turnover is the brand signature — the business earns its return by charging more, so its return is only as durable as its pricing power. A thin margin with high turnover is the volume-operator signature — retail, distribution, commodities — where the return comes from recycling capital fast, so its durability rests on operational discipline and scale. And a high equity multiplier carrying an otherwise ordinary business is the leverage signature — the most fragile, because it has bought return with risk, and the risk shows up exactly when times turn hard. When you decompose an ROE, the first question is which lever dominates, and the second is whether that lever can keep pulling.

Watch which lever moves. The decomposition is even more useful across time than at a point. If ROE rises, split the rise: did margin improve (better pricing or mix), did turnover improve (assets working harder, better working capital), or did leverage rise (more debt)? The first two are earned; the third is borrowed. A rising ROE driven entirely by a rising equity multiplier is a company manufacturing a better-looking return by taking on risk, and it should be read as a caution, not a triumph. When margin and turnover are flat and ROE still climbs, leverage is the only lever left, and it has been pulled.

The extended split, when you need it. The three-term version is enough most of the time. When you want to go deeper, the margin term itself splits further — into the operating margin, the drag from interest (the interest burden), and the drag from tax (the tax burden) — giving a five-part decomposition. That extended form separates a genuinely more profitable operation from one that merely pays less tax this year or benefited from a one-off drop in interest. But start with the three levers; they answer the durability question on their own, and the five-part version is a refinement, not a replacement.

Every sector has a characteristic DuPont signature — a typical shape of margin, turnover and leverage — and knowing the shape tells you at a glance whether a company is normal for its industry or doing something unusual. The most important case is the one where the signature inverts the manufacturing instinct entirely.

The inversion to hold onto is the bank. For a manufacturer, an equity multiplier of six would be a red flag — debt doing the work that operations should. For a bank, a multiplier of twelve is not a warning at all; it is the entire business model, because a bank is funded by deposits and lends a multiple of its equity by design. So a bank's DuPont ROE is dominated by leverage, and reading that leverage as danger — as you rightly would in a manufacturer — gets the bank exactly wrong. This is the same lesson the return-ratios module reached from another angle: the tool must be read against the business. DuPont does not tell you whether leverage is good or bad in the abstract; it tells you how much of the return leans on it, and you decide whether that lean is the model or the risk, by knowing what business you are looking at. For the cyclical, the caution is different: the margin lever is not stable, so any one year's decomposition is a reading of where the cycle is, not of what the business structurally is.

A composite consumer-durables maker has grown its ROE from 16% to 25% over three years, and the market has re-rated it as a quality compounder. Decompose the rise before you agree. illustrative

Split the starting point and the endpoint into the three levers. Net margin: 8% then, 8% now — flat. Asset turnover: 1.1 then, 1.1 now — flat. Equity multiplier: 1.8 then, 2.8 now — up sharply. So the entire nine-point rise in ROE came from leverage. The company borrowed to fund a mix of expansion and a large share buyback, and the extra debt magnified an unchanged operating return into a much higher one. The operations did not improve at all; the balance sheet simply got riskier. The "quality compounder" story, which imagines a business getting structurally better, is not what the numbers say. What the numbers say is that an ordinary-and-unchanged business has been financially engineered to show a better ROE.

That does not automatically make it a bad company — leverage used to fund genuinely high-returning expansion can be sensible, and a buyback at a low price can create value. But it completely changes what you are betting on. You are no longer betting on a widening moat or improving operations; you are betting that the business can service materially more debt through whatever its end-markets throw at it, and consumer durables are cyclical. The DuPont split turned a comfortable "returns are improving" into a sharper, more honest question: can this now-more-leveraged business carry its debt through a downturn? The single rising ROE hid that question completely; the decomposition put it front and centre. The habit is to never accept a rising ROE as improvement until you have seen which lever raised it.

Set the three levers — net margin, asset turnover and the equity multiplier — and watch the same return on equity assemble three different ways. Push the leverage lever up and the ROE rises, but so does the fragility: a return built on debt is the one a downturn breaks first, while the same ROE earned through margin or turnover is operationally sound.

Do the arithmetic once, by hand, so the identity stops being abstract. Here are two years of a composite manufacturer's figures — just five lines from the statements — and the whole decomposition falls out of them. illustrative

Work down the table. Net margin: ₹80 on ₹1,000 is 8.0%; ₹104 on ₹1,300 is 8.0% — unchanged. Asset turnover: ₹1,000 over ₹900 is 1.11; ₹1,300 over ₹1,180 is 1.10 — effectively flat. Equity multiplier: ₹900 over ₹500 is 1.80; ₹1,180 over ₹430 is 2.74 — up sharply. Multiply the three back together to confirm the identity holds: Year 1, 0.08 × 1.11 × 1.80 = 0.160, i.e. 16.0%; Year 2, 0.08 × 1.10 × 2.74 = 0.241, i.e. 24.2%. The arithmetic ties out, which is the point of an identity — it is always true by construction.

Now read what moved. ROE jumped eight points, from 16.0% to 24.2%, and two of the three levers did not budge — margin held at 8.0%, turnover held at 1.1. The entire gain came from the equity multiplier rising from 1.80 to 2.74. And look at why it rose: assets grew from ₹900 to ₹1,180 (funded by new debt) while equity actually fell from ₹500 to ₹430 — the fingerprint of a share buyback. So the "improving return" is neither better pricing nor better efficiency; it is more debt and a smaller equity base, exactly the leverage signature. Five lines from the statements, three divisions, and one multiplication told you the whole story — and told you to ask the downturn question, not to celebrate.

DuPont tells you which lever is pulling, but not whether that lever will keep pulling. It can show that a return rests on a fat margin, but not whether the pricing power behind that margin is a durable moat or a temporary gift about to be competed away. It can show that a return rests on high turnover, but not whether the operational edge producing that turnover is defensible. The decomposition locates the source of the return; judging the durability of that source is a separate enquiry into competitive position that the three numbers cannot settle.

Nor can it tell you, on its own, whether leverage is prudent or reckless. A high equity multiplier is the model for a bank and a warning for a manufacturer, and DuPont is silent on which it is looking at — you supply that judgement by knowing the business and by asking whether the cash flows are steady enough to carry the debt through bad times. The decomposition quantifies how much the return leans on leverage; it does not tell you whether that lean is safe. That depends on the stability of the operating cash flows, which is the leverage module's subject.

And, like every ratio, DuPont inherits the honesty of its inputs. If the margin is flattered by a one-off gain, or the asset base has been shrunk by write-offs or kept off the books through leases, the levers will point in misleading directions. A turnover that looks impressive because assets were written down is not real efficiency; a margin that looks stable because a mix shift is hidden inside it is not real pricing power. The decomposition is only as truthful as the margin, the assets and the equity that go into it, and whether those are truthful is a question for the forensics part, not for DuPont.

How it comes up. When ROE improves, a good analyst asks management to attribute the gain to the three levers rather than celebrate the headline. The question sounds like this: "ROE went from 18% to 25%. Can you break that into margin, asset turnover and leverage for us? How much of the improvement is operating and how much is the higher debt and the buyback?" The analyst is forcing the decomposition into the open.

A good answer, verbatim-style.

"Roughly half the improvement is operating — margin is up about 150 basis points on mix, and asset turnover improved as the new lines filled. The other half is leverage: we raised debt to fund the expansion and bought back about 4% of equity, so the multiplier went up. We're comfortable with the debt because our cash flows are stable, but I'd be clear that not all of the ROE gain is operational."

That answer splits the rise into the three levers, quantifies each, and is honest that part of the gain is financial rather than operating.

An evasive answer, verbatim-style.

"We're delighted with the improvement in ROE to 25%, which reflects the strength of our franchise and our disciplined execution. It's a testament to the quality of the business and gives us confidence in delivering superior returns going forward."

Notice what it does. It attributes a leverage-driven gain to "franchise" and "quality," which are margin-and-moat words, and never mentions the debt or the buyback that actually did the work. It is selling a financial-engineering result as an operating one.

The follow-up nobody asks, and what its absence means. "Give us the DuPont walk — how many basis points of the ROE gain came from margin, from turnover, and from the equity multiplier?" That forces the exact attribution. If the room lets "improvement reflects our franchise strength" stand without the split, either the gain is mostly leverage and management would rather call it quality, or the analysts have stopped decomposing and are taking the headline. A rising ROE that management attributes to "quality" but will not split into levers is almost always a leverage story wearing a moat's clothes.

1. Treating equal ROEs as equal businesses. The same return can be margin-led, turnover-led or leverage-led, and those are three different investments with three different risks. The decomposition, not the headline, tells them apart.

2. Cheering a rising ROE without splitting the rise. If margin and turnover are flat and ROE still climbs, leverage did it. A return improved by borrowing is riskier, not better, and should be read as a caution.

3. Reading leverage the same way in every business. A high equity multiplier is a warning in a manufacturer and the normal model in a bank. DuPont quantifies the lean on leverage; the business decides whether that lean is danger or design.

4. Mistaking written-down assets for real efficiency. An asset turnover that looks high because the asset base was shrunk by write-offs is not operational excellence. Check what happened to the denominator before crediting the turnover.

5. Reading a cyclical's decomposition as a signature. The margin lever swings with the cycle, so a single-year DuPont on a cyclical is a snapshot of where the cycle is, not of what the business structurally is. Read it across the cycle.

6. Stopping at the three levers when a one-off is hiding in the margin. A margin flattered by a one-time gain, or by a temporary drop in tax or interest, will mislead the split. When precision matters, extend the decomposition to separate operating margin from the interest and tax burdens.