What the DuPont decomposition tells you that ROE alone does not
Return on equity answers one question: how many cents of profit does each dollar of book equity produce? The DuPont decomposition answers the far more useful follow-up: why. It splits ROE into three factors that multiply together — how profitable each sale is, how much sales the asset base generates, and how much of that asset base is funded by someone other than the shareholder.
The three-step form was developed inside the DuPont Company in the early twentieth century as an internal control tool, and it survives because each factor maps onto a different part of the business and a different management lever. Net profit margin is an income statement story about pricing, cost and tax. Asset turnover is an operating story about how hard the factories, inventory and receivables are working. The equity multiplier is a financing story that lives entirely on the right-hand side of the balance sheet.
Two companies can both report 20% ROE and be nothing alike. One earns 2% margins on assets that turn over three times, financed conservatively. The other earns 20% margins on assets that barely turn over once, financed with debt. They face different risks, respond to different shocks, and deserve different multiples. Only the decomposition makes that visible.
The three factors, and why they multiply
Write the identity out and the algebra is transparent:
ROE = (NI / Rev) × (Rev / A) × (A / E)
Revenue cancels between the first two terms; total assets cancels between the second and third; what is left is NI / E, which is ROE by definition. The decomposition is therefore an identity, not a model. It cannot be wrong, and it cannot add information to ROE. What it does is partition the number into three quantities that each have an independent economic meaning.
Net profit margin (NI ÷ Revenue) is the share of every sales dollar that survives cost of goods, operating expenses, interest and tax. It rises with pricing power and falls with competition, input costs and — importantly — with interest expense, which is why the margin term is not purely operational. It is expressed in percent.
Asset turnover (Revenue ÷ Total assets) is dollars of sales per dollar of assets. High-turnover businesses hold little capital relative to what they sell: distributors, discount retailers, service firms. Low-turnover businesses are capital-intensive: utilities, refineries, telecoms, hotels. It is expressed as a multiple.
Equity multiplier (Total assets ÷ Equity) is the leverage term. Because assets equal liabilities plus equity, this multiple is 1 for a company with no liabilities at all and rises as liabilities grow. A multiplier of 2.5 means every dollar of equity supports $2.50 of assets, so 60% of the balance sheet is funded by creditors, suppliers and other non-shareholders. It is expressed as a multiple.
The first two factors multiply to return on assets, the return the business earns on everything it controls regardless of who funded it. The third factor then translates that into a shareholder return. That is the cleanest way to read the identity: ROE = ROA × Equity multiplier. Leverage does not create return; it scales whatever return the assets already produce, in whichever direction that return points.
Whether you use closing or average balance sheet figures matters. Net income is a flow over a period; assets and equity are stocks at a point in time. Averaging the opening and closing balances matches the two better, and it matters most when the balance sheet moved a lot during the year through an acquisition, a large issuance or a buyback. Tick the averaging box and enter the comparative column figures.
Worked example: $250M of profit on $3,200M of revenue
Take the default figures. Net income is $250 million, revenue is $3,200 million, total assets are $2,800 million and shareholders' equity is $1,100 million, all at period end.
- Net profit margin. 250 ÷ 3,200 = 0.078125, or 7.8125%. Just under eight cents of profit per sales dollar.
- Asset turnover. 3,200 ÷ 2,800 = 1.1429×. Each dollar of assets generates $1.14 of sales in the year.
- Equity multiplier. 2,800 ÷ 1,100 = 2.5455×. Every dollar of equity carries $2.55 of assets, so liabilities fund 2,800 − 1,100 = $1,700 million, which is 60.7% of the balance sheet.
- Multiply the three. 0.078125 × 1.1429 × 2.5455 = 0.22727, or 22.73%.
- Check directly. 250 ÷ 1,100 = 0.22727 = 22.73%. The identity holds, as it must.
- Return on assets. 0.078125 × 1.1429 = 0.089286, or 8.93%. Confirm: 250 ÷ 2,800 = 8.93%.
- Leverage contribution. 8.93% × 2.5455 = 22.73%. Of the 22.73% ROE, 8.93 points come from the assets and the remaining 13.80 points are the amplification produced by financing 60.7% of those assets with liabilities.
Now suppose the company issues $300 million of new shares and uses the proceeds to retire $300 million of debt. Total assets stay at $2,800 million, so margin and turnover are untouched; equity rises to $1,100 + $300 = $1,400 million and the multiplier falls to 2,800 ÷ 1,400 = 2.0. ROE becomes 8.93% × 2.0 = 17.86%, a fall of 4.87 percentage points with no change whatsoever in the operations.
Be careful about which action does this. Retiring the same $300 million of debt with cash already on the balance sheet leaves equity at $1,100 million, so ROE stays at 250 ÷ 1,100 = 22.73%; assets fall to $2,500 million, turnover rises to 3,200 ÷ 2,500 = 1.28× and the multiplier falls to 2,500 ÷ 1,100 = 2.2727×, and the two moves cancel exactly (0.078125 × 1.28 × 2.2727 = 0.22727). Only a change in the equity base moves ROE; a change in the mix of assets and liabilities reshuffles the drivers between themselves.
How to read the three numbers
Read the drivers against the company's own history first and its direct competitors second. Cross-industry comparison of any single driver is close to meaningless, because the three trade off against each other by business model.
Net profit margin is the driver most sensitive to competition. A margin falling while revenue grows usually means the company is buying volume with price. A margin rising while revenue is flat is often cost discipline, but check whether it is a one-off tax item or a gain on disposal, because both sit in net income.
Asset turnover is the one management can improve without touching price. Working capital is the usual place to look: receivable days, inventory days and payable days move turnover directly, and the cash conversion cycle quantifies the effect. Falling turnover after a large capital programme is expected; falling turnover with flat capital expenditure suggests demand is weakening or inventory is building.
The equity multiplier is where judgement is needed. A rule of thumb among analysts is that a multiplier above roughly 3× in a non-financial company deserves a look at the debt schedule, and banks typically run near or above 10× because deposits and other liabilities fund most of their balance sheet. But the multiplier includes all liabilities, not just borrowings — payables, deferred revenue and lease liabilities all raise it — so pair it with an explicit borrowing measure such as the debt-to-equity ratio before you conclude anything about solvency.
For the ROE result itself, a common practitioner benchmark is that a sustained ROE in the mid-teens or above, achieved without an unusual equity multiplier, indicates a business earning well above a typical cost of equity. Test that properly rather than assuming it: estimate your cost of equity, blend it with the after-tax cost of debt in the weighted average cost of capital calculator, and compare. Value is created only when the return exceeds the cost of the capital that funded it.
One trap deserves naming. A company can raise ROE by shrinking equity — through buybacks, large dividends or accumulated losses — without improving anything operationally. Whether that helps shareholders depends on the price paid and the return the deployed cash was earning, which the DuPont identity does not tell you. When ROE jumps, check whether the numerator moved or the denominator did, and if it was the denominator, look at return on invested capital, which is far less sensitive to capital structure.
How different business models reach a similar ROE
| Business model | Net margin | Asset turnover | Equity multiplier | ROE |
|---|---|---|---|---|
| Discount grocer | 2.0% | 3.00× | 2.50× | 15.00% |
| Specialty retailer | 6.0% | 1.80× | 2.00× | 21.60% |
| Heavy manufacturer | 8.0% | 0.90× | 2.20× | 15.84% |
| Enterprise software | 20.0% | 0.60× | 1.50× | 18.00% |
| Regulated utility | 12.0% | 0.35× | 3.20× | 13.44% |
| Commercial bank | 25.0% | 0.04× | 10.00× | 10.00% |
The grocer and the software firm land within three points of each other from opposite directions: 2% margins turned three times a year against 20% margins turned 0.6 times a year.
Three-step, five-step and the extended DuPont
The five-step version splits the margin term further, into a tax burden (net income ÷ pre-tax income), an interest burden (pre-tax income ÷ EBIT) and an operating margin (EBIT ÷ revenue). Multiplied by asset turnover and the equity multiplier, it still collapses to net income ÷ equity. The advantage is that it separates the two things the three-step version blends: leverage raises the equity multiplier but simultaneously depresses the interest burden, so the net effect of debt on ROE is not simply the multiplier.
Use the three-step form for a fast read and for comparing across periods. Move to the five-step form when interest expense or the effective tax rate has changed materially, because that is exactly when the three-step margin term hides the story.
Mistakes and limitations to keep in mind
- Mixing closing and average balances between drivers. If you average assets for turnover, average them for the multiplier too, or the three factors will no longer multiply to ROE.
- Treating the identity as an explanation. The decomposition is arithmetic. It shows where a change came from; it does not tell you why margins moved or whether the leverage is prudent.
- Reading a high equity multiplier as debt. It measures all liabilities. A company funded largely by supplier credit and deferred revenue can show a high multiplier with no borrowings at all.
- Comparing ROE across capital structures. Two identical businesses with different debt loads will report different ROEs. Compare return on assets or return on invested capital when the financing differs.
- Ignoring the book value problem. Equity is a historic-cost accounting number. Years of buybacks above book value, or a large goodwill write-off, can shrink it enough to inflate ROE well beyond anything the operations justify — and equity can even go negative, at which point ROE stops being interpretable.
- Using one year. A single period picks up one-off gains, restructuring charges and tax settlements. Look at three to five years of the three drivers side by side before drawing a conclusion.
- Forgetting that a loss inverts the reading. When net income is negative the margin and both return measures turn negative while turnover and the multiplier do not, so leverage makes the shareholder return worse rather than better for as long as equity stays positive.
Where DuPont fits among return measures
Use the DuPont decomposition when the question is what changed in a shareholder return you already have. Use a different tool when the question is different.
Return on assets strips out financing entirely and is the right comparison when two companies carry different debt. Return on invested capital goes further and measures after-tax operating profit against the debt and equity actually invested in operations, excluding excess cash; it is the measure most directly comparable with the weighted average cost of capital, and the spread between them is the cleanest single indicator of value creation.
On the per-share side, earnings per share takes the same net income and divides it by shares rather than by equity, which makes it sensitive to issuance and buybacks in a way ROE is not. Between them, ROE tells you how productively the capital is being used and EPS tells you how that productivity is being shared out.
None of these are defined by an accounting standard. Net income, revenue, assets and equity come from financial statements prepared under US GAAP or IFRS, but the ratios themselves are analytical conventions, which is why two data providers can publish different ROEs for the same company. When you compare, make sure you know which balance sheet date and which definition of equity — including or excluding non-controlling interests — the other figure used.
Key terms
- Equity multiplier
- Total assets divided by shareholders' equity. Equals 1 + (total liabilities ÷ equity), so it is 1.0 for a company with no liabilities and rises with every dollar of liability.
- Asset turnover
- Revenue divided by total assets. A measure of asset productivity: how much a business sells for each dollar of resources it controls.
- Return on assets
- Net income divided by total assets, equal to net margin multiplied by asset turnover. The return the business earns before considering who funded the assets.
- Book equity
- The accounting value of shareholders' claim: contributed capital plus accumulated retained earnings, less treasury stock and accumulated losses. It is not market capitalisation.
- Value creation
- Earning a return on capital above the cost of that capital. A high ROE achieved at a cost of equity that is higher still destroys value rather than creating it.