ROIC vs ROE vs ROCE: what each measures, with one worked example

ROE measures profit against shareholders' money alone; ROIC measures after-tax operating profit against all the capital — debt and equity — put to work in the business; ROCE measures pre-tax operating profit against the capital employed (assets minus current liabilities). All three are a profit figure over a capital base, which is exactly why they get confused. The differences live entirely in two choices — which profit goes on top, and which capital goes on the bottom — and those choices are what make one ratio flatter a levered company, another ignore its cash pile, and a third swing with the business cycle. This article unpacks each formula, shows where each one misleads, and then computes all three for the same company from its 10-K so every figure is reproducible.

Each formula, unpacked

ROE — return on equity

Net income ÷ total shareholders' equity. The numerator is the bottom line: profit after interest, after tax, after everything. The denominator is the book value that belongs to shareholders. ROE answers a narrow, owner's-eye question: for every euro of equity in the business, how much profit came back to equity holders this year? What it excludes is the rest of the balance sheet — debt is nowhere in the ratio, which is the single most important thing to remember about it. Two companies with identical operations can post very different ROEs purely because one financed itself with more debt.

ROIC — return on invested capital

NOPAT ÷ invested capital. The numerator, NOPAT (net operating profit after tax), strips out the financing decision: it takes operating profit — EBIT — and applies tax, but adds back nothing for interest, because interest is a return to lenders, not a cost of operations. The denominator, invested capital, is the total capital funding the operations: debt plus equity, often with excess cash netted out. ROIC therefore answers the operator's question: how much after-tax profit does the business generate per euro of capital tied up in it, regardless of how that capital was financed? Because both the numerator and denominator are capital-structure-neutral, ROIC cannot be inflated by adding debt — the property ROE lacks.

ROCE — return on capital employed

EBIT ÷ capital employed. ROCE keeps EBIT on top — pre-tax operating profit, so no tax adjustment — and divides by capital employed, defined as total assets minus current liabilities (equivalently, equity plus all non-current liabilities). It answers roughly the same operator's question as ROIC, but before tax and with a balance-sheet-side definition of the capital base rather than a financing-side one. The practical consequence: ROCE and ROIC usually move together, but ROCE runs higher because its numerator is pre-tax, and the two can diverge when a company's current-liability structure (large payables, deferred revenue) differs sharply from its debt structure.

Where each one misleads

Leverage flatters ROE. This is the classic trap. Adding debt shrinks the equity base and, as long as the business earns more than its after-tax cost of borrowing, raises the return on what equity remains. A 15% ROE built on a debt-free balance sheet and a 15% ROE built on heavy borrowing describe very different risk. The tell is to compare ROE against ROIC: when ROE sits far above ROIC, the gap is the contribution of leverage, not operating quality. When they sit close together — as they do for a lightly levered company — ROE is largely reporting the same thing ROIC is.

Goodwill and intangibles distort ROIC. Invested capital is a balance-sheet number, and acquisition accounting loads the balance sheet with goodwill. A serial acquirer that paid large premiums carries a big goodwill balance, which inflates invested capital and depresses reported ROIC — even if the operating business is excellent. The mirror problem is a company that has expensed decades of R&D or written goodwill down: its capital base looks artificially small and its ROIC artificially high. This is why some analysts compute ROIC both with and without goodwill; the two can tell opposite stories about the same firm.

Cyclicality distorts ROCE (and any single-year ratio). ROCE built from one year's EBIT is a snapshot, and for a cyclical business the snapshot depends entirely on where in the cycle it was taken. A miner or chemicals producer can print a 25% ROCE at the peak and a 4% ROCE in the trough on an unchanged asset base. None of these ratios is designed to be read as a single number for a cyclical company — the honest version is an average across a full cycle, which is why quality frameworks that lean on these metrics almost always measure them over five to ten years, not one.

One worked example: Microsoft's FY2025 10-K

To make the differences concrete, here are all three computed for one real large-cap using only figures from its annual report — Microsoft's Form 10-K for the fiscal year ended 30 June 2025, filed with the SEC under accession number 0000950170-25-100235 on EDGAR. Every input below is a named line item from that filing. This worked example is illustrative analysis of a public filing — not an assessment of the company for any investor.

The inputs. From the Consolidated Income Statements (fiscal 2025): revenue $281,724M, operating income (EBIT) $128,528M, income before income taxes $123,627M, provision for income taxes $21,795M, and net income $101,832M. From the Consolidated Balance Sheets (as of 30 June 2025): total stockholders' equity $343,479M, long-term debt $40,152M, current portion of long-term debt $2,999M, total assets $619,003M, and total current liabilities $141,218M.

ROE. Net income over year-end equity:

ROE = $101,832M ÷ $343,479M = 0.296 = 29.6%

ROIC. First NOPAT. The effective tax rate is the provision over pre-tax income, $21,795M ÷ $123,627M = 17.6%, so NOPAT = EBIT × (1 − 0.176) = $128,528M × 0.824 = $105,869M. Invested capital is total debt plus equity: ($40,152M + $2,999M) + $343,479M = $386,630M. Then:

ROIC = $105,869M ÷ $386,630M = 0.274 = 27.4%

ROCE. Pre-tax EBIT over capital employed, where capital employed = total assets − total current liabilities = $619,003M − $141,218M = $477,785M:

ROCE = $128,528M ÷ $477,785M = 0.269 = 26.9%

Read the three together and the anatomy shows through. ROE (29.6%) sits slightly above ROIC (27.4%) — a small gap, because Microsoft carries little debt, so leverage adds only a few points; a heavily borrowed company would show ROE towering over ROIC. ROCE (26.9%) lands just below ROIC despite using pre-tax profit, because its capital-employed denominator includes large non-debt current liabilities. Three numbers within four points of each other, each computed from the same audited statements, each answering a subtly different question.

Which metric quality frameworks weight, and why

Systematic quality strategies do not use all three interchangeably; they pick the one that matches what they are trying to isolate. MSCI's World Quality Index builds its quality signal from return on equity, debt-to-equity and earnings variability — it uses ROE, then controls the leverage distortion separately with the debt-to-equity term rather than switching to ROIC. Value-and-quality strategies in the tradition of Joel Greenblatt's Magic Formula rank on a return-on-capital measure instead — operating earnings over tangible capital employed — precisely because it is capital-structure-neutral and harder to game with financing. The rule of thumb: frameworks that want a pure operating-quality signal reach for ROIC or ROCE; frameworks that want the equity holder's realised return, and are willing to model leverage as its own factor, use ROE. Acutic's own fundamental-quality pillar reads return on equity among its inputs and scores balance-sheet leverage as a separate component — the construction is documented on the methodology page.

Why two data providers show different ROIC for the same company

Run the same company through two financial-data sites and the ROIC figures rarely match — not because one is wrong, but because ROIC has no single official definition. The worked example makes the size of the effect concrete. Using invested capital as debt plus equity, Microsoft's ROIC is 27.4%. Net out the $30,242M of cash and cash equivalents (many providers treat cash as not “invested” in operations) and the denominator falls to $356,388M, lifting ROIC to 29.7%. Net out short-term investments too — another $64,323M — and it jumps to 36.2%. Same profit, same filing, three defensible answers spanning nine percentage points, driven entirely by how each provider treats the cash pile.

The denominators are only the start. Some providers use year-end balances; others average the opening and closing balance sheet. Microsoft's ROE is 29.6% on year-end equity but 33.3% on average equity, because equity grew sharply during the year. Others adjust NOPAT's tax rate to a normalised or statutory figure rather than the reported effective rate; still others add operating leases into invested capital or strip goodwill out of it. None of this is error. It means an ROIC number is only comparable to another ROIC number computed the same way — which is the practical case for preferring a provider that publishes its exact definition over one that just prints a figure.

Using them in a screen

The mechanics of putting these ratios into a screen follow directly from their properties. Because the metrics are not comparable across sectors — capital-light software runs structurally higher returns on capital than capital-heavy utilities — a screen ranks each company against its sector peers rather than the whole market, exactly as a quality score does. Because single-year figures are noisy for cyclicals, a screen typically averages the ratio over several years or requires the threshold to be met in most of them. Because the definition drives the number, a screen has to fix one definition and apply it uniformly to every company, so the comparison is apples to apples. A screener that lets you filter on a return metric, disclose its exact formula, and rank within sector is doing all three; that is the design of the Acutic screener. A screen's output is a ranked shortlist to research, never a conclusion — the metrics tell you where the capital returns are high, not why, and only the filings answer the why.

Further reading: what a stock quality score measures — how ratios like these become one comparable number — how to read a 10-K in 30 minutes — where every input above comes from — and the stock research checklist — the routine these metrics plug into. Create free account.

Acutic provides investment research and educational analysis under MAR Art. 20 / § 85 WpHG. Acutic does not provide investment advice (Anlageberatung per § 1 Abs. 1a S. 2 Nr. 1a KWG / Art. 4(1)(4) MiFID II), portfolio management, or any other licensed investment service. No content in this article constitutes a personal recommendation.