Guide

Return on capital employed explained

Harbor Industrial’s quality sleeve screened on double-digit return on equity — median ROE ran 24% across twenty-eight capital-goods and industrial services names. Portfolio post-mortem after the 2024 de-rating showed that seven of the worst performers carried ROE above 20% while return on capital employed (ROCE) averaged 9.1%, below Harbor’s estimated WACC of 10.5%. The gap traced to goodwill-heavy roll-ups, stretched payables inflating short-term capital, and leverage boosting equity returns without operating improvement. After adding a ROCE floor and WACC-spread gate, value-destructive compounders fell from 27% to 8% of the book and forward-looking economic-profit estimates improved 140 bps.

Return on capital employed asks a blunt question: for every dollar of capital tied up in the business (equity plus interest-bearing debt, net of cash), how much operating profit does management generate after tax? Unlike ROE, ROCE cannot be inflated by shrinking the equity base through buybacks alone. Unlike headline operating margin, it penalizes asset-heavy or acquisition-driven balance sheets. This guide covers the formula, NOPAT and capital employed definitions, ROCE vs ROE vs ROIC, pairing ROCE with WACC for value-creation tests, links to DuPont analysis, the Harbor Industrial refactor, a technique decision table, pitfalls, and an investor checklist.

The formula and what it measures

The standard definition of ROCE is:

ROCE = EBIT × (1 − Tax Rate) ÷ Capital Employed

Or equivalently, using net operating profit after tax (NOPAT):

ROCE = NOPAT ÷ Capital Employed

Where:

NOPAT = operating profit after tax, before financing costs — typically EBIT × (1 − effective tax rate), or net income plus after-tax interest expense.
Capital employed = total assets − current liabilities (excluding interest-bearing debt) — or, equivalently, shareholders’ equity plus long-term debt minus cash and cash equivalents.

Example: EBIT of $200M, tax rate 25%, NOPAT = $150M. Capital employed (equity $800M + debt $400M − cash $100M) = $1,100M. ROCE = $150M ÷ $1,100M = 13.6%. If WACC is 10%, the company creates economic value; if WACC is 15%, it destroys it despite positive accounting earnings.

ROCE is a pre-financing return: it measures operating performance before capital structure choices split rewards between debt and equity holders. That makes it useful for comparing businesses with different leverage policies on equal footing.

Building NOPAT and capital employed consistently

Inconsistency kills cross-company ROCE screens. Standardize these inputs:

NOPAT construction

Start from operating income (EBIT), not EBITDA — depreciation is a real economic cost for asset-heavy firms.
Use a normalized tax rate (statutory or 3-year effective average), not a one-time tax benefit that inflates NOPAT.
Strip restructuring charges and large litigation settlements from EBIT when they are genuinely non-recurring; document each adjustment.
For banks and insurers, ROCE variants use different numerators — do not apply industrial NOPAT formulas blindly.

Capital employed construction

Include goodwill and intangibles from acquisitions — ignoring them flatters serial acquirers.
Use average capital employed (opening + closing ÷ 2) for growing or shrinking balance sheets; point-in-time snapshots distort M&A years.
Decide whether operating leases belong in capital employed under ASC 842; be consistent across peers.
Subtract excess cash only above a working-capital buffer — stripping all cash overstates ROCE for cash-rich tech names.

Document your template once and apply it across the universe. Harbor Industrial’s error was mixing EBITDA-based numerators with equity-only denominators on half the sleeve.

ROCE vs ROE vs ROIC

These ratios answer related but distinct questions:

Ratio	Numerator	Denominator	Best for
ROE	Net income	Shareholders’ equity	Per-share owner returns; DuPont decomposition
ROCE	NOPAT (EBIT after tax)	Equity + debt − cash	Operating efficiency across capital structures
ROIC	NOPAT	Invested capital (often ex-cash, ex-non-op assets)	Value creation vs WACC; moat analysis

ROCE and ROIC converge when definitions align; practitioners differ on whether to exclude goodwill or non-operating assets from invested capital. ROE diverges most when leverage is high: a firm can show 25% ROE with 9% ROCE if debt is doing the heavy lifting — check equity multiplier alongside both metrics.

Rule of thumb for industrials: ROCE should exceed WACC by 2–4 percentage points to justify a growth premium. ROE above cost of equity without ROCE above WACC often signals financial engineering, not operational moat.

ROCE, margins, and asset turnover

ROCE decomposes like a DuPont identity:

ROCE = NOPAT Margin × Capital Turnover

Where NOPAT margin = NOPAT ÷ revenue and capital turnover = revenue ÷ capital employed. A business can earn high ROCE through fat margins (luxury, software) or high turnover on thin margins (distribution, grocery). Comparing only net profit margin misses whether the margin is earned on a bloated or efficient asset base.

Pair ROCE with asset turnover: declining turnover with flat margins often precedes ROCE compression when capex ramps or acquisitions integrate poorly. For cyclicals, average ROCE over a full cycle beats peak-year snapshots.

WACC spread and economic profit

ROCE becomes actionable when compared to the cost of capital:

Economic Profit Spread = ROCE − WACC

Positive spread implies the company earns more on deployed capital than providers of debt and equity require — the foundation of economic value added (EVA) and durable moats. Negative spread with positive net income means accounting profits mask value destruction: growth actually shrinks shareholder wealth unless ROCE rises.

Stress-test WACC assumptions. A 50 bps error in cost of equity moves the spread materially for 11–12% ROCE businesses sitting near breakeven. Use a range (base, bear, bull) rather than a point estimate when screening.

Link to sustainable growth rate: high SGR built on ROE inflated by leverage fails if ROCE sits below WACC — retained earnings reinvest into negative-spread projects.

Harbor Industrial refactor

Harbor Industrial’s sleeve previously ranked holdings by ROE and 3-year revenue CAGR. The 2024 drawdown exposed names where ROE looked healthy but ROCE trailed WACC for three consecutive years — often post-acquisition goodwill with integration delays and debt-funded buybacks shrinking the equity denominator.

New rules (applied over two rebalance cycles):

Compute TTM NOPAT and average capital employed; derive ROCE; require ROCE ≥ WACC − 1% for new positions.
Flag holdings where ROE exceeds ROCE by more than 12 percentage points — leverage or equity-shrink distortion review.
For serial acquirers, stress-test ROCE assuming 10% goodwill impairment on deals closed in the last 36 months.
Exclude names where ROCE fell two consecutive years while net debt rose >0.4× EBITDA.

Outcomes: value-destructive bucket shrank from 27% to 8% of NAV; median sleeve ROCE rose from 10.8% to 13.2% while median ROE only fell from 22.1% to 19.4% — the book became less flashy on equity returns but more grounded in operating capital efficiency. Simulated recession drawdown improved 140 bps.

Technique decision table

Approach	Best for	Weak when
Operating margin alone	Quick profitability screen, same-asset-base peers	Ignores capital intensity and leverage
ROE	Owner returns, DuPont, dividend growth linkage	Distorted by buybacks, leverage, one-off equity items
ROCE	Cross-leverage comparison, industrial and cyclical moats	Definition drift vs ROIC; goodwill treatment debates
ROIC (invested capital)	EVA, moat durability, capital allocation	Non-operating asset exclusions vary by analyst
ROA	Banks, asset-heavy regulated utilities	Understates returns when leverage is structurally high
ROCE − WACC spread	Value creation, growth quality gates	WACC estimation error; sector beta assumptions

Common pitfalls

EBITDA-based numerators — ignoring depreciation overstates returns for capex-heavy businesses.
Excluding goodwill — flatters acquirers; paid premiums are real capital at risk until proven otherwise.
Point-in-time capital employed — M&A and divestitures need averaging; year-end snapshots lie.
Inconsistent cash subtraction — some analysts strip all cash; others only excess cash; peer ranks shuffle.
Peak-cycle NOPAT — commodity upcycles inflate EBIT; ROCE mean-reverts with margins.
Ignoring lease capitalization — pre-ASC 842 comparisons to modern peers are apples-to-oranges.
ROCE without WACC — 12% ROCE is excellent at 8% WACC, mediocre at 14% WACC.
Financials and REITs — standard industrial ROCE formulas do not apply; use sector-specific return metrics.

Investor checklist

Compute TTM NOPAT from EBIT and a normalized tax rate; document adjustments.
Calculate average capital employed (equity + debt − cash) over the fiscal year.
Derive ROCE; compare to estimated WACC and record the spread.
Contrast ROCE to ROE; investigate gaps wider than 10–12 percentage points.
Decompose into NOPAT margin and capital turnover; identify the driver.
For acquirers, stress-test ROCE with goodwill impairment scenarios.
Track 3–5 year ROCE trend, not a single peak year.
Cross-check with free cash flow yield and capex intensity.
Benchmark ROCE to sector median and closest operational peers.
Document thesis: what ROCE and spread must hold for growth to create value?

Key takeaways

ROCE measures operating return on all employed capital — equity plus debt, before financing splits.
ROE can lie when leverage or buybacks inflate equity returns — ROCE grounds the story in operations.
Compare ROCE to WACC — positive spread is the minimum bar for value-creating growth.
Define NOPAT and capital employed consistently — goodwill and averaging choices matter.
Harbor cut value-destructive holdings from 27% to 8% by pairing ROE screens with ROCE-WACC gates.