Jump through this article9 sections
On this page
ROE tells you how much profit a company earns for every dollar shareholders have left in the business. The catch: that denominator — equity — can be shrunk by buybacks, debt, or years of losses, making the ratio glow on paper without any improvement in the underlying business.
The formula#
ROE = Net Income / Shareholders' EquityThe DuPont breakdown shows what actually drives the number:
| Driver | What it captures |
|---|---|
| Net profit margin | How much of each revenue dollar survives as profit |
| Asset turnover | How efficiently the asset base generates revenue |
| Financial leverage | How much of those assets are funded by debt |
ROE = net margin × asset turnover × leverage. A company can boost ROE by earning more, selling more efficiently, or borrowing more — and the first two are healthy signals; the third is not.
ROE interpretation bands#
| ROE | What it signals |
|---|---|
| Below 10% | Below the cost of equity for most businesses — capital is being destroyed |
| 10–20% | Solid; expected range for banks, utilities, and industrials |
| 20–30% | Strong — usually indicates a durable competitive advantage |
| Above 30% | Exceptional when driven by earnings; suspect when driven by leverage or buybacks |
| Consistently above 15% for 10+ years | Buffett's moat signal — pricing power is almost certain |
When ROE becomes noise#
Buybacks shrink equity, not just share count. A company repurchasing $700B of stock reduces retained earnings faster than profits accumulate — equity shrinks, ROE inflates, and the business may be exactly the same.
Debt amplifies ROE the same way. A retailer earning 8% return on assets with 3× leverage reports 24% ROE. The business hasn't improved; the balance sheet changed. ROE doesn't separate the two.
One-time items spike the numerator. A large asset sale, a tax reform windfall, or a legal settlement flows through net income for one year. The following year ROE reverts — but any screen run in that year flags the company as exceptional.
Buffett, Coca-Cola, and Apple: two very different ROEs#
Warren Buffett's 1988 Coca-Cola investment started with one observation: the company had earned above 20% ROE for ten consecutive years with minimal debt. That sustained signal told him the brand — not the balance sheet — was generating the returns. Coke still earns above 40% ROE today.
Apple's ROE tells a different story. In 2012, Apple earned ~30% ROE on a solid equity base. Then a $700B buyback program compressed equity to near-zero, and ROE crossed 100% in 2018.
Today, Apple Inc. (AAPL) sits at 146.7% ROE — a number that reflects extraordinary earnings power atop an intentionally hollowed-out equity base. The distinction matters.
What the market looks like right now#
Across 2,816 US stocks with reported earnings, the median ROE is 11.6%.
1,180 companies sit below 10.0% — burning capital or early-stage with unproven unit economics. 370 sit above 30.0%, either because of a real moat or because of financial engineering. The split is worth checking.
Sector context#
Healthcare posts the highest average ROE at 17.9%. Real Estate sits at the bottom with 8.9% — asset-heavy structures and thin margins compress equity returns in that sector.
Where ROE breaks down#
Negative equity inverts the ratio. McDonald's has had negative shareholders' equity for years because buybacks outpaced accumulated profits — ROE becomes negative or undefined even though the business earns ~$10B a year.
Leverage multiplies ROE and risk simultaneously. A bank earning 12% ROE on 10× leverage is running a fundamentally different risk profile than a software company with 12% ROE on no debt. ROE does not encode that difference.
Capital-light accounting inflates it. A company that leases all its assets carries minimal equity and shows high asset turnover — ROE rises without any change in economic returns. A manufacturer with the same profit but owned assets looks worse on the same screen.
Temporary earnings spike the ratio. Patent expirations, commodity cycles, and one-time tax events lift net income for one year. ROE lags the reversal by 12 months — any screen run in the peak year will flag the company as high-quality.
How to use ROE well#
- Require 15%+ consistently over a decade — a single high year is noise; a decade of consistency is a moat signal. Buffett's threshold was "above 15% with little debt for 10 years."
- Pair with ROIC — if ROE far exceeds ROIC, the gap is almost always debt. ROIC measures returns on all capital and can't be gamed with leverage alone.
- Run DuPont — split ROE into margin, turnover, and leverage. Rising ROE driven only by rising leverage is a yellow flag, not a green one.
- Check the equity trend — if equity is declining faster than earnings are growing, buybacks are doing the work. That is not the same business quality as earned ROE.
- Adjust for one-time items — strip out asset-sale gains and tax-reform windfalls before comparing ROE across years or against peers.
Bottom line#
ROE is a quality lens, not a quality verdict.
A stock at 8% ROE can be worth owning if leverage is zero, the business is scaling, and ROE is rising. A stock at 40% ROE can disappoint if buybacks are papering over flat earnings or if a debt binge drove the expansion.
The question ROE forces is: what is generating this return — the business, the balance sheet, or financial engineering?