Home Depot earned 110% on equity last year. Nobody earns 110%

Home Depot earned a 110% return on equity in the financial year that ended in February 2026.

Take that at face value and it is one of the great businesses in the history of commerce. Put a dollar in, get a dollar and ten cents back, every year, forever. Warren Buffett has spent sixty years compounding at around 20% and is considered the best there has ever been at it.

Home Depot is an excellent business. It is not five times better than Buffett. The number is not a lie and it is not a typo. It is a correctly computed answer to a question that stopped being useful some time ago.

Equity is not what you put in

Return on equity is net income divided by shareholders' equity, and the trouble is entirely in the denominator.

Shareholders' equity is not the money invested in the business. It is what is left on the balance sheet after you subtract everything the company owes from everything it owns. It is a residual. It is the plug.

Which means you can change it without touching the business at all.

Buy back stock and the cash leaves the asset side while the buyback is charged against equity, so the denominator shrinks. Nothing about the stores, the staff, the supply chain or the earnings has moved. The ratio goes up because you made the bottom of the fraction smaller.

Home Depot has been buying back stock for years and was still doing it through 2026. At its February 2026 year end it carried $60.9 billion of debt against $12.8 billion of equity. The 110% is that $12.8 billion doing the work of a denominator far too small to be doing it.

You can watch the effect move. A year earlier, at the February 2025 year end, equity was only $6.6 billion and the same calculation gave 223%. Earnings barely moved between the two years. The business did not become half as good. The denominator doubled.

Run it far enough and the number breaks

McDonald's has gone one step further. Its shareholders' equity is negative $1.79 billion.

That is not distress. McDonald's earned $8.56 billion in 2025. It has bought back so much of its own stock over so many years that the residual went through zero and out the other side, which is a thing a residual is perfectly entitled to do.

But now put it in the formula. Positive earnings divided by negative equity gives a return on equity of negative 478%, and a screen sorting on ROE will file one of the most reliably profitable companies of the last fifty years somewhere below a business that is actually failing.

The number has not become inaccurate. It has become meaningless, which is worse, because an inaccurate number announces itself.

The denominator you cannot move money out of

Return on invested capital asks a different question: what does the business earn on all the capital inside it, debt and equity together.

Because it counts both, moving money from one to the other does nothing. Pay for a buyback by borrowing and equity falls, debt rises, invested capital is unchanged. The ratio does not budge. That is the whole point of it.

On that measure Home Depot earned 21.6% and McDonald's 23.3%. For scale: across the 971 companies in our universe with a positive invested capital balance, the median was 9.8%, the top quarter began at 17.2%, and the top tenth at 28.9%.

So both are genuinely strong, sitting in the top quarter without reaching the top tenth. That is the honest version of the 110%.

The gap is the interesting part

The cases worth your attention are the ones where the two numbers disagree sharply and the business is not obviously exceptional.

Oracle returned 40.2% on equity in the year to May 2026. On invested capital it returned 9.4%, against that 9.8% median. Elite by one measure, slightly below average by the other, and the distance between them is $159.7 billion of debt sitting on $42.5 billion of equity.

None of that makes Oracle a bad company, and it is spending enormously on data centres, which depresses the ratio while the spending is going in and before the revenue comes out. But it does mean that if you ranked businesses by return on equity, Oracle would appear above hundreds of companies that earn more on every dollar actually working inside them. You would not be measuring the quality of the business. You would be measuring how much of it was financed with debt, which is a decision the finance department made, not a property of the operation.

That is the same failure as reading a low P/E as a discount: a ratio quietly answering a different question than the one you asked.

Where ROIC lets you down too

It is not a magic number, and three things dent it.

Acquisitions inflate the denominator. Goodwill from an overpriced purchase sits in invested capital forever, so a serial acquirer can look mediocre on ROIC while running fine underlying economics. Sometimes that is the truth telling you something. Sometimes it is an accounting scar.

Asset-light businesses flatter it. A company whose real assets are a brand and some engineers has very little invested capital, so the ratio gets large fast. Mastercard cleared 84% in 2025. That is real, and it also means small changes in the denominator swing the number around.

And a wonderful business is still a bad investment at the wrong price. ROIC says nothing about what you pay. It measures the machine, not the deal.

Which is why we do not rank companies on any single ratio, and why we publish what the pillars are without publishing how they are weighed. The methodology has the shape of it.

How to spot it from the outside

You do not need our tables for the first pass. Two glances at any company's balance sheet do most of the work.

First, the equity line. If it is small next to the debt line, or negative, then any return on equity quoted for that company is describing the financing, and you can stop reading it as a quality signal right there.

Second, the buyback history. A company that has retired a large share of its own stock over the years has been shrinking the denominator that whole time, so its ROE has been inflating without the business changing at all.

Neither glance tells you whether the business is good. They tell you whether the popular number about it means anything, which is the cheaper thing to learn first.

The short version

If you only ever look at one of the two, look at return on invested capital. Return on equity tells you what the business earns and what the balance sheet has been doing, mixed together, with no label saying which is which.

And when the two disagree by a factor of four, that gap is not noise. It is the answer to a question you had not thought to ask.

---

Every figure comes from our own annual financials table, read on 9 September 2026, using each company's most recently filed annual report: Home Depot to 1 February 2026, Oracle to 31 May 2026, McDonald's and Mastercard to 31 December 2025. Universe statistics cover the 971 companies with a positive invested capital balance in their latest filed year.