Return on equity

Return on equity is net income divided by shareholders equity. A company earning $200m on $1bn of equity returns 20%, meaning it produced twenty cents in the year for each dollar of owner capital in the business.

It is the closest single figure to the question of whether a business is a good one, because it measures output against what the owners have put in rather than against revenue or assets. Sustained high returns are uncommon and usually indicate something durable: a brand, a network, a licence, or a cost position competitors cannot match. Without such an advantage, returns compete back down towards the cost of capital.

Equity here is the book value on the balance sheet, and most calculations use the average of the opening and closing figures rather than the year-end one, since profit was earned across the year rather than on the last day.

Equity is the denominator, and borrowing shrinks it. A company can raise its return on equity by funding itself with more debt and less owner capital, with no improvement in the underlying operation, and the ratio improves as the balance sheet becomes more fragile.

The DuPont decomposition splits the ratio into three parts and makes the source visible: net margin, asset turnover and the equity multiplier, which is assets divided by equity and is the leverage term. Two companies can reach a similar ROE through entirely different combinations.

Company A earns a fifth of its revenue and borrows almost nothing. Company B earns a twenty-fifth of its revenue and is financed five times over. B posts the higher return on equity, and a screen sorted on ROE puts it first.

Return on invested capital counts debt and equity together in the denominator, so it measures the return on all the money in the business regardless of where it came from. It is the version leverage cannot flatter.

Read side by side, the two answer different questions: ROE describes the return to owners as the company is currently financed, and ROIC describes the quality of the operation underneath. A high ROE with a mediocre ROIC is a financing decision. Both high is a business.

The screener carries ROE, ROIC and ROCE as separate metrics for that reason, so a filter can require both rather than either.

Sustained figures in the mid teens and above are the conventional marker of a business earning more than its cost of capital, and duration matters more than level. A decade in the high teens through a downturn is stronger evidence than a single year at 40%, which is often a cyclical peak or a one-off gain.

Comparison is only meaningful within a sector or against the company's own history. Banks run structurally high leverage by design and are measured against their own range, not against a software company's.

A very high figure is worth checking against the denominator before it is read as quality, for the reason in the next section.

Years of buybacks reduce book equity directly, which sends ROE to levels that describe the accounting rather than the performance. Several large, well-known companies show returns on equity in the hundreds of percent for this reason alone.

Taken far enough, equity turns negative, and a negative denominator returns a negative ROE for a profitable company. The ratio has stopped being a measure at that point rather than reporting a bad result, which is why screens usually exclude it.

Equity is also historical: it records what was contributed and retained, not what the assets are worth now. A company whose value sits in brands or research that was expensed rather than capitalised shows a small equity base and a correspondingly high return, which is a statement about accounting conventions as much as about the business.

Reading a high ROE as a quality signal without checking leverage, which is the single most common error with this ratio.

Comparing across industries, where capital structures differ by design.

Using a single year. Net income carries one-off items, and a gain on an asset sale flows straight into the numerator.

Ignoring negative or near-zero equity, where the ratio produces figures that look extreme in either direction and mean nothing.

Treating ROE and ROIC as interchangeable. The gap between them is the effect of borrowing, which is the thing worth measuring.