Return on Assets: What the Ratio Says About a Business, and What It Hides

The ratio, and the two halves it is made of

Take one company and keep it for the whole article. Its balance sheet holds goodwill of 1,200 from an acquisition, net property and equipment of 4,800, inventory of 1,500, receivables of 2,000 and cash of 1,500 — total assets of 11,000. On the other side, 2,000 of current liabilities (supplier credit, accrued wages, tax not yet paid), 3,000 of long-term debt and 6,000 of equity, which adds back to 11,000. It sells 14,000, earns an operating profit of 1,400, pays 180 of interest and 305 of tax at 25%, and keeps 915.

Return on assets is 915 ÷ 11,000 = 8.3%. That single number is the product of two very different things, and separating them is the first useful thing you can do with it: net margin, 915 ÷ 14,000 = 6.54%, multiplied by asset turnover, 14,000 ÷ 11,000 = 1.2727. A business can reach the same 8.3% by charging a lot on a slow, heavy balance sheet or by charging almost nothing on a fast, light one. Two competitors with identical ROA can therefore be opposite businesses, and the ratio on its own will never tell you which you are looking at.

The numerator does not match the denominator

Net income belongs to the shareholders: interest has already been taken out of it and paid to the lenders. Total assets belongs to everyone: it is the whole pot, funded by shareholders, lenders and suppliers together. Dividing one by the other measures a return to one group over capital provided by three, which is why the ratio falls when a company borrows and rises when it repays, without anything happening in the business at all.

There are two clean repairs. Lift the numerator above the financing line: EBIT ÷ total assets = 1,400 ÷ 11,000 = 12.7%, which measures what the assets earned before anyone was paid for lending them. Or keep net income and add back the after-tax cost of the debt: (915 + 180 × 0.75) ÷ 11,000 = 1,050 ÷ 11,000 = 9.5%, which is the version an analyst uses when comparing companies with different capital structures. The calculator on this page deliberately implements the plain form, net income over total assets, because that is the one a bank, a supplier and a company registry will all compute from published accounts — but knowing that 8.3%, 9.5% and 12.7% are the same company on the same day is the point.

Move the denominator and you get the rest of the family

Capital employed is what the providers of long-term money actually had to put up. On this balance sheet it is 9,000, and three routes reach it: total assets less current liabilities (11,000 − 2,000), non-current assets plus working capital (6,000 + 3,000), and equity plus non-current liabilities (6,000 + 3,000). They agree because they are the same subtraction read from different sides of the sheet — and when they do not agree on your accounts, you have found a classification error, not a choice of method.

Return on net assets keeps net income and swaps the denominator: 915 ÷ 9,000 = 10.2%. The whole move is worth ×1.2222, which is 11,000 ÷ 9,000 — the ratio of what you hold to what you had to fund. What left the denominator is the free financing: the supplier who has not been paid yet, the wages accrued at month end, the tax owed to the treasury. A business that pays suppliers in 90 days and collects in 30 will show a much better return on net assets than on total assets, and the gap is not efficiency, it is somebody else's money. Return on capital employed then keeps that denominator and lifts the numerator to EBIT: 1,400 ÷ 9,000 = 15.6%, a further ×1.5301.

So the family is a two-by-two and not a ladder. One axis is the numerator: before financing and tax, or after. The other is the denominator: everything the company holds, or only what long-term investors had to fund. Return on equity sits in a third corner — profit for shareholders over shareholders' money — and it rises with borrowing rather than falling, which is why a company can post a rising return on equity and a falling return on assets in the same year and both be true.

What the denominator hides

Cash first. The 1,500 sitting in the bank earns nothing operational, yet it is in the denominator and drags the ratio down. Strip it out and the same company reads 915 ÷ 9,500 = 9.6% on assets, and 1,400 ÷ 7,500 = 18.7% on capital employed. Neither number is more honest than the other: the first says the balance sheet is over-capitalised, the second says the trade is good. A company holding a war chest for an acquisition and a company that cannot collect its invoices will look alike on the raw ratio and nothing alike once you split operating assets from surplus cash.

Then history. The 1,200 of goodwill exists because this company bought its growth; an identical business that built the same customers itself would carry 9,800 of assets and read 9.3%. Then age. Hold profit constant and let the plant depreciate: with net equipment of 7,200 the ratio is 6.8%, at 4,800 it is 8.3%, at 2,400 it is 10.6% and at 800 it is 13.1%. Nothing improved. The denominator shrank by 800 a year while the numerator stayed still, and the business quietly became a candidate for a capital expenditure it is not showing. Revaluation runs the same lever the other way: measure the same property at 7,000 instead of 4,800 under the revaluation model of IAS 16 and the ratio falls to 6.9%.

Leases are the largest single distortion and the one most people miss. Since IFRS 16 took effect in 2019, a lessee puts a right-of-use asset and a lease liability on the balance sheet for almost every lease, which inflates total assets and deflates every return computed on them. A French company preparing individual accounts under the plan comptable général does the opposite: crédit-bail stays off the balance sheet and the payments run through operating charges, so the same shop, the same machine and the same rent produce a materially higher return on assets in the French accounts than in the group's IFRS accounts. That is not a discrepancy to reconcile, it is two frameworks answering two questions, and the ratio is only comparable within one of them.

No standard defines it, which is why you must

Accounting frameworks define line items, not ratios. That is not an oversight: it is why the European securities regulator issued guidelines on alternative performance measures in October 2015, applicable to information published from 3 July 2016, whose definition of such a measure covers any financial measure of performance, position or cash flows that the reporting framework does not define. Under those guidelines a listed issuer must define the measure and the basis on which it is calculated, and reconcile it to the most directly reconcilable line item in the financial statements. Return on assets, return on net assets and return on capital employed all sit squarely inside that description.

Two things are about to move the ground under these ratios. IFRS 18, issued in April 2024 and effective for annual periods beginning on or after 1 January 2027, replaces IAS 1 and for the first time defines an operating profit subtotal, along with disclosure of management-defined performance measures — so the numerator of a ROCE will finally have a standardised starting point. And in France, the modernisation of the financial statements has already landed: in the model income statement of the plan comptable général, proceeds from the disposal of tangible and intangible fixed assets and the carrying amount of what was sold both sit inside operating income and operating charges, and the exceptional result is now reserved for items directly linked to a major, unusual event. Sell a building and your operating profit rises while your asset base falls: both halves of a ROCE move in your favour for a reason that is not trading.

The practical rule that falls out of all this is short. Write your definition down, including which assets are averaged and over what period, and then never change it without restating the history. Compare the company with itself across years, and with competitors only when you can see their definition. And when someone quotes you a sector average, ask what the denominator contained: on the numbers above, one business produced 6.8%, 8.3%, 9.3%, 9.5%, 9.6%, 10.2%, 12.7%, 13.1% and 15.6% without a single euro of trading changing hands.

Frequently asked questions

Should I use year-end total assets or the average for the year?

The average, because the numerator covers a whole year and the denominator would otherwise be a photograph of one day. The simple average of opening and closing balances is enough in a stable year; use monthly or quarterly averages when the balance sheet moved sharply, which above all means an acquisition, a large capital expenditure or a disposal near year end. The distortion is not small: buy a business in December and a year-end denominator carries twelve months of new assets against one month of their profit. The calculator takes one figure for total assets, so compute the average yourself and enter that.

Is a rising return on assets always good news?

No, and the counter-example is common enough to check for first. Depreciation shrinks the denominator every year on its own, so a company that stops replacing its equipment posts a rising ratio for as long as the old machines keep running: on the figures above, the same unchanged profit reads 6.8% against new plant and 13.1% against plant near the end of its life. Before congratulating anyone, look at gross fixed assets rather than net, at the ratio of accumulated depreciation to gross cost, and at capital expenditure against the depreciation charge. If investment has been below depreciation for several years, the improving ratio is the sound of the asset base draining, not of the business improving.

Return on assets, on equity or on capital employed — which should I actually track?

Track the one that matches the decision. If you are deciding whether an operation is worth running at all, use return on capital employed and compare it with your cost of capital: it is the only one of the three built to be compared with a financing rate, because both sides then refer to the same pool of money. If you are a shareholder asking what your own money earned, use return on equity, and remember that borrowing more raises it mechanically. Return on assets is the outsider's ratio: it is what a supplier, a credit insurer or a competitor can compute from filed accounts without knowing anything about your financing. Its virtue is availability, not precision.

What is a good return on assets?

The question has no cross-sector answer, and the reason is arithmetic rather than opinion. The denominator of a bank contains its loan book; the denominator of a software firm contains almost nothing; the denominator of a utility contains a network built over decades. A number that is excellent in one is impossible in another, and the published sector averages you will find are averages of definitions as much as of businesses. Two comparisons do work. The first is your own ratio over several years on an unchanged definition. The second is the ratio against your cost of capital, using return on capital employed rather than return on assets, because a business earning less on capital employed than it pays for capital is destroying value however respectable the headline percentage looks.

Does leasing rather than buying change my return on assets?

It used to change it enormously and now depends on which set of books you mean. Under IFRS 16, in force since 2019, leasing puts a right-of-use asset in the denominator, so the ratio ends up close to what buying would have produced — that was the point of the standard. Under US GAAP the balance sheet treatment is the same, and the difference sits in the income statement instead, because an operating lease there keeps a single straight-line lease cost inside operating expenses. In a French individual account under the plan comptable général, crédit-bail stays off the balance sheet entirely, so leasing still flatters the ratio exactly as it did before 2019. Before you compare two companies on return on assets, check whether either of them leases and under which framework each reports.