Return on Assets (ROA) vs ROE: The Honest Profit Test

In FY25, two well-known Indian companies posted almost exactly the same return on equity. Cipla earned about 16.9% on its shareholders' money; Adani Green Energy earned about 16.5% (Inve data, FY25; figures cross-checked against company filings, FY25). A screener sorting on ROE would have set them side by side as near-equals. They are not even close. On return on assets — profit measured against everything the business put to work — Cipla earned 14.1% and Adani Green earned 1.8%. One number says the two are twins. The other says one of them is roughly eight times as productive with the capital it controls. (This is an illustration of a ratio, not a view on either stock.)

That gap is the whole point of this article. A company can lift its ROE simply by borrowing more. It cannot do the same to its ROA, because the denominator — total assets — grows when you borrow rather than shrinks. When ROE runs far ahead of ROA, you are not looking at a better business. You are looking at a more leveraged one. ROA is the number a careful analyst reaches for when a high ROE looks too good, and across the data it is the one that quietly catches the difference between a management that improves the business and one that just adds debt to the same business.

What does return on assets actually measure?

Return on assets answers a deliberately blunt question: for every rupee of assets the company controls, how much profit does it generate?

Total assets is the whole base — factories, inventory, receivables, cash, goodwill, everything the balance sheet records — regardless of whether shareholders or lenders funded it. That is the point. ROE looks only at the equity slice; ROA looks at the entire asset base. A business that produces ₹14 of profit on ₹100 of assets is genuinely more productive than one that produces ₹2 on the same ₹100, no matter how either financed those assets — which is, give or take rounding, exactly the Cipla-versus-Adani-Green spread above.

Because the denominator includes debt-funded assets, ROA is insensitive to the financing trick that inflates ROE. Pile on debt and your equity shrinks while assets grow — ROE jumps, but ROA, measured against the now-larger asset base, does not budge unless the new assets actually earn their keep. This is what makes ROA the honest profitability test: it rewards businesses that use their assets well and refuses to reward businesses that merely borrow more.

Why is ROA harder to manipulate than ROE?

ROE's vulnerability is structural. Its denominator — equity, measured as net worth (share capital plus reserves and surplus) — is the most malleable line on the balance sheet. Take on debt, buy back shares, book a loss, write down equity, and the denominator shrinks. A smaller denominator produces a bigger ROE with no improvement in the actual business. We unpack this fully in our piece on what's a good ROE in India and when to distrust it.

ROA closes that escape hatch. Whatever a company does with its capital structure, the assets are the assets. Borrow ₹500 crore and your total assets rise by ₹500 crore — so to keep ROA steady, that borrowed money has to actually generate proportional profit. If it doesn't, ROA falls and tells you the truth that ROE was hiding: the new capital is not earning its cost.

This is why ROA is the metric that survives a credit cycle. During easy-money years, leverage flatters ROE across whole sectors, and the leveraged businesses look as good as the productive ones. When rates rise and lenders tighten, the leveraged ROE collapses while the productive one holds — and the ROA, which was telling you the difference all along, is vindicated. The owner who watched ROA was never fooled by the boom.

The honest caveat: ROA is not un-gameable, and it is not perfectly financing-neutral either. Because its numerator is net profit after finance costs, a heavily indebted company's interest bill still drags its ROA — what ROA neutralises is the equity-shrinking denominator trick, not the interest drag of debt itself. The metric the curriculum designates as fully capital-structure-neutral is ROCE (operating profit ÷ capital employed), which uses pre-interest profit precisely to strip financing out — and tellingly, on FY25 numbers Cipla's ROCE was about 23% against Adani Green's roughly 7% (company filings, FY25), the same hierarchy ROA already revealed. Off-balance-sheet leverage, aggressive asset write-downs, and the treatment of leases and goodwill can all distort ROA too. But compared with ROE, it is far harder to dress up, because you cannot make the asset base disappear the way you can shrink equity.

A worked example: same ROE, very different honesty

Here is the FY25 record — real companies, audited full-year figures — sorted not by ROE but by ROA. Every name in this list earned a return on equity between roughly 16% and 20%. A screener would have called them a cohort. Read down the ROA column instead.

Source: Inve data, FY25 (net profit = sum of the four FY25 quarters from financialreport; assets, borrowings and net worth from the March-2025 balance sheet); ROE computed on closing net worth. Figures for Cipla and Adani Green were cross-checked line-for-line against company filings, FY25, and match. Illustration of a ratio, not a view on any of these stocks; REC, being a lender, is shown for the leverage pattern, and financial companies are read on their own scale (see below).

Look at what the two columns disagree about. The ROE column is almost flat — a 16-to-20 band from top to bottom. The ROA column falls by a factor of nearly eight. And the Debt/Equity column rises in near-lockstep with the ROA fall, from one paisa of debt per rupee of equity at the top to ₹6.59 at the bottom. The companies that show the same ROE on the smallest asset productivity are, without exception, the ones carrying the most debt. The ROE is identical; the source of the ROE is not. At the top it is earned; at the bottom it is, to a large degree, borrowed.

That is the entire thesis, and it needed no hypothetical to make it. The leverage didn't create profit out of nothing — it spread a thin layer of asset-level return across a small equity base until the equity-level return looked respectable. ROA is the only column in that table that refused to play along.

One company, watched over four years

Stack the forces on a single name and the mechanism becomes impossible to miss. Take Adani Green Energy — and read it as a case in arithmetic, not a verdict on the company.

Between FY23 and FY25 its total assets grew from about ₹66,900 crore to ₹1,10,764 crore — roughly ₹44,000 crore of new asset base, funded largely by borrowings that climbed from about ₹54,200 crore to ₹80,040 crore over the same window (Inve data, FY23–FY25 balance sheets). That is a business building furiously. And through all of it, ROA barely moved: 1.5% in FY23, 1.4% in FY24, 1.8% in FY25. The asset base ballooned by two-thirds; the return on each rupee of it stayed pinned near 1.5%. ROE, meanwhile, swung from 13.3% to 16.5% — because a thinner and thinner equity sliver was being asked to carry a heavier and heavier asset base.

Three forces are converging on this one case, and that is exactly why it teaches well. First, a capital-intensive build whose assets earn a low single-digit return by their nature. Second, leverage stacked high enough — ₹6.59 of debt per rupee of equity — to lift the equity-level number into respectable territory. Third, and this is the part a screener will never show you, a management commitment about that leverage that quietly went silent.

On the Q2 FY26 call, management guided net debt to run-rate EBITDA of "4x to 5x for the next 2-3 years," noting the operational ratio was then 4.4x. By the Q3 FY26 call the disclosed figure had moved the other way: "The company's net debt stands at INR 76,000 crores," at "an overall debt to EBITDA run rate of 5.6x" (Adani Green Q3 FY26 concall, parsed by Concall AI) — above the guided ceiling, with management now saying it expected to stay in that higher range. By Q4 FY26, the ratio was not addressed at all. Inve's Promise Tracker records that leverage-discipline commitment as having quietly gone silent — guided as a range, breached, then dropped from the deck.

This is where the screen and the concall record have to be read together. The screen showed a flat ROA against a rising ROE for years — the financing-flatters-equity signature, visible in the numbers alone. The concall record showed management putting a number on its leverage discipline and then declining to repeat it once the number moved the wrong way. Either signal is suggestive. Together they are a far sharper question than either alone: is the equity return being earned by the assets, or manufactured by the balance sheet — and is management still willing to be measured on the leverage that makes the difference? (Illustration of how ROA and the guidance record interact; not a recommendation on Adani Green.)

What is a good ROA, and how does sector change it?

There is no universal threshold, because asset intensity differs enormously by business model — and ROA is very sensitive to it. The FY25 table above already shows it: Cipla's 14.1% and Power Grid's 5.8% are not a quality ranking, because a pharma company and a national transmission grid are not in the same physical business.

Asset-light businesses — software, consumer brands, services, much of pharma — need few assets to produce profit, so they post high ROA, often into the teens. Cipla's 14.1% and Dr Reddy's 11.9% (Inve data, FY25) are characteristic. For these, a low ROA is a genuine warning.

Asset-heavy businesses — manufacturing, infrastructure, utilities, telecom — carry enormous fixed-asset bases, so even excellent operators show ROA in the mid-single digits. Power Grid at 5.8% and Bharti Airtel at 6.0% (Inve data, FY25) are not weak operators; they are running businesses where each rupee of revenue demands a great deal of installed asset. Judging a transmission utility against a software firm on raw ROA is meaningless; judge it against its own history and its direct peers.

Banks and financials are a special case. Their "assets" are loans, and they operate at very high leverage by design, so they run very low ROA — often around 1–2% for a healthy bank — alongside high ROE. (In our FY25 set, HDFC Bank's ROA was about 1.7% and ICICI Bank's about 2.1%, both on healthy double-digit ROE — Inve data, FY25.) For financial companies, ROA is read on its own scale — a bank doing 1.8% ROA may be best-in-class — and alongside NIM, GNPA, and capital adequacy.

The consistent rules: compare ROA only within a sector, weight the trend over the absolute level, and treat a steadily rising ROA as the signature of a business genuinely getting more productive with its capital. A falling ROA against a rising ROE is the combination that should make you look harder — it almost always means leverage is doing work that the assets are not. For a closer look at the same divergence read through a third lens, see what the gap between ROCE and ROE reveals.

Where we could be wrong

The honest counter-argument deserves to be made at its strongest, because for the right kind of business it is correct.

Some businesses should run high leverage and low ROA, and punishing them for it is a category error. A regulated transmission utility like Power Grid earns a regulator-set return on a giant rate base; debt is the cheapest way to fund that base, and a 5.8% ROA with a 16.8% ROE is not financing theatre — it is the intended design of a low-risk, contracted-cash-flow business. A renewable developer signing 25-year power-purchase agreements has, in principle, a similar logic: predictable contracted revenue can responsibly support a lot of debt, and building fast while capital is available may be the right strategic call even if it pins ROA low for years. On that reading, Adani Green's flat ROA is not a warning at all but the expected shape of a long-duration infrastructure build, and the leverage is matched to long-dated, contracted cash flows.

We think that defence is real and it is why the ladder table is sorted as a spectrum, not a blacklist — REC, a lender, sits near the bottom by design, not by failing. But it does not dissolve the article's point; it sharpens it. The defence rests entirely on the durability and contracted nature of the cash flows that service the debt — which is precisely the thing management was putting a number on when it guided 4x–5x leverage, and precisely the number that went quiet once the actual figure crossed 5.6x. The metric (ROA) tells you the equity return is leverage-dependent; it cannot tell you whether the leverage is prudent. For that you need the cash-flow durability and the management's own willingness to keep being measured on it. The ratio raises the question. It does not close it.

Where ROA meets management's capital story

ROA is, at bottom, a verdict on capital allocation — and capital allocation is exactly what management commits to and gets graded on every quarter. "We'll improve asset utilisation," "returns on the new capex will exceed our cost of capital," "we'll hold leverage in this range" — these are all claims about ROA, made on the concall and rarely checked afterward.

This is the gap the screen can't see. Across the commitments tracked on Inve — more than 13,000 of them, over 1,500-plus listed companies — barely half are delivered as guided, and roughly 35% of companies have at least one commitment that simply went silent and was never mentioned again (Inve data, as of 2026-06-12). Capital-allocation and return-improvement guidance sits among the classes most prone to that quiet disappearance — the Adani Green leverage commitment above is one specific, parsed instance of it. A management that spends heavily on new capacity and guides that "returns will follow" is making an ROA claim; if ROA then drifts sideways for years while the commitment vanishes from the deck, the screen shows you the flat number but not the abandoned commitment. You have to track the commitment to connect the two.

Inve's Promise Tracker is built for exactly that — holding management's forward commitments next to what later happened, quarter after quarter, across a whole portfolio, which is the one thing no investor can do by hand for 10 or 15 stocks. Read together, ROA and the delivery record become a sharper filter than either alone: ROA tells you whether the asset base is actually earning, and the commitment record tells you whether management keeps facing that question or quietly stops mentioning it.

How to use ROA in your own analysis

A workable sequence:

1. Compute ROA on average total assets for five to seven years. The trend is the signal; a single year can be distorted by a large acquisition or asset write-down. (Adani Green's 1.5% → 1.4% → 1.8% over FY23–FY25 is a trend; one year would have told you little.)
2. Compare ROE and ROA side by side. A large and widening spread means leverage is amplifying equity returns — confirm with the Debt/Equity figure and interest coverage, exactly as the FY25 ladder above does.
3. Sector-anchor the benchmark. Read a manufacturer against manufacturers and a software firm against software firms; never across business models. Cipla's 14% and Power Grid's 6% are both fine — for what they are.
4. Watch ROA through a downturn. A business whose ROA holds up when profits are pressured is genuinely productive; one whose ROA was always thin was relying on leverage.
5. Cross-check the capital story. If management has guided to better returns on capital — or to a leverage ceiling — check whether they keep reporting against it. The KPI Screener ranks return metrics across peers with trend and confidence flags so the comparison stays investment-meaningful.

Frequently asked questions

ROA does not flatter and cannot be borrowed into looking better than the business is. That is precisely why it is less popular than ROE and more useful — it strips away the financing theatre and asks the plain question of whether the assets earn their keep. The owner's question is the one the FY25 table forces: if two companies show me the same return on my equity, which one is earning it on its assets and which is renting it from its lenders — and when the cycle turns and the lenders want more, which return survives? Watch the spread between ROE and ROA, judge ROA against its own sector and its own history, and pair the number with what management has actually committed to on returns and leverage. A business whose ROA rises steadily, whose ROE doesn't outrun it on borrowed money, and whose management keeps facing the question is a rare and durable thing.