Valuation Ratios — Cheap or Expensive?

A share price on its own is a number without a scale. ₹80 is not cheap and ₹8,000 is not expensive, because the price tells you only what one slice costs — not how big the cake is or how many slices it was cut into. A company can turn one share into ten with a stock split on a Tuesday and nothing whatsoever about the business has changed, yet the price falls by ninety per cent. Any statement about value has to divide the price by something the business produces.

That division is the whole of it. A valuation ratio — a multiple — puts the market's price on top and one unit of business output underneath, and the answer reads as a plain sentence: this is what the market is currently charging for one rupee of this company's annual profit, or of its net assets, or of its sales.

Which denominator you pick is not a detail, because it decides what the ratio can see. Profit sits at the very bottom of the income statement, after interest, depreciation and tax, so a multiple built on profit is sensitive to how the company is financed and how aggressively it writes down its machines. Sales sit at the top, untouched by any of that, so a multiple built on sales is blind to whether the company makes money at all. Every multiple below trades one blindness for another, and knowing which blindness you have accepted is more useful than the number itself.

The price-to-earnings ratio is the multiple everyone starts with because its denominator is the number every business is judged on. Divide the share price by earnings per share and you get the number of years of current profit the market is asking you to pay up front for a claim on that profit stream.

A P/E of 20 is often described as a twenty-year payback. That is a useful first intuition and a bad second one, because it silently assumes profit stays frozen at today's level forever. It never does. What a P/E really encodes is the market's collective view on three things at once: how fast profit will grow, how confident it is that the growth arrives, and how much capital the company must consume to produce it. A high P/E is not a claim that the shares are dear; it is a claim that the profit line is going somewhere. Whether that claim is right is a research question, not an arithmetic one.

There are two versions in circulation and they are not interchangeable. Trailing P/E uses the last four reported quarters of actual, audited earnings — it is a fact. Forward P/E uses an estimate of the next twelve months — it is an opinion, usually someone else's, and it is almost always the lower of the two simply because the estimate assumes growth. When a screener or a news story quotes a P/E without saying which one it is, you cannot use the number.

P/E has a denominator that can go to zero, go negative, or be temporarily inflated by something that will never happen again. In each of those cases the ratio still computes — screeners will happily print it — but it has stopped describing the business.

The most obvious failure is a loss-making company. Divide a positive price by a negative EPS and you get a negative P/E, which is not a low valuation; it is an undefined one. A company earning almost nothing produces the opposite artefact: an EPS of ten paise against a ₹300 price gives a P/E of 3,000, which is not extreme expense but a denominator too small to divide by. Both cases mean the same thing — move to a multiple whose denominator survives, usually EV/Sales or EV/EBITDA.

The subtler failures are the ones that cost real money, because the ratio looks perfectly reasonable. A one-off gain — land sold, a subsidiary divested, a tax write-back — lands in profit after tax, inflates EPS for exactly four quarters, and drags the trailing P/E down to a level the operating business never earned. The other is a cyclical business at the top of its cycle, which gets its own section below because it is the most expensive trap in this entire topic.

If a high P/E is really a claim about growth, the obvious next move is to divide the P/E by the growth rate and see what you are paying per unit of growth. That is PEG. A company on a P/E of 40 growing earnings at 40 per cent a year carries a PEG of 1.0, and so does a company on a P/E of 12 growing at 12 per cent. The rule of thumb attached to it says a PEG near 1.0 is fair.

That rule of thumb has no theoretical foundation. It is a heuristic that became folklore. More importantly, PEG smuggles in assumptions that P/E at least had the decency to leave visible.

The first is whose growth number you used. Trailing growth is history and may never repeat. Forward growth is a forecast, and the forecast usually comes from the same optimism that produced the high P/E, so the two sides of the ratio are not independent. The second assumption is duration: a company growing at 25 per cent for two more years and one growing at 25 per cent for ten more years produce the same PEG and are not remotely the same proposition. The third is that PEG treats all growth as equal, when growth funded by a business earning 30 per cent on its capital is worth far more than the same growth bought with debt at a 9 per cent return — a distinction the return-ratios topic in this Academy handles properly. PEG is a conversation starter, not a conclusion.

Price to book compares market capitalisation against book value: what the accounts say the shareholders' share of the company's assets is worth after every liability has been settled. It is the multiple of the balance sheet rather than the income statement, and that changes who it suits.

It suits banks and NBFCs almost perfectly. A lender's assets are financial — loans, investments, cash — and they are carried at values much closer to what they could actually be realised for than a factory is. Book value for a bank is close to a real, checkable quantity, which is why banking analysis in India runs on P/B paired with return on equity rather than on P/E. It also suits heavy manufacturing, shipping, cement and metals, where the balance sheet genuinely holds the productive capacity of the business.

It fails, and fails badly, for asset-light businesses. Indian accounting expenses brand-building, research and software development as they are incurred rather than capitalising them, so a consumer brand's most valuable asset — the reason a shopper reaches past a cheaper packet — appears nowhere on its balance sheet. The same is true of an IT services firm whose entire capacity is people. Book value understates the real asset base, so P/B comes out at a number that looks alarming and means nothing. Buybacks make it worse: repurchasing shares reduces both cash and equity, so book value falls and P/B rises mechanically without any change in the business.

P/B is also close to useless on its own. A low P/B on a business earning 4 per cent on its equity and a low P/B on a business earning 22 per cent are different facts, because book value is only worth paying for in proportion to what the company can earn on it. Read it next to return on equity or not at all.

Market capitalisation prices the equity. It does not price the business, because whoever owns the equity also owns a set of obligations. Enterprise value fixes that, and the plainest way to understand it is to imagine buying the company outright.

You would pay the shareholders for their shares — that is market capitalisation. Then you would discover you had inherited the company's borrowings, which you must eventually repay, so add the debt. But you would also find cash sitting in the company's bank accounts, and that cash is now yours to use against the purchase price, so subtract it. What you are left with is the true cost of taking control of the operating business: market cap plus debt minus cash.

That matters the moment two companies are financed differently. Take two businesses running identical factories and earning identical operating profit, where one is debt-free and the other is carrying substantial borrowings. The indebted one pays interest, so its profit after tax is lower, so its P/E looks higher — and a reader concludes the debt-free company is the better-run business when the difference is entirely in the financing. EV/EBITDA removes the distortion from both ends at once: enterprise value counts the debt on top, and EBITDA is measured before interest is paid.

EBITDA is earnings before interest, tax, depreciation and amortisation — what the business earned before paying its lenders, before paying the taxman, and before writing down the cost of its machines. Because it strips out interest, tax rates and depreciation policy, it is the most comparable profit measure across companies with different capital structures, different tax positions and different asset ages. That is also its weakness: capital expenditure is real, and a business that must keep replacing its plant is genuinely worse off than one that does not, even though EBITDA treats them identically.

Some businesses are worth studying before they earn anything. A young platform, a company deliberately spending its margin on acquiring customers, or a manufacturer in the first years after a large capacity expansion may all report losses while building something real. P/E is undefined and EV/EBITDA may be too, so the denominator has to move up the income statement to the one line that is always positive: revenue.

Price to sales divides market capitalisation by revenue. EV/Sales divides enterprise value by revenue, and it is the better of the two for the same reason enterprise value is generally the better numerator — a loss-making business is often the one carrying debt, and ignoring that debt makes it look cheaper than it is.

The blind spot is total and worth stating bluntly: a rupee of revenue earned at a 3 per cent margin and a rupee earned at a 30 per cent margin count exactly the same. A sales multiple is therefore only meaningful between businesses you believe will end up with similar margins. Applied across business models it is not a comparison at all. The honest way to use it is as a placeholder — you are implicitly forecasting the margin the business will settle at, so state that forecast out loud rather than letting the multiple hide it.

Dividend yield turns the question around. Instead of asking what you are paying, it asks what the company hands back: dividend per share divided by the share price, expressed as a percentage. It is the only multiple on this page that measures cash actually reaching your bank account rather than an accounting quantity.

Its most important trap is arithmetic. Yield rises when the price falls, so a yield that has climbed sharply is often reporting a falling share price rather than a generous board. The second trap is sustainability: check whether the dividend is covered by free cash flow and what proportion of profit is being paid out, because a payout ratio that has crept above what the business generates is being funded from the balance sheet. Special or one-time dividends distort the trailing yield for a full year.

Price to free cash flow is the strictest multiple in common use. Free cash flow is what the business generated from operations after paying for the capital expenditure needed to keep running — the cash genuinely available to repay lenders, pay dividends or reinvest. It is far harder to manage upward than profit, because accounting judgement moves profit around while cash either arrived or it did not. Its own blind spot is lumpiness: a company part-way through a large expansion will show weak or negative free cash flow for two or three years by design, and the multiple will look terrible while the business is doing exactly what it should. Read it across a full capex cycle, not off one year, and read it alongside the cash flow statement topic in this Academy.

Numbers help, and they must tie out, so take an illustrative mid-cap called Bharat Cables Ltd — figures illustrative throughout, not a real company. Every multiple below is computed from the same set of inputs, so you can see how one balance sheet and one income statement feed eight different ratios.

Read the table down the third column and notice how little the multiples agree with each other about what kind of company this is. On EV/EBITDA it looks unremarkable. On price to free cash flow it looks demanding. On dividend yield it looks modest. None of those is the answer — they are eight different questions about the same business, and the work is in deciding which question matters for this particular kind of company.

Keep this table next to any screener output. The third column is the one people skip and the fourth is the one that decides whether the number in front of you is information or noise.

Here is the part most explanations skip. A multiple is not a measurement. It is a shorthand for a whole set of assumptions that somebody, somewhere, has already made about the business — and shorthand is only legible if you know the longhand it stands for.

Four things drive what multiple a business deserves, and none of them is visible in the ratio itself. The first is return on capital: a company that turns every reinvested rupee into 25 paise of profit compounds, and a company that turns it into 8 paise does not, so growth is worth far more to the first. The second is durability — not how fast it grows but for how many more years, which is a question about competitive position and belongs to the economic moat topic. The third is capital intensity: two businesses can report the same profit while one needs continuous capital expenditure and the other needs almost none, and only one of them ever hands cash to its owners. The fourth is the risk attached to the whole stream, which sets the rate at which future profit is discounted back.

Put those four together and it becomes obvious why an identical number carries opposite meanings across sectors. A P/E of 45 on a consumer staples business with high returns on capital, negligible capital expenditure and decades of visible demand is a statement about duration. The same 45 on a capital-goods company whose order book has just been filled by one government cycle is a statement about a moment. Neither is a verdict. Both are compressed forecasts, and the discounted cash flow topic in this Academy unpacks exactly what a multiple is compressing.

The practical consequence: never compare multiples across sectors, and be suspicious of comparisons within a sector where the business models differ. The relative and peer valuation topic that follows this one is entirely about doing that comparison honestly.

Commodity and deeply cyclical businesses — metals, sugar, shipping, chemicals, refining, some parts of real estate — break P/E in a way that is almost perfectly designed to mislead, because the ratio reaches its most attractive-looking level at the worst possible point in the cycle.

Take a second illustrative company, Deccan Alloys Ltd, again with figures that are illustrative and not those of any real company. At the top of a metals cycle, realisations are high, the plant is running flat out, and fixed costs are spread across maximum volume, so margins expand far beyond anything the business earns on average. Say it reports EPS of ₹50 while the share trades at ₹350. The P/E is 7. It screens as one of the lowest multiples on the exchange.

Two years later the cycle normalises. Realisations fall, utilisation drops, the same fixed costs are spread thinner, and EPS falls to ₹6 while the share trades at ₹250. The P/E is now 42. Nothing about the factory, the management or the product changed. The only thing that moved was where the business sat in its cycle — and the multiple was at its most inviting precisely when its denominator was least repeatable.

Analysts who work on cyclicals do not fix this by finding a better multiple. They change the denominator to something that survives the cycle: normalised or mid-cycle earnings, which is average earnings across a full peak-to-trough period rather than the last four quarters. They lean harder on P/B, because a steel plant's book value is far steadier than a steel plant's earnings. And they watch the physical drivers — capacity utilisation, the spread between input and output prices, the inventory sitting in the channel — because those turn before the reported earnings do.

Multiples are a triage tool. They tell you where to spend your reading time, and they are very good at that. What they cannot do is deliver a conclusion, because every one of them compresses a forecast into a single number and then hides the forecast.

Three habits keep them honest. First, always say what the multiple is being compared against — its own sector, the company's own history, or the broad market — because a bare number carries no information at all. Second, always pair a multiple with the quality measure that justifies it: P/E next to earnings growth and return on capital, P/B next to return on equity, EV/EBITDA next to how much of that EBITDA survives capital expenditure. Third, before treating a low multiple as attractive, actively look for the reason it is low. There is almost always a reason — a cycle at its peak, a one-off in the earnings, a governance issue, a customer concentration, a regulatory change already visible to the people who own the stock.

The corresponding discipline on the other side: a high multiple is not a warning either. It is a hypothesis that growth and durability will arrive, and the way to test it is to write down what would have to be true, then check the filings against it. That is the whole exercise. The number is the question, not the answer.