How Corporate Actions Affect Shares and Per-Share Metrics

Every per-share figure is a total divided by a share count. The business moves the top; a corporate action moves the bottom. So a figure can jump while nothing whatever has happened to the company. Restatement puts a whole history onto one count so the series measures one thing throughout, and a per-share series that crosses an unrestated action is not untidy but wrong.

Eleven subjects have already covered what happens to a company inside a public market, each taking a single action apart: what a buyback does and where the money comes from, why a rights issue and a buyback point in opposite directions, how a company raises equity a second time, what separates a bonus issue from a split, and how to read the announcement that says any of it is coming.

Set the five actions side by side instead of taking them one at a time, and a single common feature appears. There is exactly one, and once that one feature is held, nothing about what each individual action does has to be memorised. All of them work on the same lever, and it is the bottom of a fraction.

The lever sounds like a small idea. It is not. The mistake it prevents is the commonest quantitative error made about listed companies, it is made by careful people using correct figures, and it survives review precisely because every individual number in it is right.

Why does a per-share figure move when nothing has happened to the business?

Take any figure with the words per share in it and write it out as a fraction. Earnings per share is profit after tax over the number of shares. Book value per share is net worthWhat is left of a company's assets once every liability has been settled. It is the balance sheet total that belongs to the ordinary shareholders as a group, and it is a single figure for the whole company rather than a per-share one. over the number of shares. Dividend per share is the dividend the company paid over the number of shares. Cash flow per share, the same shape again.

Now notice what sits where. The top of every one of those fractions is a company total, and a company total is the output of a year of trading: goods sold, prices realised, costs paid, tax settled. Nothing a company announces on a Friday afternoon changes what it sold last year. The bottom of every one of those fractions is a share count, and a share count is an administrative number in a register that can be changed by a resolution and a filing.

The business moves the numerator and a corporate action moves the denominator, and the two have almost nothing to do with each other. Held apart, no action in this sequence can mislead. Allowed to blur, a denominator event gets read as though it were a numerator event. An administrative change to a register gets read as news about a company's trading.

The household version. A cake baked for four people is cut into eight slices. The cake is sitting on the table where everyone can see it. So everyone notices immediately that the slices are smaller, and nobody concludes that less cake was baked. A company's profit is not sitting on the table where anyone can see it. All most readers ever see is the slice. When the slice halves they reach for the only explanation available to them, that something went wrong at the bakery.

Which actions push the count up, which push it down, and what else moves with them?

Five actions have appeared in this sequence. Set out in one place, something becomes visible that is invisible while they are read one at a time.

Read against each other, the last column and the second are locked together and always point opposite ways. The direction of the count change alone settles which way every per-share figure moves, before any calculator is touched. Count up, per-share figure down. Count down, per-share figure up. No action anywhere in this sequence breaks the rule. The rule is not a fact about corporate actions at all. The rule is a fact about division.

The buyback is the odd one out and worth naming for that reason. The buyback is the only action here that takes the count down. A falling count on its own lifts a per-share figure without anything happening to the business. A rise in earnings per share is exactly the sort of number that gets quoted on its own, so the exception earns its keep.

Now the third column, the one people skip. Two of these actions leave the company's resources completely untouched. A bonus issue and a stock split move nothing but the register: no money comes in, no money goes out, the factory is the same factory and the bank balance is the same bank balance. The other three all move cash, in one direction or the other, and that difference turns out to decide almost everything that follows.

What does restatement actually do to a published series?

Sarvani Coatings Limited has done two of these things, and both are in its published past. At the start of year one it split its shares, reducing the face valueThe nominal amount printed against each share when it is created, used for company law and accounting purposes. It is an administrative figure and has no connection to what a share changes hands for. from Rs 10/- to Rs 2/- and turning one share into five. The split took the count from 2,40,00,000 shares to 12,00,00,000. At the end of year one it made a one for one bonus issue, taking the count from 12,00,00,000 to 24,00,00,000. Neither action brought in a rupee and neither took one out.

Now hold year one's profit after tax still, at Rs 143 crore, and ask what earnings per share was. On the 2,40,00,000 shares in issue when the year opened, Rs 143 crore over 2,40,00,000 is Rs 59.58/-. On the 12,00,00,000 shares after the split, it is Rs 11.92/-. On the 24,00,00,000 shares after the bonus, it is Rs 5.96/-.

Three completely different numbers, all arithmetically correct, all describing the same Rs 143 crore of profit from the same year of trading. That is not a paradox and it is not sloppiness. Three divisions of one number produce three answers, and that is exactly why a per-share figure cannot be left to stand on whatever count happened to exist on the day it was first printed.

So restatement steps in and does one thing. Restatement takes every historical per-share figure and recomputes it on the count that exists now. The whole series is then divided by the same number, and a series divided by one number measures the same thing at every point. Sarvani Coatings publishes year one as Rs 5.96/-, and that is the only one of the three figures that may sit in a row beside year two and year three.

Here is the part that trips people, and it is worth saying flatly. Restatement changes no company figure whatsoever. Profit after tax for year one was Rs 143 crore before the restatement and is Rs 143 crore after it. Net worth is untouched. The dividend that was actually paid, in rupees, into actual bank accounts, is untouched. Not one number in the audited accounts moves. The divisor moves, and nothing else. When somebody says a company restated its results, the two meanings are not remotely the same event, so the question to put back is which one is meant.

The everyday version is a shop that used to quote a price per dozen and now quotes it per piece. Every historical price on the board gets rewritten so a customer can compare this week with last week. Not one rupee of past takings changed. The board changed.

Which actions get restated backwards, and which count only from their own date?

One distinction decides whether any of the above applies, and it is the one most readers never get told. The distinction sits in the third column of the table above, the one about cash.

A bonus issue and a stock split change the count and change nothing else. The company has exactly the resources it had the day before: same plant, same inventory, same bank balance, same everything. A holder who had one per cent of the company before still has one per cent afterwards. Because no resources moved, there is no honest way to say that year one had fewer shares behind it than year three in any economically meaningful sense, so the whole history is rewritten as though today's count had always existed. So a bonus issue and a split get restated backwards, all the way to the start of the series.

A rights issue, a follow-on offering and a buyback are different animals. Each one moves real money in or out of the company. A follow-on offering that raises Rs 552 crore genuinely gives the company Rs 552 crore it did not have before, and the new shareholders genuinely paid for their shares. Year one did not have that money and year one's profit was earned without it. Restating year one as though those shares had always been there would credit the earlier years with capital they never had. So prior years are left exactly as published, and the new count applies only from the day it happens.

Within the year an action falls in, the count is therefore time-weighted rather than switched: shares that existed for three months count for three months. The result is the weighted average share countThe share count used to compute a per-share figure for a period in which the number of shares changed part way through, weighting each count by the fraction of the period it was in issue. How it is built is set out in the accounting sequence., and how to build one belongs to the accounting sequence. One further wrinkle: a rights issue priced below the market carries a bonus element folded inside it, and the accounting standard sets out how that part is separated and handled. The standard is named in the sources below.

The test runs in about four seconds, and running fast is why it gets run at all. One question is put to any action met: did anything enter or leave the company? If nothing did, the past gets rewritten. If something did, the past stands and only the future carries the new count.

What happens when two actions land inside the same series?

One action is easy. Two is where careful people get it wrong, and Sarvani Coatings has exactly two.

The split multiplied the count by five. The bonus multiplied it by two. So by what does the pair of them multiply the count? A great many readers reach for seven, and seven has a certain plausibility to it: two events, one worth five and one worth two, so add them up. Factors multiply, they never add, and the register settles the argument without any need for an opinion. Two crore forty lakh shares became twelve crore, and twelve crore became twenty four crore. Twenty four crore over two crore forty lakh is ten.

Using seven restates year one onto 16,80,00,000 shares instead of 24,00,00,000, and every per-share figure produced from it comes out 42.9 per cent too large. Not obviously broken. Not flagged by anything. Just wrong by a margin that would swamp anything the business had actually done.

The other half of this is easier and worth stating because it removes a worry. The order does not matter. Split first then bonus gives 2,40,00,000 times five times two. Bonus first then split gives 2,40,00,000 times two times five. Multiplication does not care which way round it is done. Both land on 24,00,00,000. The chronology of the actions is therefore never needed to compute the combined factor. The list of them is enough.

How can a restated series be known to be right?

There is a check for this and it takes fifteen seconds, a good deal less time than the arithmetic took.

Sarvani Coatings' restated earnings per share runs Rs 5.96/-, Rs 8.21/- and Rs 11.58/- across the three published years. Its profit after tax over the same three years runs Rs 143 crore, Rs 197 crore and Rs 278 crore. Growth in the total from Rs 143 crore to Rs 278 crore is 94.4 per cent. Growth in the per-share figure from Rs 5.96/- to Rs 11.58/- is 94.4 per cent.

The two rates agree, and they agree for a reason that cannot fail: when the divisor is the same at both ends, the growth in the fraction has to equal the growth in the numerator. Nothing about Sarvani Coatings makes that true. The identity holds for any three numbers divided by a constant, and holding everywhere is what makes it a reliable test of somebody else's work rather than a coincidence.

One caution catches people regularly. Run on the rounded published figures, the check takes Rs 5.96/- to Rs 11.58/- and comes out at 94.3 per cent rather than 94.4. Nothing is wrong. Rs 5.96/- is a rounded Rs 5.9583/- and Rs 11.58/- is a rounded Rs 11.5833/-, and a ratio of two roundings carries both of their errors. Run before the rounding, on the totals and the counts, the two agree to the last decimal anyone cares to carry.

Which figures can a share count never reach?

Restatement learnt properly then tends to get over-applied, and over-application produces confident nonsense. The over-reach is worth heading off.

Go back to the fraction. A figure is exposed to the share count if, and only if, the share count appears in it. Earnings per share carries it underneath. So does cash flow per share. So does book value per share, and so does the dividend expressed the same way, so all four of them move. The historical share price series carries it implicitly. So Sarvani Coatings' illustrative Rs 486/- becomes Rs 243/- after a one for one bonus and Rs 97.20/- after a one into five split. And every multiple built on a per-share figure inherits the exposure from whichever figure it is built on.

The two price adjustments show the idea from the holder's side rather than the company's, and they repay a pause. A one for one bonus takes the quoted price from Rs 486/- to Rs 243/- and hands every holder a second share for each one they had. Twice as many shares at half the price is the same money. Nothing was handed to anybody and nothing was taken from anybody.

The rights issue reaches the same place with real money crossing, and one division checks it. A holder of twenty shares at the illustrative Rs 486/- takes up the single new share the entitlement offers at Rs 350/-. The holder has put out Rs 10,070/- altogether against a finishing holding of twenty one shares. Each of the twenty one therefore stands at Rs 10,070/- over twenty one, or Rs 479.52/-. The value given up on the twenty already held and the value picked up on the one just bought are not merely close to each other. The two are the same quantity: twenty times the Rs 136/- gap between the two prices, spread across twenty one shares, or Rs 129.52/- on either side of the ledger. Enlarging a count cannot by itself move what a holder holds, so nothing whatever is left over. Round the per-share drop first, multiply that by twenty, and the two sides stop agreeing; the daylight opening up between them is no residue in the arithmetic. The gap is rounding, done a step too early, the identical fault named above when two growth rates were compared. A holder who lets the entitlement lapse is dealt with separately.

Now look at what is left. Revenue is a total. Profit after tax is a total. Capital employedNet worth plus borrowings: the whole pool of long-term money the business is running on, whoever it came from. Sarvani Coatings runs on Rs 1,726 crore of it, being net worth of Rs 1,486 crore with borrowings of Rs 240 crore standing alongside. is a total. The margin on earnings before interest, tax, depreciation and amortisation (EBITDA) is Rs 446 crore over Rs 2,415 crore, or 18.47 per cent. Nowhere in that expression is there room for a share count. Cost of materials at 54.0 per cent of revenue, the same. Return on capital employed at 20.5 per cent, the same. A ratio of two company totals has no denominator a corporate action can reach, so restating one is not a refinement but a mistake. An analyst who reaches to adjust a margin for a bonus issue is about to divide a number by something it was never divided by.

And now the caveat that most treatments of this leave out. Left out, the caveat makes the rule false. Look at the buyback. Return on equity is profit after tax over net worth, two totals, so on the rule above it should be untouchable. It is not. Buying in 0.40 crore shares at Rs 600/- takes Rs 240 crore of cash out of the company, so net worth falls from Rs 1,486 crore to Rs 1,246 crore, and return on equity moves from 18.7 per cent to 22.3 per cent.

Has the rule broken? No, and the reason is the sharpest point here. Return on equity did not move because the count moved. The balance sheet moved, and the ratio moved with it. The count never reaches a ratio of two totals, but the cash that travels with some of the actions certainly does, and the two effects arrive together and get confused with each other constantly. A bonus issue or a split moves no cash at all, and leaves every one of those ratios exactly where it was. A buyback, a rights issue or a follow-on offering will move any ratio built on net worth, on cash or on borrowings, and will do it through the balance sheet rather than through the register.

How is an unrestated series spotted when nobody has said anything?

In practice a corporate action history rarely arrives alongside a set of figures. A spreadsheet somebody built arrives instead, or a screen, or a table in a document with no notes under it. So the detection method has to work from the numbers alone, and happily it does.

Put the per-share series beside the corresponding total and look at their shapes. If the total climbs smoothly while the per-share figure lurches, breaks or reverses, the count moved and the series was not put onto one basis. The pattern has no other honest cause. A per-share figure is nothing but that total divided by something, and a smooth numerator can only produce a lurching quotient if the denominator lurched.

Take Sarvani Coatings' own numbers and build the series the naive way, using each year's figure exactly as it was first published. Year one was first reported on the 12,00,00,000 shares then in issue, giving Rs 11.92/-. Years two and three were reported on 24,00,00,000, giving Rs 8.21/- and Rs 11.58/-. So the series reads Rs 11.92/-, then Rs 8.21/-, then Rs 11.58/-. Profit after tax over the same three years reads Rs 143 crore, Rs 197 crore, Rs 278 crore.

Look at year two. Profit after tax rose 37.8 per cent. Earnings per share fell 31.1 per cent. Two numbers describing the same year, moving hard in opposite directions, and neither of them is wrong. The fingerprint is exactly that, and it is hard to miss once seen.

What should be checked before any per-share series is used at all?

All of the above collapses into four questions, in order, and running them is a habit rather than a piece of knowledge.

The second question is doing something particular. The question does not ask which action does what. The question asks one factual thing about the company, whether money moved, and lets the answer choose the treatment. So the question survives meeting an action this sequence never covered.

And notice what the third question does not require. The corporate action history is not needed, nor the announcements, nor the record dateThe single day on which a company reads its register to decide who is entitled to a corporate action. How it is fixed and disclosed is covered separately. or anything filed anywhere. The totals are all it needs, and they sit in the same table as the per-share figures nine times out of ten. So the third question is a check that actually gets run rather than one that is merely intended.