Signalling: What a Payout Change Tells the Market

A payout change carries information because reversing it is expensive. Sankalp Industrial Systems Limited, invented, lifted its regular dividend from Rs 1.80 to Rs 2.60 a share across five declarations, and every lift is a level the next declaration gets measured against. The Rs 1.00 special was declared once and sets no level. Structuring it as a special is what keeps it out of the level.

The mechanism underneath that answer is cost, and there is nothing subtler behind it. Because everybody would make it, a statement anybody could make carries no information. A statement that is expensive to make and more expensive to take back separates the companies willing to carry the expense from the ones that are not, and it is that separation, not the announcement, that a reader is reading. Everything below is that one idea applied to a five year record.

What turns an action into a signal and leaves an announcement as noise?

Consider a shopkeeper on a busy street. In January he tells every customer he is staying put for the next ten years. In February he pays two years of rent in advance and repaints the shutter with his name on it. Both are statements about the same thing. Only the second one costs him money he cannot get back, and only the second one changes what an onlooker would bet on. The words were free, so everybody says them, so nobody listens. The advance rent was not free, so it separates him from the shopkeeper next door who would like to say the same thing but cannot afford to.

A signal is an action whose information comes from its price, not from its content. The content of a dividend declaration is almost nothing: a rupee figure and a date. The price is real cash leaving the business, and, more importantly, a level that everyone who reads the company will hold it to next year. The level, not the cash, is where nearly all the information sits, and the level is the half a reader skips.

Notice what this does not require. The mechanism does not require anybody to believe anything about how people read accounts, or to have a theory of what goes on in a reader's head. The separation happens whether or not the reader is thoughtful. It happens at the point where a boardDirectors carrying the legal authority to recommend a distribution, sitting above the managers who run the business week to week. decides what the company can afford to be held to. A company that could not sustain Rs 2.60 a share, and knew as much, would be walking knowingly into a visible reversal a year later. Some would still do it. Most would not, and that is enough to make the number worth reading.

Why is a regular dividend expensive to take back, and a special one not?

Start with the arithmetic. The sum is smaller than people expect. Sankalp Industrial Systems Limited moved its regular dividend from Rs 2.40 to Rs 2.60 a share in Year 0. On a share countHow many shares are in issue at a moment. Issuing more raises it; cancelling some lowers it. of 20,00,00,000 that extra Rs 0.20 is Rs 4,00,00,000 of cash a year. Against a profit attributable to ownersWhatever survives once lenders, the tax authority and outside holders in a subsidiary have each taken their slice. of Rs 1,38,00,00,000 in the same year, that is a small sum. The company would not notice it.

Now look at what it did to the company's position. Before the declaration, the level everyone held Sankalp Industrial Systems Limited to was Rs 2.40 a share, costing Rs 48,00,00,000 a year. After it, the level is Rs 2.60, costing Rs 52,00,00,000 a year, and the company has to keep clearing that bar every single year or be seen not to. The rise cost Rs 4,00,00,000 once; the commitment it created has to be paid again every year afterwards, and that asymmetry is the whole of why a regular dividend carries information.

Suppose next year the board declares Rs 2.40 again. In cash terms the saving is the same Rs 4,00,00,000 the company just spent, and that is nothing at all. In reading terms it is a reversal, and a reversal is the one thing on a payout record that everybody notices. The company will spend the following year explaining it. The year of explaining is the cost. The cost is real, so the willingness to accept it is real information.

A special dividend is the same cash with the second half removed. A special is declared once, labelled a special at the declaration, and produces no successor. Next year's comparison starts from the regular line and the special is not in it. The missing successor is not a loophole or a trick of presentation. It is the entire purpose of having two instruments instead of one. A board that has cash to return this year and no reason to think it will have the same cash next year needs a way to hand it over without buying a permanent obligation, and this is that way.

Why does the regular dividend move in such small steps?

If a reversal is the expensive event, a company will arrange its dividend so that reversals almost never happen. The way to do that is to keep the level well below what the business can carry, and to move it in steps small enough that a bad year does not force one back. John Lintner set this out in 1956, in Distribution of Incomes of Corporations Among Dividends, Retained Earnings and Taxes, and it is the oldest useful observation about payout: companies do not set a dividend each year from that year's profit; they adjust a level slowly toward a target share of profit, and they resist cutting it almost regardless of what the year did.

Smoothing leaves two marks on the record at once, so it is visible from outside without any knowledge of the board's discussion. The rupees rise steadily, and the payout ratioTake the dividend for a period, divide by the earnings of that same period, and the answer is a share of profit rather than a sum of money. those rupees produce stays roughly still. Here is the whole record for Sankalp Industrial Systems Limited, with both marks in the same table.

The earnings per share column and the dividend column both climb. The ratio column does not climb; it wanders inside a band 1.79 percentage pointsThe gap between two percentages, counted in whole units of the scale. A move from 38.00 to 39.00 is one point, and it is not a one per cent move. wide, from a low of 37.50 to a high of 39.29, and finishes at 37.68, which is 0.18 points from where it started. Nobody chose that outcome and nobody engineered it. The narrow band falls straight out of the two growth rates: across the four steps the dividend grew 44.44 per cent and earnings per share grew 43.75 per cent, and two growth rates 0.69 points apart cannot pull a ratio far.

The share count moved once, from 20,75,00,000 to 20,00,00,000, and a share buybackCash handed back by purchasing shares from holders who choose to sell, after which those shares stop existing. executed at the end of Year minus 2 did it. Buybacks are a separate subject, settled under the buyback. The buyback matters here for one reason: the earnings per share line steps up between Year minus 2 and Year minus 1 by more than the profit line did, and a reader tracking ratios rather than rupees has to know the denominator changed.

How many rises are actually in this record, and what did each one commit to?

Count carefully. The count is where a reader first goes wrong. The record holds five declared levels and four risers between them, not five risers. Rs 1.80 is the earliest level in the record and nothing is recorded before it, so no rise into that year can be read; it is where the record starts, and it is a level rather than a movement. From there the company moved four times, each time by exactly Rs 0.20.

Read the four risers one at a time. Each is a separate statement and they are not interchangeable. The opening level of Rs 1.80 was already costing Rs 37,35,00,000 a year at the count then in issue, and that is the bar the first riser was raised from. Riser 1 took the level to Rs 2.00 and committed the company to Rs 41,50,00,000 a year at the share count then in issue. Riser 2 took it to Rs 2.20 and Rs 45,65,00,000. Riser 3 took it to Rs 2.40, and by then the buyback had cut the count, so the commitment is Rs 48,00,00,000. Riser 4 took it to Rs 2.60 and Rs 52,00,00,000. Each level is a bar the company then had to clear again, and each was cleared. Clearing them is the only reason the series reads as a series at all.

What does a company say by holding the step at exactly Rs 0.20?

Look at the last two columns of that table together. In rupees the step never changed. The same Rs 0.20 sits on a larger base each time, so in percentage terms the step shrank every year, from 11.11 to 10.00 to 9.09 to 8.33. A company that decided the dividend afresh each year, from whatever cash happened to be spare, would produce a step that jumped about; a step that repeats to the paisa says the decision was taken once and then simply kept.

Compare it with what the business was actually doing underneath. Earnings per share moved 8.33, then 7.69, then 12.50, then 9.52 per cent across the same four transitions. The earnings line is lumpy. Year minus 1 in particular is a big jump, and part of that jump is the share count having just fallen. The dividend step walked straight through all of that without noticing. A reader who understands only one thing about smoothing should understand this: the dividend line is deliberately smoother than the earnings line, and the smoothing is the policy showing through the numbers.

Why declare a special instead of simply lifting the regular dividend?

In Year 0 Sankalp Industrial Systems Limited declared Rs 1.00 a share as a special dividend, being Rs 20,00,00,000, alongside the regular Rs 2.60 a share, being Rs 52,00,00,000. Total cash out that year was Rs 72,00,00,000 and the total dividend per share was Rs 3.60.

Now run the counterfactual. The counterfactual is where the teaching sits. Suppose the board had wanted to move the same Rs 72,00,00,000 and had simply raised the regular dividend to Rs 3.60 a share. The cash out this year would be identical to the rupee. The difference is entirely in the year after. On the special route, next year starts from Rs 2.60 and the ordinary step takes it to Rs 2.80. On the regular route, next year starts from Rs 3.60, and the company has committed itself to Rs 72,00,00,000 a year for as long as it wants to avoid a reversal. The regular route therefore locks in Rs 20,00,00,000 a year more than the company actually committed to.

The special dividend buys the company the right to hand over Rs 20,00,00,000 once instead of promising Rs 20,00,00,000 every year, and the rupees leaving the business this year are the same either way. Put next to the Rs 4,00,00,000 that riser 4 cost, the special is five times as much cash and a great deal less commitment, which is a useful reminder that the size of a payout and the weight of a signal are separate things.

The shape of the announcement therefore matters more than the amount in it. A board that quietly slipped the extra rupee into the regular line and then said in the accompanying note that the rise was one off would have made a statement that costs nothing. Words are free, and the regular line would read Rs 3.60 to everybody who came to it later. Using the separate instrument is the costly, and therefore credible, way to say the same thing. The structure carries what the sentence cannot.

Where does the idea that a dividend carries information come from?

There is a well known argument, set out by Merton Miller and Franco Modigliani in the Journal of Business in 1961, that under a set of stated conditions the payout decision by itself does not change what a company is worth: the value comes from what the assets earn, and slicing the same cash differently between a dividend today and a share of a larger business tomorrow does not create anything. Dividend irrelevance is covered separately.

The same paper separates two things that get muddled. If the payout decision does not change value, then any movement a reader associates with a payout announcement cannot be coming from the slicing; it has to be coming from what the announcement revealed about the business. The dividend is read not as a source of value but as a piece of evidence, and that distinction is the reason the word signalling exists at all. A company with more information about its own prospects than the reader has takes a costly action, and the reader updates on the action rather than on the cash.

Two names, two ideas, and they are complementary rather than competing. Lintner describes how the level actually moves: slowly, toward a target share of profit, with cuts avoided. Miller and Modigliani explain why a slow moving level ends up being informative even though the payout decision itself is not creating value.

What can a payout change not convey?

Here is the uncomfortable part. The number carries information, and it is worth being precise about how little. A signal is a narrow instrument, and readers routinely ask it to carry a whole thesis.

A rise in the regular dividend indicates that the board was willing to accept a higher recurring obligation. The rise does not indicate why. The reason could be a genuine view that the earnings supporting it are durable. The reason could be a cash balance the company cannot deploy at a decent return, a much less cheerful reason for the same action. The reason could be a habit four years old that nobody wanted to be the one to break. All three produce Rs 2.60 and the record cannot separate them.

Nor is a signal a forecast. Reading a signal as a forecast sounds sophisticated, and sounding sophisticated is what makes the error hard to shake. A costly statement is evidence about willingness at the moment it was made, and it is not a prediction, not an undertaking and not a floor under next year's profit. The company can be entirely sincere in Year 0 and still face a Year 1 that makes Rs 2.60 uncomfortable. The company committed to the level, not to the world in which the level is easy.

What does this record not contain, and why is the gap left open?

An account of signalling invites an ending about what happened next. What did the share price do when the special was declared? Did the fourth riser move anything? A reader arrives with those questions, and the record holds an answer to none of them.

The reason is specific rather than squeamish. The case record behind this subject area holds one share price, Rs 90.00, at Year 0. The record holds no price for Year minus 4, Year minus 3, Year minus 2 or Year minus 1. A dividend yieldMeasured against what a share costs on the day rather than against what the business earned, so the answer moves whenever the price moves. can therefore be built for exactly one of the five years and not for the other four. No response to any of the six declarations was ever recorded for this listedShares traded on an exchange, which attaches disclosure duties that a privately held business never picks up. company, so the record holds nothing whatever about how anybody read them.

Supplying a reaction here would not be a small liberty; it would plant a fabricated fact into a shared record that many other guides read from, and every one of them would inherit it as though it had always been there. A price move invented to round off one account becomes, several steps later, a figure somebody computes a return from. So the gap is named instead of filled, and the boundary of what a signal is gets drawn where it actually falls: the signal is what the sender committed to, and what any reader then did with it is a separate question that this record cannot answer.

For completeness, the one yield the record does support: at Rs 90.00 the Year 0 regular dividend of Rs 2.60 is 2.89 per cent. The yield is arithmetic on a price and a declaration, and it measures nothing about anybody's reaction. The figure would read the same whether the announcement was met with enthusiasm, indifference or nothing at all.

How is a payout record read for its information content?

The routine is short, and each pass depends on the one before it, so run it in order. Run it on any payout history, not only this one.

Until the regular is separated from the special, every comparison between years mixes two different kinds of statement, so pass one separates them. Here that means pulling the Rs 1.00 out of Year 0 and reading the regular line as Rs 1.80, Rs 2.00, Rs 2.20, Rs 2.40, Rs 2.60. Pass two looks at the direction and the size of the risers in what is left: four risers, every one of them upward, every one of them Rs 0.20. Four identical steps are a rhythm rather than a set of separate decisions. Pass three asks what the ratio did while the rupees moved: a band of 1.79 points, so the rises followed earnings instead of stretching past them.

The three passes answer three different questions, and it is worth being explicit about which is which. A reader who runs them out of order tends to answer the third question with data from the first.

A control has to show a consequence. The consequence a reader would want here is what the market did with each declaration, and that is exactly the thing no part of this record holds.

Sankalp Industrial Systems Limited is invented.
Educational material. Not advice on any investment, tax, budget or market position.