Guidance, Forecast, Estimate and Result: Four Numbers

Guidance is what the company says before the period about measures it chose to speak on. A forecast is what a reader builds from drivers before the period. An estimate is a forecast that has been published, and can therefore be collected with others. A result is what the company reports after the period. Four quantities, four authors, four dates, and only the last one measures anything.

Underneath all four sits one distinction, and it does the whole job. A statement about a period is not the same object as a measurement of it. Three of these are statements, made before anybody knew the answer, by three different parties with three different things at stake. The fourth is a count made afterwards under a set of rules. Every difference between the four comes out of that one split.

What is guidance, on its own?

Guidance is a statement made by the company itself, before a period has finished, about measures the company selected. Every part of that sentence is doing work. The company is speaking about the company, so the speaker and the subject are the same party. The company speaks early, when nobody has counted anything. And the company speaks about measures it picked. The choice of what to guide on is itself a decision somebody took in a room.

Think of a shopkeeper telling a supplier in April roughly how much stock he expects to take through the year. He is not obliged to be right. He is obliged to have meant it when he said it, and he will be remembered by it. The shopkeeper is close to the position a listed company is in. The company version of the same promise is made in public, and it carries duties attached to a listed issuer speaking about a period that has not happened yet.

Guidance is the only one of the four where the party being described and the party doing the describing are the same. That is its strength and its weakness at once. Nobody knows the business better, and nobody has more at stake in how it sounds.

Guidance also does not have to be a number. At the start of year three, Sarvani Coatings Limited guided to high single digit volume growth and said it expected gross marginWhat share of revenue survives the cost of the materials that went into the product, stated as a percentage. The accounting layer of this library teaches how it is built. to hold near the prior year level of 44.0 per cent. Two statements, one a phrase and one a level, neither of them a promise. How such statements are pulled apart, dated and scored is a subject of its own, set out under reading guidance statement by statement. The definition itself stays narrow: a company, speaking early, about measures it chose.

What is a forecast, on its own?

A forecast is a construction made by a reader, before the period, out of drivers the reader named. A forecast is built rather than announced. Somebody decided that revenue equals units times price, wrote down what they think each of those does, and let the arithmetic produce a total. Pulling on any figure in a forecast should bring a driver away with it.

The everyday version is working out a household electricity bill for next summer. The bill is not guessed. The units the air conditioner draws, the hours it is expected to run and the tariff go in, and the bill falls out. When the tariff changes in June the bill is rebuilt that evening, and nobody is owed an explanation for the new number.

A forecast is the only one of the four its author can change this afternoon without telling a single person. That freedom is not sloppiness. Revising in private is what a private working number is for. The moment it stops being revisable in private, it has become something else.

A build for Sarvani Coatings' year three, constructed this way, gave revenue of Rs 2,415 crore, earnings before interest, tax, depreciation and amortisation (EBITDA)Short for earnings before interest, tax and the two non cash charges, depreciation and amortisation. A rung part way down the profit ladder, built in the accounting layer of this library rather than here. of Rs 446 crore and profit after tax of Rs 278 crore. How such a build is assembled, driver by driver, is taught under building a forecast from drivers. The build belongs to whoever made it and to nobody else.

What is an estimate, on its own?

An estimate is a forecast that has been published. The definition stops there, and it sounds too small to matter. Nothing inside the number changes when it is published. The drivers are the same drivers, the arithmetic is the same arithmetic. The change is that the number is now outside its author's file, where other people can see it, quote it, and gather it into a pile with other people's numbers.

Aggregation is the point. Once nine numbers about the same year exist in public, somebody can average them, quote a range, and talk about how the group of readers is positioned. None of that is possible while each of the nine sits on its author's laptop. Publication changes nothing in the number and everything about what can be done to it.

For Sarvani Coatings' year three the published set ran to nine numbers, with a mean profit after tax of Rs 268 crore, a lowest published figure of Rs 255 crore, a highest of Rs 284 crore, and a mean of Rs 11.17/- on the per share line. How that set gets assembled, weighted and revised is set out under the aggregated estimate.

One warning about the word itself. In accounting, an estimate means something completely different: a measurement judgement made inside the statements, such as how much of a receivable will never be collected. The accounting sense belongs to the accounting layer of this library. The word carries no clue as to which of the two rooms it is being used in, so the sense has to be checked each time it appears.

What is a result, on its own?

A result is a measurement made by the company after the period has ended, prepared on a stated basisThe set of definitions and accounting choices a figure was prepared under. Change the basis and the same twelve months can produce a different number without anybody being wrong. and subject to review. A result is the only one of the four where somebody has actually counted something.

But counting is not the same as observing a fact of nature. The number on a weighing scale depends on whether the shoes stayed on, whether the weighing came before or after breakfast, and whether the scale was zeroed. None of those make the reading dishonest. The shoes, the breakfast and the zeroing make it a reading taken under conditions, and the conditions have to travel with the number. A result is a measurement made under rules rather than a fact of nature. The basis therefore has to be stated, and the same year can honestly be presented more than one way.

Here is Sarvani Coatings' year three as the statements carry it, the twelve months to 31 March of year three.

The reported ladder ties from top to bottom, and revenue rose 13.9 per cent between year two and year three, a one year figure measured between two completed years. The earnings per shareThe year's profit spread across every share in issue. How that share count is set, and when it has to be restated, belongs to the accounting and listings layers of this library. figure for the same year was Rs 11.58/-. All of it is a measurement, and all of it was made on a basis somebody chose.

Set side by side, where do the four actually differ?

Each of the four now has a definition that leans on none of the others, so the four can be put in one frame. Four attributes separate them: who made it, when, what it owes anybody, and what kind of thing it is. Read down the columns rather than across, and one fact jumps out immediately.

No two of the four share all four attributes, so no two of them are ever directly comparable until the attribute being held still is said out loud. Guidance and a forecast share their date and nothing else. A forecast and an estimate share their author and their method and differ on obligation. Guidance and a result share an author and differ on everything else. The guidance against result comparison is the useful one, and also the one people run without saying so.

Which of them do people mix up, and what does each mix-up cost?

Three confusions do almost all the damage, and each has a price that can be named.

The first is treating a beat against the aggregated mean as a beat against guidance. It is not. The mean is what other readers wrote down; guidance is what management said. Nothing management promised is tested by an average of nine strangers. The cost is praising a company for clearing a bar it never set.

The second is treating a beat against guidance as a beat against an analyst's own build. Also not. If that build was more ambitious than the guide, the company can clear its guide and still land well below it. The cost is that the analyst stops checking their own work at the moment it needed checking.

The third is treating a miss against an analyst's own build as evidence that the business disappointed. The build is the analyst's own construction. Missing it is first of all news about the drivers inside it. The cost of skipping that step is that a modelling error is quietly converted into a judgement about somebody else's management.

A results commentary routinely runs at least two of these together inside a single paragraph, and the reader is left with an impression that no sentence in it actually supports.

Sarvani Coatings' year three shows the shape at its cleanest. Two things were guided at the start of the year, and they went in opposite directions.

Now add the other comparison. Set against the nine published numbers, profit after tax of Rs 278 crore came in 3.7 per cent above the mean of Rs 268 crore, and the same 3.7 per cent shows up on the per share line, Rs 11.58/- against Rs 11.17/-. So in one year the company missed its own guide and beat the aggregated mean, and a reader can write both sentences truthfully in the same paragraph.

There is a second reason the beat is weaker than it sounds. The nine published numbers stretched from Rs 255 crore up to Rs 284 crore. The Rs 278 crore that actually arrived sat inside that band, Rs 23 crore above the bottom and Rs 6 crore below the top, so at least one reader had already written down a figure higher than the year delivered.

Why is even the result more than one number?

Everything so far has treated the result as the one solid object in the set. The result is the most solid of the four, and it is still not one number. At the EBITDA line, Sarvani Coatings' year three can be written three ways, and only one of them is in the statements.

The reported figure is Rs 446 crore. Inside that year sat a restructuring chargeA cost booked for reorganising part of a business, which management usually argues will not repeat. Which costs may be treated this way is an accounting question, settled in the accounting layer of this library. of Rs 6 crore. Management added it back on the argument that it does not repeat, giving Rs 452 crore. But the same year also contained a provision write backAn amount set aside in an earlier period and released in this one because it is no longer expected to be needed. A write back flatters the current year in the same way a one off cost depresses it. of Rs 4 crore, which went the other way and flattered the year, and which the management version left in place. Taking that out as well gives Rs 448 crore.

The add back is small in itself, 1.3 per cent of reported EBITDA. The asymmetry is what matters: an unfavourable item was removed and a favourable one of similar size was not, and the Rs 4 crore between Rs 452 crore and Rs 448 crore is the whole of that one sidedness. Only the Rs 446 crore appears in the statements; the other two are constructions, and one of the two is built to look through the bad news without looking through the good.

What does a reader do when all four are on the desk at once?

Four rules, and they are all versions of the same instinct: keep things that were made by different people at different times in different columns.

The four are held apart. Guidance goes in one column, the analyst's build in another, the published set in a third, the result in a fourth. Four columns and never a merged cell. The moment two of them share a box, the comparison has begun between objects that do not match on author or date.

Each is compared against its own kind over time. The current year's guide against last year's guide shows whether management is getting more or less cautious. The current year's build error against last year's build error shows whether the analyst is getting better. Neither of those questions can be answered by looking across.

The quantity, the author and the date are named on both sides before any comparison is written down. If all three cannot be filled in on both sides, what is in hand is not a comparison but a sentence.

And never net. If a build was light on volume and heavy on realisationThe average price actually collected per unit sold, after discounts and after the mix of what was sold. Where it comes from is built in the revenue material of this sequence., nettingSetting one gap against another and reporting only the difference, so both of the original gaps vanish from view even though each carried a separate piece of information. the two gives a small error and destroys both pieces of information. Two errors that cancel are not one small error; they are two errors nobody is looking at any more.

Who actually has all four in front of them at once?

Meghna Iyer, on the morning after the filing. She is the analyst whose working method the procedures in this sequence follow, and her desk at that moment holds four separate things: the guide Ravindra Setlur gave at the start of the year, her own build from the spring, the aggregated set she can see on a screen, and the release that landed last night.

She does not write one paragraph. She writes four lines, each a complete comparison with its quantity, author and date on both sides. Volume growth, 6.0 per cent against a phrase from April: missed. Gross margin, 46.0 per cent against a guide to hold near 44.0 per cent: beaten by 2.0 points in one year. Profit after tax, Rs 278 crore against her own build of Rs 278 crore: matched at the total, drivers not yet checked. Profit after tax, Rs 278 crore against a mean of Rs 268 crore: beaten by 3.7 per cent, inside a range that ran to Rs 284 crore.

Four lines, four verdicts, no arithmetic between them. The discipline is not cleverness, it is refusing to let four different comparisons collapse into one impression. A lender reading the same release does the same thing for different reasons, caring about the result and the basis and almost nothing about the aggregate. A household does it too, without calling it that: the school says fees will rise a little, the household budgets its own figure, the neighbours swap what they have heard, and then the bill arrives. Nobody sensible averages those four.

Sarvani Coatings Limited, Nandivarman Paints Limited, Kesaria Surface Solutions Limited and Thottam Chemicals Limited, along with Meghna Iyer and Ravindra Setlur, are invented.
Educational material. Not advice on any investment, tax, budget or market position.