Company Filings as a Research Source: What Each Document Offers

Six figures, all six correct: which were read and which were made?

A working sheet comes before any definition. A definition would have to argue for something the reader already accepts, so the sheet can go down first. Anjani Stationers Private Limited, an invented register maker, had a second year, and here it is written out the way anybody writes one out: revenue Rs 2,70,00,000/-, profit before taxWhat is left once every expense including the cost of borrowing has come off, and before the tax charge itself. Rs 38,00,000/-, finance costsWhat the business paid for the use of borrowed money during the period. Rs 3,50,000/-, operating profit Rs 41,50,000/-, a standing bill of Rs 74,00,000/- and contribution of Rs 1,15,50,000/-.

Every one of those six is correct. Not one of them is a rounding, a guess or a slip. The question the sheet does not answer is this: which of the six could a reader walk over to a set of published accounts and physically point at?

Three of them can be pointed at. One has to be assembled on the spot out of two lines that are printed. Two are not anywhere in the document at any distance. Those last two rest on a division of cost that nobody was ever asked to publish. The sheet gives no way to tell those three cases apart. The six figures are the same colour, the same size, the same handwriting, in one column, and by the time somebody else reads the sheet the difference has vanished completely.

The sheet can be sorted without anything being looked up, and sorting it once makes a sheet like it uncomfortable to write ever again.

Secondary Research: what does working from somebody else's record provide?

The act sounds grander than it is, so it deserves plain naming first. Somebody inside a business typed a figure up. Somebody else checked it. The figure was then put out where an outsider could get at it. The reader did none of those three things, and the reader's entire access to that business runs through what that record chose to print. Reading somebody else's record is the whole of the act, and both its strength and its ceiling fall straight out of it.

The strength is repeatability. A second reader can open the same document and land on exactly the same figure as the first. The claim sounds unremarkable until it is set beside the fact that almost nothing else anybody might learn about a business behaves that way. Two people asked about a business give two answers, both honest. One document opened twice gives one answer, twice. When two readers disagree about a printed figure they have somewhere to go, and the somewhere settles it.

The ceiling is the same property seen from the other side. A record settles only what it recorded, and it recorded what the presentation rules asked for rather than what any particular reader happened to want to know. There is no amount of care, patience or rereading that gets a figure out of a document that never carried it. The ceiling is not a defect in the document and not a defect in the reader. The ceiling is the shape of the route.

Weighing a set of accounts against a transcript, a presentation or somebody's written commentary is a different question, set out under Source Hierarchy: Ranking Evidence From Filing to Commentary. The part nothing else reaches is the distance from the front of a single document to the figure finally written down, and that distance is the subject from here on.

How far from the front of a document can a figure sit?

Most readers arrive believing a filing is one flat thing, held or not held. A filing is not flat. The document has depth, and reading it is a descent into that depth. A figure sits at one of four distances from the front of the document, and its distance is a harder fact about it than its value is. Distance inverts the thing everybody cares about. The value gives what the figure says. The distance gives what it would take to correct the figure, and correcting a figure is exactly what two readers who disagree have to work out.

The four carry the rest of this guide. First, printed on the face of the statement, where the reader copies it. Second, printed in a note behind the statement, where the reader copies it after turning to the note. Third, built by the reader out of things that are printed, where the work is an addition anybody else would also do. Fourth, not in the document at all, where the reader supplies a judgement and the arithmetic then runs on top of that judgement rather than on top of the document.

Something happens on the way down. At the top the document is doing all the work and the reader none. At the bottom the reader is doing the work and the document is only supplying the ingredients. The authority of the document falls away step by step and the reader's own contribution rises to fill the gap, and a figure at the fourth distance carries almost none of the document's authority even though it was worked out with the document open.

What is actually printed on the face of the statement?

The face of a statement of profit and loss is short, and its shortness is intended. The face prints revenue from operationsThe sales of the trading business itself, kept apart from anything the business earned on the side., taken before other income. Under it comes cost of materials consumedWhat the raw material actually used up in the period cost, rather than what was bought during it., with changes in inventoriesThe adjustment that turns what a business made in a period into what it sold in that period. beside it. Then comes employee benefits expenseEverything the period cost in wages, salaries and the payments that travel alongside them. as one line. Beside it sits depreciation and amortisationThe slice of an asset's cost charged against this period rather than against the period it was bought in. as one line. Other expenses follow as one line. Then finance costs, and then profit before tax.

The names on the face are not the business's choice of words. The presentation rules require them, and a reader who has never seen this particular business before can therefore still find the line they want. A movement in one of those heads is readable precisely because somebody was required to print that head, and a quantity nobody was ever asked to present is not readable at any price and no amount of effort changes that.

The reader's move at this distance is a transcription and nothing more. Calling it a transcription sounds like faint praise and is not. The strongest thing that can be said about a transcription is that anybody else doing it lands in the same place. Repeatability is exactly the property the remaining distances lose, one step at a time.

What was put in a note instead, and why was it put there?

One step behind the face, the document opens out. The composition of employee benefits expense sits in a note. So does the anatomy of the single line called other expenses, and the note behind that line is where rent stops being mixed in with insurance, with packing, with carriage outwardWhat it cost the business to get goods to the buyer after they had been sold., and with the audit fee.

Here is the part readers get backwards, so say it in the plainest possible words. A note is not a weaker place than the face. A note is a more detailed place, holding the same money broken into its parts, and a reader who stops at the face has stopped one step short of the only breakdown the document is going to give them. Nothing about a figure becomes less reliable by sitting one step further back. Only the distance from the front changes.

The shape is familiar from ordinary life. A phone bill prints one total for the month on the front and the individual calls on the sheet behind. Nobody thinks the sheet behind is the untrustworthy part. The sheet behind is where the number actually rung at eleven at night turns up, and the front has never once carried that. A set of accounts works the same way: the face gives the total, the note gives what the total was made of, and both are printed by the same people at the same time.

So where in the document does operating profit appear?

Nowhere, and this is the moment most readers refuse to believe what they are being told. Operating profit is the single most quoted figure in the whole trade. Operating profit turns up in every summary, every comparison and every conversation about whether a business is doing well. Operating profit is not a line in the document, it never was a line in the document, and every reader who has ever quoted it assembled it themselves out of two lines that are printed.

The assembly takes one step. Profit before tax comes off the face, and for Anjani Stationers' second year it is Rs 38,00,000/-. Finance costs come off the face at Rs 3,50,000/-. The two are added. Rs 38,00,000/- plus Rs 3,50,000/- gives Rs 41,50,000/-, and that is the operating profit, sitting at the third distance because it was made rather than found.

Now the reassuring half, and it matters as much as the surprising half. An addition of two printed numbers has exactly one answer, so every reader who does that addition lands on the same Rs 41,50,000/-. The figure was built, and it is nonetheless as settled as anything copied. Handed the document, somebody in another city produces the same figure without a word passing between the two readers. Agreement without conferring is what the third distance keeps, and it is the property the fourth distance loses.

Where does contribution come from, if it is not in the document?

Contribution is what is left of revenue after only the costs that move with each sale. Working it out first requires dividing the business's costs into the ones that stand still whatever happens and the ones that rise with every extra register sold. No presentation regime asks a business to set out how its costs behave, and so the division appears nowhere in the filing, at any distance.

So the division has to come from the reader. The printed heads and the notes behind them are examined, a view is formed about which costs stand still, and a line is drawn. Contribution is then assembled on top of that line, and it is never a line item anywhere. The split is an estimate, not a disclosure, and that label has to travel with the figure everywhere it goes.

Consider what this does to two careful readers. Both open the same document. Both read every note. Both are competent and neither makes an arithmetic slip. One draws the line in one place and reports contribution of Rs 1,15,50,000/-; the other draws it slightly differently and reports something else. The two readers now disagree, both are defensible, and the document never held the answer, so there is nothing anywhere in it they can turn to in order to settle the matter. Neither reader has failed to take care. Disagreement of that kind is what the fourth distance is.

Which two distances get confused, and what test separates them?

Operating profit and contribution are both described as built. On a sheet they look identical: a name, an amount, no annotation. An addition anybody can repeat and a judgement nobody else will reproduce are both called building, and they are not remotely the same act.

There is one test and it takes ten seconds. Handed the document, must a second reader land where the first reader landed? For operating profit they must. One sum has exactly one answer. For contribution they need not. A judgement sits in the middle of the calculation, and two reasonable people can make it differently. The gap between repeating and judging is the whole difference, and it is invisible on the sheet.

The test holds better as a question than as a rule. A question survives into situations no rule anticipated. Must a second reader land here? If yes, what was built is repeatable. If they need not, something was supplied, and whatever was supplied has to be written down beside the figure or it will be lost inside it.

What do published accounts record about how a business works, and what did they never record?

The register maker's own side of the question is where the title earns itself. A set of published accounts records the size of a movement in a printed head, exactly and without argument. If employee benefits expense went up, the accounts say so and say by how much, and nobody can argue it away. Nobody was ever asked to present how a cost behaves, and so the accounts do not record it, and every question worth asking about a business sits downstream of that one absence.

The reconciliation stops meaning anything very quickly. Contribution of Rs 1,15,50,000/- less the standing bill of Rs 74,00,000/- gives Rs 41,50,000/-, the same operating profit the addition produced one step higher up. The agreement is not a check on anything. The split was drawn so that what remains after the standing bill is the operating profit the reader already had in hand, so the two could not have disagreed with each other whatever the split had been. Written out, the inputs of both routes carry the same number twice, sitting in the middle of both. The split is an estimate, not a disclosure, and no arithmetic done on top of it can turn it into one.

Now take a movement and see the two kinds of thing sitting inside one total. Anjani Stationers' standing bill rose by Rs 24,40,000/-, and that rise divides into employee benefits of Rs 6,00,000/-, the fixed part of other operating costs of Rs 11,40,000/- and depreciation and amortisation of Rs 7,00,000/-. The three sum to Rs 24,40,000/- exactly, to the last rupee, with nothing left over.

Look at where each of the three could be found. Employee benefits expense carries a printed head of its own, and so does depreciation and amortisation, so both movements are simply read. Rs 6,00,000/- plus Rs 7,00,000/- is Rs 13,00,000/-, or 53.28 per cent of the rise. No regime asks anybody to set out how a cost behaves, so the fixed part of other operating costs carries no head of its own anywhere, and it accounts for the remaining 46.72 per cent. So a total landing on the exact rupee still carries a portion nobody ever published, and the exactness is what conceals it. Arithmetic that closes perfectly is the last place anybody thinks to look. The split is an estimate, not a disclosure, and the label travels with it here as it does everywhere else above.

Here is the moment a reader leaves the document without noticing. The rise in employee benefits expense can be read, and by how much it rose can be read, and both of those are transcriptions. The instant a reader says that part of that rise stands still whether or not another register is sold, the reading has stopped and the supplying has begun. The sentence sounds the same, comes out of the same mouth and goes onto the same sheet. None of this is a failing of the document; it is the boundary of one route, and that boundary is the whole point.

What has to be written beside a figure before the sheet is put down?

Four things, and no more than four. Which document the figure came out of. Which of the four distances it sat at. Whatever was done to it, if anything was. And the date it was read. Four items are the whole annotation, they take a few seconds a line, and they are the difference between a sheet somebody can act on and a sheet somebody has to rebuild from scratch.

Take the fourth item seriously. A date is the one people leave off. A date is not decoration on a figure; it is the only thing that tells a later reader whether the line is still worth anything, and without it a correct figure and a stale figure look exactly alike. The rung tells a later reader what to do with the figure, and the date tells them whether to bother.

Two institutions stand behind the names the printed lines carry, and each governs a different part of that naming. Every criterion, every numbered standard and every date on which something took effect has to be read at the source rather than written from memory. A source with no reading date beside it is a claim that cannot be aged. A reader going to those sources writes down the day of reading.

So how does the opening sheet finally sort?

The six figures from the opening sort out as follows. Revenue of Rs 2,70,00,000/-, profit before tax of Rs 38,00,000/- and finance costs of Rs 3,50,000/- were read off the face. Operating profit of Rs 41,50,000/- was built, by adding two of those printed lines, and every reader who does that addition arrives at the same place. The standing bill of Rs 74,00,000/- and contribution of Rs 1,15,50,000/- rest on a division of cost that nobody published, so the split is an estimate, not a disclosure, and a second reader need not arrive at either.

The two margins set side by side show the sorting doing real work. On unchanged revenue of Rs 2,70,00,000/-, operating profit of Rs 41,50,000/- reads 15.37 per cent and the assembled contribution of Rs 1,15,50,000/- reads 42.78 per cent. Both are margins. Both are correct. Both sit in the same column of the same sheet. The two margins do not sit at the same distance from the document, and a reader who quotes the second one as though it were as settled as the first has quoted their own judgement back to somebody as though it were somebody else's record.

A figure without its distance beside it is not finished, and a sheet worth passing on is one whose every line states where it sat and what was done to it.