Earnings Management: Where Discretion Ends and Distortion Begins

Earnings management covers everything from choosing one permitted estimate over another to deliberate misstatement. The term says nothing until somebody states where on that range they mean. Accounts require judgement: a useful life, a provision, a cost formula. Making them is ordinary work, not an accusation. The difficulty is that the same act sits anywhere on the range depending on intent. Intent is the one thing a reader cannot see.

Here is what sits underneath that. A set of accounts is not a measurement of the past. A set of accounts is a set of statements about the past that require several statements about the future to be made first. This year's profit cannot be stated until somebody has said how long the machines bought during it will last, and which of the invoices sitting in receivables will actually be collected. Nobody knows either of those things. Somebody has to write a number anyway. Earnings managementA loose label covering any use of accounting choice, estimate or timing that affects the reported profit figure. The label stretches from ordinary judgement at one end to deliberate misstatement at the other, so the label means very little on its own. is the loose label for what happens in the space between those two facts, and the label stretches so far that two people using it are often not discussing the same thing at all.

The pieces are already in hand. Accounting policies, estimates and errors are treated differently from one another, and a change of estimate is a forward-looking event rather than a correction of the past. From the receivables work, Anjani Stationers Private Limited raised its doubtful debt provision from Rs 3,00,000 to Rs 9,00,000, and the Rs 6,00,000 charge splits into Rs 2,23,000 that the ageing arithmetic produces on its own and Rs 3,77,000 that came from changing the rates applied. Earnings quality has already been defined, and so has the reason cash conversion is the check on it. The range those pieces sit on runs from honest judgement to misstatement, and the published evidence stops well short of the far end of it.

What are the four positions between honest judgement and fraud?

Almost every argument about earnings management is really an argument about which of four quite different things somebody means. So get the range on the table first. Take a household first. The people in it set aside money each month for a wedding two years away. In January they guess the cost from what their cousin's wedding cost last year. In March a caterer quotes higher, and they raise the set-aside to match. In July the son sees the running total, decides it looks alarming, and quietly lowers the set-aside so the number on the fridge looks calmer. In September somebody writes a figure on the fridge that no quote, no cousin and no arithmetic supports. Four acts, one household, and only the last one is lying.

Only the fourth position is unambiguously wrong, and the second and third look identical from outside. The resemblance between the second and the third is what makes the whole subject difficult rather than merely disputed. Position one is applying judgement honestly: the estimate comes from the evidence available and the assumption behind it is stated. Position two is choosing consistently inside a permitted rangeThe set of treatments a rule actually allows for one situation. Where two or more are permitted, picking any of them is compliant, so the choice cannot be settled by pointing at the rule., where two treatments are both allowed, one is picked and then kept. Position three is choosing opportunisticallyMaking a permitted choice because of the effect it has on the reported number rather than because of what the underlying facts suggest. The choice itself breaks no rule; the reason for it is the whole issue. inside that same range, where the pick is made because of what it does to the reported figure. Position four steps outside the range altogether and produces a number the facts cannot reach.

Notice what separates position two from position three. The difference is not the treatment. The treatment is the same, and permitted in both. The difference is not the disclosure. Both are disclosed. The difference is not the arithmetic. The arithmetic is identical. The only thing that differs is why the choice was made, and that difference is invisible in every document a reader outside the business will ever see. Two companies can publish the same figure, computed the same way, disclosed the same way, and sit in different places on this range, and nothing in either filing will tell them apart.

Why does accounting permit judgement at all?

Because the alternative is worse, and it is worth being concrete about why rather than treating this as a concession. Think about what a rule with no judgement in it would have to do. The rule would have to say that every machine of a certain kind lasts exactly eight years, that exactly four per cent of every receivables book goes bad, that stock is worth exactly what was paid for it whatever has happened since. Each of those statements would be applied identically to a printing press run two shifts a day in a coastal warehouse and one run three days a week in a dry one. The number produced would be perfectly consistent, perfectly comparable, perfectly auditable, and wrong about both businesses.

The uncertainty sits in the facts, not in the accounting, so removing the judgement does not make the accounts more accurate, it only makes them more precise about something nobody knows. That distinction is the heart of it. Accuracy is being close to the truth. Precision is being definite. A rule that hands down one number for every situation buys precision by giving up accuracy, and it gives that accuracy up silently. The resulting figure carries no visible sign that it was never about that particular business at all. DiscretionThe space a rule deliberately leaves for the preparer of the accounts to apply judgement to their own facts, because the rule cannot know those facts in advance. is what a rule leaves open on purpose. The person writing the rule in 2015 cannot know what a particular binding machine in a particular warehouse will do, so the rule leaves that space open.

So a reader who treats every estimate as a place where something might be hidden has misunderstood what an estimate is for. An estimate is not a gap in the rules. An estimate is the rule doing the only honest thing available: admitting that the fact is a range, requiring somebody who can see the range to pick a point in it, and requiring them to say what they assumed. The presence of judgement in a set of accounts is evidence that the underlying facts are uncertain, and nothing more than that. Anjani Stationers has to decide today what proportion of a Rs 95,00,000 receivables book will be collected, when the answer will only be known in eighteen months. Somebody has to write a figure. There is no version of that decision that does not involve judgement.

Which levers move reported profit without breaking any rule?

Four of them do most of the work, and it is worth naming them plainly so that the lines to read slowly are known. Each is first of all a decision a business genuinely has to take. Naming them shows where in a filing the movement would be visible if there were one.

The first lever is the timing of discretionary spend. A workshop can run its maintenance shutdown in March or in April, train its staff this year or next, launch the new range before the season or after it. Every one of those is a real operating decision with real consequences, and every one of them also moves which year the cost lands in. The second is an estimate changeA revision to a figure that depends on a judgement about the future, such as a useful life or a provision rate. A change of estimate is new information rather than a past error. The revision is applied from the current period forward and does not restate prior years.: a useful life extended or shortened, a provision rate raised or lowered, a residual value revised. The third is the timing of a disposal that crystallises a gain. A machine due for replacement can be sold this March or next. The fourth is classification between operating and exceptional. The split does not change profit at all, but it changes which profit figure a reader anchors on.

Every one of the four is a legitimate decision that somebody in the business has to take anyway. Naming them tells a reader where to look and tells nobody what to do. That asymmetry is worth sitting with. Knowing that a useful life is a lever does not help anybody extend one. Extending one requires a defensible reason, a note that survives being read, and a set of facts that supports it. The list does help a reader notice, in about ninety seconds, that a company's depreciation charge fell while its asset base grew, and turn to the note that has to explain why.

Ind AS 8 and Ind AS 1: which documents govern a change of estimate in India?

The difference between a range and a point is a feature of the world rather than of any local rulebook. Everything above holds wherever it is read. The documents that carry the same requirements in India are the ones named below.

What patterns would a reader look for, and why do they need several years?

Four patterns come up again and again, and the single most useful thing about all four is what they have in common. The household serves again. One month's bank statement shows what was spent. One statement cannot show that the set-aside is being adjusted whenever the total looks worrying. Adjusting the set-aside whenever the total looks worrying is a habit, and a habit needs several months to become visible. Every pattern below is a habit.

The first is several estimates moving the same way in one period. One estimate moving is unremarkable; three moving together in the same direction is a question. The underlying facts behind a useful life and behind a provision rate are usually unrelated. The second is a change of estimate arriving in a period that would otherwise have missed something visible: a covenant, a stated target, last year's figure. The third is smoothingA reported profit line that moves less from year to year than the underlying cash generation does. Because estimate movements reverse over time, they can flatten reported profit without touching cash at all., where reported profit is markedly steadier than the cash the business actually generates across several years. The fourth is reversals of previous charges turning up in weak periods. A provision taken in a good year comes back as a credit in a bad one.

Every one of the four is a pattern across time, so a single period cannot reveal any of them, and that is the strongest practical argument there is for reading several years rather than one. Each of them requires something before it can be seen at all. Estimates moving together requires a prior period to compare against. A change arriving at a convenient moment requires knowing what the moment was and what the figure would have been without it. Smoothing requires at least three or four periods before the word means anything. A reversal requires the original charge to be visible somewhere behind it. Given one year of accounts, a reader cannot form any of these observations, however carefully they read.

Three settings carry the whole of the mechanism. At the default the panel draws one year: profit after tax of Rs 30,00,000 against operating cash flow of Rs 36,30,000. A spread needs changes and one year has none, so both spread readouts refuse to compute. Move to five periods with the slider at zero, and the reported line sits exactly on the cash line. The cash line swings by Rs 14,25,000 a period on average. Push the slider to 100 and the reported line goes dead flat while cash keeps swinging by that same Rs 14,25,000, so the reported spread reads Rs 0/- against a cash spread that has not moved a rupee. At every setting in between, the total of the estimate movements across the five periods reads Rs 0/-. A movement that flatters one period has to be given back in another, and that reversal is the single most useful property of the whole mechanism. And at every setting, whichever of the two reasons is selected, the readout for why the estimates moved says the same thing.

What did Anjani Stationers' own estimates do in year two?

Now run Anjani Stationers through the pattern list honestly. Setting out what moved comes first, and deciding whether the movement means anything comes after. Three estimates are in play. The provision for doubtful debts went from Rs 3,00,000 to Rs 9,00,000, a charge of Rs 6,00,000, of which Rs 2,23,000 falls straight out of applying last year's rates to a book that genuinely aged, leaving Rs 3,77,000 that came from raising the rates themselves. As a proportion of the gross receivables book the provision went from 3.8 per cent to 9.5 per cent. Depreciation went from Rs 5,00,000 to Rs 12,00,000, a rise of Rs 7,00,000, explained by the assets bought during the year and the part-year effect of assets bought partway through it. And the inventory cost formula did not change at all.

The last three rows carry the whole of the arithmetic. Rs 7,00,000 of depreciation plus Rs 2,23,000 of provision is Rs 9,23,000 that follows mechanically from things that happened: assets were bought, and the book aged. Rs 3,77,000 is the only amount here that came from somebody changing a rate. The rate change is 29.0 per cent of the combined movement. So the first thing the pattern list asks, whether several estimates moved the same way in one period, gets a genuine yes: two of the three did, and they moved together. The second thing the pattern list asks is which way they moved. Both of Anjani Stationers' movements pushed reported profit down rather than up, and an opportunistic account predicts the opposite.

A direction with no magnitude is easy to wave away, so put a size on the direction. Take the doubtful debt charge as an operating cost. A charge of that kind ordinarily sits there. Had Anjani Stationers raised the provision by only the Rs 2,23,000 that the ageing arithmetic produces at unchanged rates, its operating profit for year two would have been Rs 45,35,000 and its operating margin 16.8 per cent. Anjani Stationers reported Rs 41,58,000 and 15.4 per cent instead. The judgement in the provision moved the reported operating margin down by 1.4 percentage points in a year when the margin was already falling, and an account in which somebody was choosing to suit the number has to explain why they chose to make a bad year look worse.

A reader can talk themselves into a conclusion at exactly this point. Be careful with what has just been established. The profit-reducing direction is genuinely awkward for the simplest opportunistic story, and both movements have documented operational causes sitting behind them: a book that aged, and assets that were bought. But notice how easily the reasoning can be turned around. Somebody determined to reach a conclusion can say that depressing a bad year is itself a plan, one that sets up an easier comparison next year. The easier-comparison argument cannot be checked against anything either. A pattern that can be read as evidence whichever way it points is not evidence at all, and noticing that in one's own reasoning is the single most valuable habit in the whole subject. The honest statement is short: two estimates moved, they moved together, they moved in the direction that reduced reported profit, and both have explanations already on the record. Nothing beyond those four things is established.

What can a reader establish, and where does the evidence stop?

Four things can be established from published documents. All four are genuinely useful, and readers routinely undersell them, so it is worth being precise about each. Whether an estimate changed at all can be established by reading the note on accounting policies and estimates against the prior year's. By how much it changed can be established too, since two balance sheets and an income statement give the arithmetic. Which way it moved reported profit can be established, and that is the check most people skip. And what the business itself said about why can be established from the note, the directors' report, and anything said on the record afterwards.

One thing cannot be established at all, and that thing is intentWhy a decision was actually taken. Intent lives in the minds of the people who took it and in conversations no outside reader ever sees, and no arithmetic on published figures recovers it.. Not partially, not by inference, not by combining enough signals until the weight of them feels sufficient. Intent is why somebody made a choice, and it lives in conversations no outside reader joined, in an internal file no outside reader sees, and in the heads of people no outside reader has met. Intent is the whole of the difference between position three and position four on the range set out at the start, and no arithmetic on any published figure reaches it. The honest position for a reader outside the business is to describe what changed and stop there.

The cost of going further falls on somebody, and the cost is worth naming. An accusation that cannot be supported is made against a controller who spent three weeks on an ageing analysis, a board that asked the right questions, and an auditor who tested the assumption and signed. None of them can prove a negative, and none of them ever fully gets the accusation back off the record. The reader loses too. A view that cannot survive one question about its evidence will be dismantled the first time somebody who knows the business hears it. After that, nothing else that reader says will be weighed the same way. Describing what changed costs nothing and holds up. Concluding costs a great deal and does not.

Who reads an estimate change, and what do they do with it?

Away from the range for a moment: three different people open the same estimates note in the same week, and none of them is reading it for the reason usually assumed. None of the three is trying to work out whether anybody did anything wrong.

A lender reads an estimate change to find out whether the number its covenant is tested against has moved, an analyst reads it to split a cost line into the part that will repeat and the part that will not, and Vaidehi Rao reads it to know what she has to be able to evidence. Watch each of them. The lender's problem is mechanical. If a facility is secured on receivables and the provision against those receivables has tripled, the collateral has been marked down by the borrower's own hand, and the lender needs to know whether that is the borrower being careful or the book genuinely deteriorating. The ageing is closer to the facts, so the lender goes to the ageing table rather than to the provision. The lender draws no conclusion about anybody's motives; it re-runs its own margin against the ageing and moves on.

The analyst's use is the one that pays. A forecast needs each cost line split into what repeats and what does not, and an estimate change is exactly where that split lives. Anjani Stationers' Rs 6,00,000 provision charge is not one thing. The Rs 2,23,000 that the ageing arithmetic produces will repeat next year if the ageing stays where it is, and will grow if the ageing worsens further. The Rs 3,77,000 that came from raising the rates is a catch-up. The catch-up repeats only if the rates are raised again, and raising them is not a thing that happens every year. An analyst who models Rs 6,00,000 as recurring has built a forecast on a one-time adjustment. An analyst who models Rs 2,23,000 as recurring and treats Rs 3,77,000 as a level shift has done the work. Neither of them needed to know anybody's motives to get there.

And Vaidehi Rao, as finance controller, is reading it from the other side entirely. Her protection is not the size of the estimate; it is the contemporaneous record behind it. The record is what the ageing table showed on the day the rate was set. The record is which assets were bought, on which dates, and what the part-year arithmetic was. The record is what the board was told, and when. A judgement with a dated working behind it is defensible whatever the number turns out to be, and a judgement without one is difficult to defend even when it was right. The record matters more than the figure. That is worth knowing on either side of the accounts, because it tells a reader exactly what to ask for and tells a preparer exactly what to keep.

Anjani Stationers Private Limited, Chitra Binding Works, the Sunrise Public School group and Vaidehi Rao are invented.
Educational material. Not advice on any investment, tax, budget or market position.