Accounting Red Flags: The Patterns Worth Investigating

An accounting red flag is a pattern in published figures that raises the chance something needs explaining. A flag is an instruction to look further and never a finding. Most flags resolve into an ordinary business reason, so the work a flag creates is a specific question and a place to look for the answer, not a conclusion about the company.

A red flag comes one step after the checks. A figure looked odd, the comparison was run, and what remains is a pattern and the question of what may properly be done with it. Three things sit underneath the work that follows. The quality of earnings checks raise a flag in the first place. The frequency test, the two way adjustment and the disclosure tests settle what counts as recurring. The recognition and measurement rules behind every pattern named here are covered in the accounting material.

So what is an accounting red flag?

Consider the kirana shop at the end of a lane. One month the storeroom behind the counter is stacked to the ceiling, twice as full as it has ever been. The stack is a pattern. The stack is not yet news. The cause could be a supplier who offered a discount for a bulk order, a wedding season the shopkeeper is stocking up for, or goods he bought and cannot sell. The stack does not say which. The stack has earned one question, asked of one person, about one thing.

An accounting red flag is exactly that stack, expressed in published figures. The published version is a pattern that raises the chance something needs explaining. A red flag is a statement about where the next hour of work should go, and about nothing else. The word flag is doing precise work here and it is worth pausing on: a flag marks a place. A flag does not describe what is buried there. Planting one on a hillside tells the next person where to dig, and it tells them nothing whatsoever about what they will find.

The useful output of a flag is therefore never an adjective. The output is a question with a name and an address: what specifically do I want to know, and which published document would tell me. A reader who has produced those two things has done the whole job the flag exists to create. A reader who has produced a feeling about the company has produced nothing that can be checked.

What is a red flag not?

A red flag is not evidence of wrongdoing. A flag is not a finding. No flag is a basis for any statement about what anyone intended. The full storeroom does not establish that the shopkeeper is hiding something, and the fullest storeroom on the lane still does not.

Between a pattern and a conclusion sits evidence, and at the moment a pattern is spotted none of it has been gathered. The distance between a pattern and a conclusion is measured in evidence nobody has collected yet, and closing that distance without collecting it is the one failure worth guarding against.

Which patterns are actually worth investigating?

Seven, on the list used here. Each one is a comparison of two published figures, each one raises a specific question, and each one has a document where that question would be answered. Note what is missing from every row: any word describing what the pattern means.

Down the middle column, every entry is a question, not a suspicion. Down the first column, something more useful appears. Every pattern on that list is a comparison and never a level, so not one of them can be read off a single figure. Receivables of Rs 289 crore is not a flag. Receivables of Rs 289 crore against revenue that grew more slowly might be. The number alone carries no information at all until something is set beside it.

There is a second reason a level says nothing, and it is a reason about scale. A flag is usually small. A movement large enough to see at a glance would have been explained in the release itself, so what is left to find is faint by construction. Take working capital intensityStock on the shelf, plus customer bills not yet collected, minus what suppliers are still owed, stated as a share of one year of sales. at Sarvani Coatings Limited, an invented paint and coatings maker. In year two, the year ended 31 March, year two, working capital intensity was 13.25 per cent of revenue. In year three, the year ended 31 March, year three, it was 13.87 per cent. Plot those on an ordinary axis running from nought and the two bars are the same bar. Cut the axis to the range where the movement lives and it is impossible to miss.

Where does the accounting behind each pattern live?

Every pattern above has a mechanism sitting behind it, and none of those mechanisms is taught here. Why a provision write backMoney held back in an earlier year for a cost that came in smaller than feared, then released through the accounts. Releasing it cuts the very expense line that creating it once raised. reduces an expense line, how a timing differenceA gap between when an item counts for the accounts and when it counts for tax. A timing difference shifts the year a payment lands without changing the total across the whole of an asset's life. arises between the tax charge and the tax cheque, what accrual accountingRecording a sale or a cost in the period it happens rather than the period the cash moves. Accrual accounting is the reason a profit figure and a bank balance are two different things. does to the relationship between profit and cash: all of that is settled somewhere else, by people whose subject it is.

Two addresses cover it. Recognition, measurement and disclosure sit with Ind ASThe accounting standards Indian listed companies report under. Which standard governs a particular item is settled by the accounting material., and the Institute of Chartered Accountants of India at icai.org is where that material is published. Requirements that come from company law rather than from accounting, including what a company must do when it changes its statutory auditorThe firm appointed to examine a company's accounts and report on them to the shareholders. The examination itself is a separate subject with its own material., sit under the Companies Act, and the Ministry of Corporate Affairs at mca.gov.in is where that lives. The list above settles which comparison to run and which document answers it. How the accounting that produced the number works is covered by the accounting material.

The division of labour keeps the pattern list honest. A reader who half remembers a standard will build a flag on the half they remember, and a flag built on a misremembered rule is worse than no flag at all. Such a flag consumes the same hour and produces a question the documents cannot answer.

Why do most flags end in nothing?

The answer is one most readers find uncomfortable, so commit to a number before reading on.

A growing business funds a growing cycle: more stock on the shelf, more customer bills outstanding, and the money to carry both. Timing differences move tax payments between years without changing the total. A genuinely one time event, a fire, a flood, a factory sold, really does happen once and really is disclosed as such. Each of those produces a pattern indistinguishable, from the outside, from the pattern that a problem produces.

So a flag does something much smaller than it feels like it does. A flag raises a small probability by a modest amount. The honest response to one is therefore a question rather than a conclusion. The sensible bet on what an untested flag turns into is an ordinary explanation, and the useful move is not to bet at all but to read the note.

What does a flag oblige an analyst to do?

Three things, and firmly no more. Write the specific question the pattern raises. Name the document that would answer it. Put a date on the check, usually the next release. Those three lines are the whole obligation, and a reader who writes them has converted a feeling into an object other people can work with.

Notice what is not on that list. Not an adjective. Not a view. Not a sentence about the company at all. A flag with no question attached is an impression. An impression cannot be checked, cannot be handed to a colleague and cannot be closed. Six months later the person holding the impression still has the impression, which has quietly hardened by sitting there, and the person holding the written flag has either an answer or the second observation in a series.

How is a flag written down without accusing anyone?

The movement is set out below, and there is one line to write. The worked version follows after that.

The difference between those sentences is not tone. The second one describes something anybody can recompute from the published statements and names where the explanation would be found. The first and third describe something nobody has established: one asserts a purpose, the other asserts a future. Neither is reachable from any arithmetic that has been run.

A written claim about a listed issuer carries conduct obligations, so the wording is not a matter of politeness. The wrong sentence exposes the person who wrote it and everybody who forwards it. The practical rule is simple enough to apply while tired: a sentence that would still be true had nobody guessed at anyone's purpose can be written down. If removing the guess empties the sentence, the guess was the sentence, and it does not belong in the file.

What are the two ways this goes wrong?

The two failures look like one failure and are not. They need opposite fixes. The first is ignoring a flag. Ignoring comes from time pressure: there were nine things on the list, the release landed at four in the afternoon, and the tax note was the one that did not get opened. The fix is mechanical. Keep a list, run every row, and record the rows that produced nothing as carefully as the rows that produced something.

The second is concluding from a flag. Concluding comes from somewhere less obvious and much harder to police, the wish for the work to be finished. A pattern plus a conclusion feels like a finished job. A pattern plus a question feels like homework. An omission leaves the question open and a conclusion stops the looking, so concluding is the more damaging of the two. The analyst who wrote down a question opens the tax note next quarter. The analyst who wrote down a verdict has no reason to open anything, and has quietly stopped being able to be surprised by the company.

What happens when the whole list is run over one year?

Here is the list run over Sarvani Coatings Limited, year three, the year ended 31 March, year three, reported exactly as far as the evidence goes and not one step further. Meghna Iyer, the analyst doing it, finishes with nine checks recorded, four written questions, four named documents and no view about the company whatsoever.

The four patterns that raised a question

Four rows, four questions, four addresses. One date sits on all of them, the next results release. Notice that not one row contains a word about what any of it means.

The three checks that produced nothing, which are output too

A list that reports only its hits has not been run, it has been skimmed. So these go in the file with the same care as the four above. Cash conversion in year three runs at 1.09 times: profit after tax was Rs 278 crore, the operating cash flow behind it was Rs 304 crore, and so profit and cash are not pulling apart in this year. The quarterly splits and the two segment lines each sum to the published revenue of Rs 2,415 crore, so no unannounced restatement is sitting in the numbers. Capital spending of Rs 186 crore runs at 2.02 times depreciation of Rs 92 crore, which is high. The balance sheet carries Rs 118 crore of a coatings line not yet commissioned, so the statements answered that one themselves before anybody had to ask.

The two checks the record cannot answer either way

The unanswerable checks are the part most easily fudged, and they must not be. Only one year of item disclosure exists, and testing repetition needs at least two, so whether items described as one time appear repeatedly cannot be tested at all. The record carries no disclosure about the accounting policy or the statutory auditor in either direction, so that check cannot be tested either. An absence of disclosure is not a clean result. Both go in the file marked as unanswered rather than as passed. A check that could not be run and a check that came back clean are different outcomes, and recording them the same way is how a gap becomes an assumption.

The other half of the skill is knowing which document to open, and it is a bigger half than it looks. Four questions went to four different places, and none of them to the same place twice. A reader who raises the right pattern and then opens the wrong document has burned the hour and still has the flag.

Who actually runs a list like this, and when

A credit officer at a bank does, every year, before renewing a working capital limit. The bank is lending against stock and unpaid customer bills, so a cycle that lengthens is the bank's money staying out longer. The officer is not looking for wrongdoing and would be embarrassed to be asked about it. The officer is looking for the question to put in the annual review file. If the limit is renewed and the cycle lengthens again next year, the file already shows the question was asked once.

An analyst does it before an investment committee, for a duller reason: because somebody in that room will ask, and arriving with a written question and a named document is a different meeting from arriving with a feeling. A household does a version of it too. The shop down the lane is stocking twice as much and has taken on a second delivery boy. Nothing is concluded. The change is noticed, and on the next visit to the shop the question is how the new supplier is working out. Noticing and asking is the entire method, scaled down to a lane.

The effective tax rateThe tax charge shown in the accounts divided by profit before tax, written as a percentage. The effective tax rate is an outcome of many separate items, not a rate anybody sets. at 25.1 per cent in year three is a good example of how the professional version stays disciplined. The rate is a result, computed from Rs 93 crore over Rs 371 crore. The rate is not high or low, aggressive or conservative. The 25.1 per cent is one half of a comparison whose other half is the Rs 88 crore actually paid, and the only sentence anyone should write about it is the one that names the gap and points at the note to the accountsThe numbered explanations printed behind the statements, where a single amount on the face is broken into the parts that make it up. where it is explained.