Recurring vs Non-Recurring Earnings: What to Forecast
Recurring earnings come from the ordinary operation of a business and belong in the base a forecast is built from. Non-recurring earnings come from an event nobody expects to happen again, and they do not. The test is cause, never size: a large recurring cost and a small one-time gain sit on opposite sides of it.
The rule is easy to agree with and hard to apply. The separation that follows rests on three things settled elsewhere. Quality of earnings as a subject named both sides and set the order of the work. The accounting layer settled what a note to the accounts is and how an amount reaches the statements. And forecast construction, covered separately, is the only reason a base matters at all: a base is not a scorecard, it is the number next year gets multiplied out from.
Three amounts sit in the notes to a published set of accounts. Two of them are charges and one of them is a benefit. Before any of them is examined, which is the reader most likely to miss?
What actually decides which side an amount belongs on?
Ask what produced the amount. Not how big it was, not whether it looks unusual against last year, not whether the commentary drew attention to it. Just the cause. If the cause is the business doing the thing it does, the amount is recurring. If the cause is an event that happened once and is not expected to arrange itself again, the amount is non-recurring.
Consider an ordinary household. A salary arrives every month, so it is recurring, and next year can be planned around it. The electricity bill arrives every month too, and it is recurring even though the summer figure is nearly double the winter one. Gold bangles sold to pay for a wedding are not recurring, and a budget built on the assumption that money appears again has quietly planned on selling something the household no longer has.
Sorting by size is the wrong test, and it fails in two opposite directions at once: a large recurring cost gets pulled out because it stands out, and a small one-time gain gets left in because nobody noticed it. Both mistakes are ordinary. The first one is loud, the second one is silent, and the second one is the more expensive of the two because it moves the base upward.
What sits on the recurring side, including the things that jump about?
Everything produced by running the business. Revenue from selling what the business sells. The cost of the materials it turns into that. The people it pays. The advertising it buys, the lorries it hires, the electricity its factories draw. For Sarvani Coatings Limited, an invented listed maker of decorative paints and industrial coatings, year three is a revenue line of Rs 2,415 crore, a materials cost of Rs 1,304 crore that runs at 54.0 per cent of that revenue and leaves a gross margin of 46.0 per cent, and other expenses of Rs 460 crore. All of it is recurring.
Here is the part people get wrong. Recurring does not mean steady. Sarvani Coatings' advertising and sales promotion of Rs 121 crore in year three could be Rs 90 crore next year or Rs 150 crore the year after, depending on what it decides to launch, and it stays firmly on the recurring side the whole time. Its third quarter is the festive one and its second carries the monsoon, so its quarterly result swings as a matter of course. None of that swing is an event. The swing is the ordinary weather of the business.
Irregular and non-recurring are two different words, and a reader who treats variability as evidence of a one-time event will strip the ordinary volatility out of a business and forecast something smoother than the thing itself. The cost of that is real, not tidy. A forecast that is too smooth understates how bad a bad year can get, and how bad a bad year can get is exactly what a person deciding anything needs to know.
A cost appears in the accounts every single year, and it swings by about half its own size from one year to the next. Which side does it belong on?
What sits on the non-recurring side, and is taking it out calling it fake?
The non-recurring side holds amounts caused by an event: money recovered under an insurance policy after a loss, a charge taken to close down a site, a gain when a piece of land is sold, a provision released because the claim behind it went away. Each of them is a thing that happened. None of them is a thing the business does.
When a genuine one-time gain is taken out of a base, is that a claim that the money was not real? No. A genuine one-time amount is real money that was really received or really spent, and removing it from a base is a statement about next year rather than a statement about this one. The cash from the insurance recovery is in the bank. Shareholders can be paid out of it. The recovery went through the accounts correctly and nobody is disputing it.
The claim being made in removing it is narrower and duller: this is not expected to happen again, so it will not be multiplied out for the next five years. That is all. The street vendor who found a dropped five hundred rupee note this morning genuinely has five hundred more rupees. He would still be foolish to budget his month on finding one every day.
A genuine one-time gain is taken out of the base the forecast will run from. Is that a claim that the money was not real?
Where does each side get disclosed, and how much work does that make?
Finding the two sides is the difference between a five minute job and a two hour one. The recurring business is right there on the face of the statementsThe main numbered lines of a published income statement, balance sheet and cash flow statement, as opposed to the explanatory notes printed after them.. Revenue, materials, employee cost, EBITDAEarnings before interest, tax, depreciation and amortisation. A profit measure taken partway down the statement, before the financing and asset-life lines., profit. The recurring business can be read in a minute.
The one-time amounts, almost always, are not there. Each one is inside the notes to the accountsThe explanatory statements printed after the main financial statements, which break down what the summarised lines are made of. Prepared under the same standards as the statements themselves., and frequently inside a larger caption rather than beside it. Sarvani Coatings' published other incomeA catch-all line for income that did not come from selling the main product: interest earned, dividends, gains on disposals and similar amounts. for year three is Rs 38 crore. The insurance recoveryMoney received from an insurer to compensate for a loss already suffered. When and how much of it is recognised belongs to the accounting standards. of Rs 9 crore is sitting inside that Rs 38 crore, and there is no line on the face that says so. Its other expenses for year three are Rs 460 crore. A restructuring chargeA cost booked for reorganising an operation, such as closing a plant or ending a product line, covering redundancies, write-downs and exit costs. of Rs 6 crore is inside that, and so, going the other way, is a provision write-backThe reversal of an amount previously set aside for an expected liability, released back through the accounts when the liability turns out to be smaller or to have disappeared. of Rs 4 crore.
Every one of the three amounts is buried inside a caption rather than printed beside it, so skipping the notes leaves a reported result that cannot be taken apart. The separation is not hard once the disclosure is in hand. Getting the disclosure is the work.
Where in Sarvani Coatings' published documents do these three amounts actually appear?
How does the separation run on a real year, item by item?
Take year three at Sarvani Coatings Limited and work all three amounts. Each one gets three fields, and all three fields matter: what caused it, which way it points, and where it sits in the ladder. The third field is the one people forget, and it decides which figures the amount can possibly move.
| Amount, year three | Cause | Direction | Where it sits | Verdict |
|---|---|---|---|---|
| Insurance recovery, Rs 9 crore | A single settled claim | Benefit | Inside other income, below the EBITDA line | Comes out |
| Restructuring charge, Rs 6 crore | One site being closed | Cost | Inside other expenses, above the EBITDA line | Goes back in |
| Provision write-back, Rs 4 crore | A claim that went away | Benefit | Reduced other expenses, above the EBITDA line | Comes out |
Two of the three point the same way as each other and the opposite way to the third. A one-directional habit goes wrong here for exactly that reason. And notice that other income is not part of how EBITDA is built, so the Rs 9 crore recovery is the only one of the three below the EBITDA line.
Start at the EBITDA line, where only two of the three can reach
Year three closed with reported EBITDA of Rs 446 crore. Against the same year's revenue of Rs 2,415 crore that works out at a margin of 18.47 per cent. Other income is not in the build, so the insurance recovery cannot touch the EBITDA line. So only two amounts are in play up here, and the two-way test takes both: put the Rs 6 crore charge back because it is not ordinary, then lift out the Rs 4 crore write-back because it is not ordinary either. Start at Rs 446 crore, add the six, subtract the four. The arithmetic lands on Rs 448 crore, the year three figure for the ordinary operating business once both directions have been given the same treatment.
Get into the habit of reaching the same figure from the cost side as well. Other expenses were Rs 460 crore for the year. Lifting the charge out drops them to Rs 454 crore, and restoring the write-back to where it belongs pushes them up to Rs 458 crore. Rebuild the line from there, taking Rs 1,111 crore of gross profit down by Rs 205 crore of employee cost and then by that Rs 458 crore, and Rs 448 crore appears a second time. Same answer, opposite direction of travel.
The insurance recovery of Rs 9 crore sits inside other income. Does taking it out change the EBITDA figure for year three?
Year three other expenses were Rs 460 crore as published. Run the two-way test on both amounts sitting inside that caption. What is the ordinary operating cost line?
Now the profit line, where all three are in play
Reported profit before tax for year three is Rs 371 crore. The notes themselves describe two of the amounts as non-recurring. Take out the Rs 9 crore recovery, add back the Rs 6 crore charge, and the result is Rs 368 crore. Rs 368 crore is the underlying figure for the year, and after tax at the published effective tax rateThe tax charge actually reported for a year divided by the profit before tax for the same year, rather than the headline rate written in law. of 25.1 per cent it is about Rs 275.6 crore, or about Rs 11.48/- a share on 24.00 crore shares. Reported earnings per share for the same year were Rs 11.58/-.
Up at the EBITDA line the two-way test removed a write-back as well as a charge. What happens to profit before tax when exactly that same test is carried down to the profit line?
The sharpest point in the whole separation is that two different tests are in play. The Rs 368 crore figure removed the two amounts the notes labelled. The Rs 448 crore figure at the EBITDA line removed three directions of amount, including the write-back that carried no label. Carry the two-way test all the way down and the Rs 4 crore write-back comes out at the profit line too, giving Rs 364 crore before tax, about Rs 272.6 crore after tax and about Rs 11.36/- a share.
So year three has three defensible profit figures: a reported Rs 371 crore, an underlying Rs 368 crore with the two labelled amounts removed, and a two-way Rs 364 crore with every amount in the note treated the same way. The whole of the difference between the second and third is method, not arithmetic. Both figures stand, and what matters is that the test actually run is named. The Rs 368 crore result was never symmetric, so describing it as though it were is the error that does the most damage.
What does each side actually do to a forecast?
The recurring side becomes the base and gets a growth rate applied to it. Applying a growth rate is its entire job. The non-recurring side gets a different job: it is removed from the base and written on a list, with the year it happened beside it. Not deleted. Listed.
The list is not bookkeeping tidiness. The list is what makes the next release readable. A second appearance of the same kind of amount is only visible to a person who wrote down the first one, and a reader who removed something last year without recording it will meet its twin next year with no memory that they have seen it before. The list is the memory.
Who decides which side an amount sits on, and what can be checked?
Management applies the label. Ravindra Setlur, the chief financial officer of Sarvani Coatings Limited, and the people working to him decide what the note says and which amounts get described as one-time. The labelling decision is not visible from outside, and why anything was labelled the way it was is not answerable from the published documents. Pretending otherwise is where research goes wrong.
Still, three things are checkable from the published documents alone, with no access, no call and no relationship. The checks are whether the amount was disclosed at all, whether the same kind of amount appeared in an earlier year, and whether the labelling was applied to amounts running in both directions. Those three, and honestly not many more.
Why must the separation run in both directions?
Because the arithmetic is not symmetric in its consequences. Removing a charge raises the figure. Removing a benefit lowers it. If only the first is ever done, every single adjustment pushes the base up, and after enough of them the base describes a business better than the one actually in front of the analyst.
Run it on year three and look at what happens. The commentary described only the Rs 6 crore restructuring charge as one-time. Remove that charge, leave both benefits inside. Profit before tax becomes Rs 377 crore. Rs 377 crore sits above the reported Rs 371 crore, a strange place for a normalised number to land on a year the reported figure was already flattered by two one-time benefits. After tax that is about Rs 11.77/- a share against a reported Rs 11.58/-.
The mistake that hides inside three defensible decisions
Meghna Iyer separates Sarvani Coatings' year three. She removes the Rs 6 crore restructuring charge because the commentary described it as one-time. Nothing she was reading flagged either benefit, so the Rs 9 crore recovery and the Rs 4 crore write-back both stay where they are. Her underlying figure comes out at Rs 377 crore before tax, higher than the reported Rs 371 crore, on a year the reported number already carried Rs 13 crore of benefits she did not touch.
Now repeat that habit for three years running. The underlying series drifts steadily above the reported series with no reconciliation written down anywhere. Nobody catches it, and here is why: each individual decision was defensible. A reviewer reading one year at a time sees one removal, agrees with it, and moves on. The drift is only visible to somebody holding all three years at once, and by then it is inside a model.
The cost is a base that no longer describes the business, carried into a forecast that then looks pleasingly conservative because it grows slowly from a starting point that was already too high. The fix is not clever: every amount in the note gets the same test in both directions before any one of them is removed.
Over three years an analyst has removed three charges from the base and no benefits at all. What has happened to that base?
What becomes of an amount that cannot be classified at all?
Sometimes the note gives the amount and not the cause. The note says an expense of some size is included and stops there. The disclosure does not say what caused it, so there is genuinely no telling whether the amount came from the business operating or from an event.
The instinct is to decide anyway. Resist it. The amount stays out of the base, goes on the list with a note saying the cause was not disclosed, and widens the range being worked in rather than being guessed onto a side. A reader who forces every amount onto one side or the other has manufactured a precision the disclosure never supported, and the model that comes out the far end looks more certain than the evidence behind it ever was.
Leaving an amount unclassified feels like a failure, and it is not. An unclassified list is part of the output of this work, not a gap in it. Handing somebody a base plus a short list of amounts that could not be placed is a more honest deliverable than handing them a single tidy number that quietly absorbed all of them.
What somebody actually does with this on a Tuesday afternoon
An analyst covering Sarvani Coatings runs the separation once a year and keeps the list forever. The list is the only asset that compounds. When the year four notes land she does not start fresh: she reads them next to three years of amounts she already wrote down, and a second restructuring charge in that context is a completely different object from a first one.
A lender does the same work for a different reason. The lender is sizing a facility against the operating result the business can be relied on to produce, so an insurance recovery that turned up once is worth nothing to it, whatever it did to the reported profit. Its version of the base is deliberately meaner than the analyst's.
And a person choosing where to put their own savings is doing a plainer version of the same thing when they ask whether last year's result was a normal year for this business. Whether last year was a normal year is the whole question. The separation is not an accounting exercise, it is the step where the analyst decides which parts of last year are worth expecting again.
A note discloses an amount but does not say enough about it for the amount to be classified. What is the correct treatment?
Which body sets what a filer must actually disclose
Nothing above this line is jurisdiction specific. The cause test asks what produced an amount and not which paragraph permits it, so it works the same way in any reporting language. Jurisdiction decides the second half of the work: what a listed maker in India is obliged to publish alongside a result, how much detail a note must carry, and what a person writing a research view is obliged to do with it.
Filing deadlines, disclosure thresholds and materiality limits are matters for the bodies that set them. Here is where each one actually lives.
- Disclosure obligations attaching to a published result, and the conduct rules for a person writing research on it, sit with the Securities and Exchange Board of India at sebi.gov.in. Read the current text before relying on any timing or limit.
- The filed result itself, with whatever notes travelled with it, is at nseindia.com and bseindia.com. Both carry the filing, and occasionally one has an attachment the other has not yet posted, so a reader who finds nothing on one should try the other before concluding the disclosure was never filed.
- How a provision is recognised, how its release is treated and how a recovery under an insurance policy is measured belong to the accounting layer, routed to the Institute of Chartered Accountants of India at icai.org and, for the Companies Act path to the notified Ind AS, to the Ministry of Corporate Affairs at mca.gov.in.
The Ind AS standards themselves are covered under financial accounting.
Where to check every one of these routes yourself
| Body | What that source settles | Site | Confirmed on |
|---|---|---|---|
| Securities and Exchange Board of India | What a listed maker must disclose alongside a published result, and the conduct rules binding a person who writes research on it | sebi.gov.in | 28 August 2026 |
| National Stock Exchange of India | The filed annual and quarterly results of a listed maker, together with the notes attached to them | nseindia.com | 28 August 2026 |
| BSE Limited | The same filings on the second exchange, worth checking when one exchange has not yet posted an attachment | bseindia.com | 28 August 2026 |
| Institute of Chartered Accountants of India | How a provision is recognised, how a release of one is treated, and how a recovery under an insurance policy is measured | icai.org | 28 August 2026 |
| Ministry of Corporate Affairs | The Companies Act route to the notified Ind AS under which a set of published accounts is prepared | mca.gov.in | 28 August 2026 |
Sarvani Coatings Limited, Nandivarman Paints Limited, Kesaria Surface Solutions Limited, Thottam Chemicals Limited, Ravindra Setlur and Meghna Iyer are invented.
Educational material. Not advice on any investment, tax, budget or market position.