Quality of Earnings: What Repeats and What Does Not

Quality of earnings asks a single question about a reported profit: how much of it should a reader expect to see again next year. Quality of earnings separates the part produced by the ordinary running of the business from the part produced by something that happened once, checks the reported figure against the cash the same period generated, and hands back a base for a forecast rather than a verdict.

Four things are already settled before the work begins. The three statements, and the mechanism by which any item is recognised, measured and disclosed, were settled in the accounting material and are named here rather than taught. The forecast being tested was built in the forecasting material. Quality is judged against that forecast rather than in the abstract. Every rupee below comes from the published ladder, the notes and the cash flow statement of one invented issuer, Sarvani Coatings Limited, so each step can be worked on paper. And a view has to be checkable, so the work ends in a figure that can be handed to somebody rather than an impression.

What is quality of earnings actually asking, and what is it not asking?

The plainest version of the problem is a shop rather than a listed company. A cloth shop in a market street reports that it made Rs 8,00,000/- last year. A prospective buyer is thinking of taking it over, so the number matters enormously, and the obvious question follows: will it make that again? The owner then mentions that Rs 2,50,000/- of it came from an insurance payout after the godown flooded, and that a supplier let him off an old dispute for Rs 40,000/-. Neither of those is a lie. Both are in the accounts, correctly. But the shop the buyer would be running next year is a Rs 5,00,000/- shop, not an Rs 8,00,000/- one, and the number on the sheet does not say so anywhere.

The buyer's question is the whole of quality of earnings, moved to a bigger stage. Notice the two questions it is deliberately not asking. Quality of earnings is not asking whether the profit is honest. Nor is it asking whether the profit was calculated correctly. Both of those are somebody else's work, and both have their own machinery. The only question here is how much of a reported figure a reader should expect to see again. Quality of earnings is a question about repetition and about nothing else.

One boundary is set now, and it governs every block below. The notes to the accountsThe detailed statements published behind the summary lines, where a company sets out what a figure is made of. What must appear there is set by the accounting standards and the Companies Act. record that an amount was set aside, released, charged or received, and the rules deciding how each of those reaches the statements are settled in the accounting material and route to the Institute of Chartered Accountants of India. The reading picks up where the recording stops: given that the item is there and correctly there, what should be done with the total it is sitting inside?

What are Recurring Earnings, and why is only this part a base?

Recurring Earnings are the part of a reported result produced by the ordinary operation of the business. Sarvani Coatings Limited makes paint, sells it through dealers, pays for resin and pigment, pays for freight, pays a sales team and advertises. Everything that flows from doing those things again next year is recurring, and that part is the only material a forecast may be built out of.

Now the trap, and it catches more readers than any other idea in quality of earnings. Recurring does not mean stable and it does not mean large: a cost that arrives every single year at a wildly different size is still recurring, and a reader who sorts items by size or by steadiness rather than by cause will file half the ordinary running of a business on the wrong side.

The case record shows how easily it happens. Advertising and sales promotion was Rs 121 crore in year three and freight and distribution was Rs 138 crore, both inside other expenses of Rs 460 crore. Neither of those is steady. Advertising rises hard before a festive season and can be cut in a bad quarter; freight moves with fuel and with where the paint is going. Both jump about. Both are caused by the ordinary business of making and moving paint, so both are utterly recurring and both will be there next year in some size.

The household version is immediate. An electricity bill in a Chennai May and an electricity bill in a Chennai January are not remotely the same number. The bill is still a recurring cost, and a household would budget for it every month of next year. Set against that the money spent repairing the roof after the cyclone. The roof repair is large, it is one line, and putting it in next year's budget would be foolish. The test that separates them is not the size and it is not the regularity. The test is the cause.

What are Non-Recurring Earnings, and why keep a list of them?

Non-Recurring Earnings are the part of a reported result produced by an event nobody expects to repeat. In year three there are three such things sitting in Sarvani Coatings' accounts, and none of them appears as a line on the face of the statements: an insurance claim of Rs 9 crore received, a restructuring chargeA cost booked when a business reorganises itself, for example closing a line or consolidating a site. When it may be recognised and how it is measured belong to the accounting standards. of Rs 6 crore, and a provision write backA credit that appears when an amount set aside earlier turns out not to be needed, so the old charge is reversed. Its recognition is settled in the accounting material, not here. of Rs 4 crore.

The instinctive move is to strike all three out and forget them. Forgetting them is the mistake. Next year is not made of the three amounts, so all three do come out of the base. Each one then goes into a list kept year after year, with the amount and the description the issuer used.

The reason to keep the list is that an issuer describing something as non-recurring in every year of a five year record has said something, and only a kept list can reveal it. One year of one-off items says almost nothing; a company gets flooded, wins a case, closes a plant. Five consecutive years of items each individually described as one-off is a different object entirely. Nobody has been caught at anything, and no such claim should be made. A five year record yields a question that arises only because somebody kept a list, and the place to ask it is the notes and the call.

What is a Non-Recurring Item, and who decides that something is one?

A Non-Recurring Item is the individual transaction underneath the total. Not the class, not the aggregate, the actual thing: this insurance recovery, this restructuring charge, this provision released, this gain on a disposal. Each one has an amount, a description and a place in the notes.

A label attaches to an item, so an item can be tested one at a time. A total carries no label and can be tested at nothing. The item is therefore the unit of the work. An amount described as Rs 6 crore of one-off costs can be questioned: what it was, when the decision was taken, which site it relates to, and whether something similar appeared last year. A line reading Rs 19 crore of exceptional items is an amount with no handle. There is no claim on the table to test, so there is nothing to ask about.

Who decides? The issuer writes the description, and the accounting standards decide what has to be disclosed and where. The honest answer is worth sitting with. Because the label is the issuer's word, published under rules the issuer did not write, the reader recomputes rather than accepts. Whether that label survives a frequency test is covered separately.

Adjusted Earnings: which amounts came out, and who chose them?

Adjusted Earnings are any figure reached by starting from a reported number and taking named amounts out of it. There is nothing improper in the idea. If a genuinely one-off cost sits inside a total, a figure with that cost removed really is a better base than the total, and a reader who refused to look at any adjusted figure would be throwing away useful work.

But the result is the least interesting thing about an adjusted figure. Two questions matter more than the number itself: which amounts came out, and who chose them. Both are answerable, and both are answerable only from the notes. Neither is answerable from the adjusted figure, and quoting the adjusted figure on its own therefore tells a reader almost nothing.

Here is why the second question bites. Nothing whatever prevents the removals running in one direction only, and in practice one direction is the ordinary case rather than the unusual one. Sarvani Coatings' management presents an adjusted earnings before interest, tax, depreciation and amortisation (EBITDA)The rung of the profit ladder read before depreciation, financing charges and tax arrive. What the line contains was settled in the accounting material. of Rs 452 crore, being reported Rs 446 crore plus the Rs 6 crore restructuring charge added back. The addition is arithmetically correct. The Rs 4 crore write back went the other way, flattered the same line, and stayed in, so the presentation is also one-sided.

Run in both directions, the test yields a third figure. Adding back the Rs 6 crore charge and taking out the Rs 4 crore write back reaches Rs 448 crore. Three figures now exist for one year: reported Rs 446 crore at a margin of 18.47 per cent, presented Rs 452 crore at 18.72 per cent, and two-way Rs 448 crore at 18.55 per cent. The presented figure is Rs 4 crore high, or 0.9 per cent of reported EBITDA and 0.17 of a margin point.

One distinction carries the rest of the work. Nothing above says the presented figure was wrong. Rs 452 crore is Rs 446 crore plus Rs 6 crore, and that arithmetic is exact. A one-sided adjustment and a two-way adjustment differ in method rather than in arithmetic, and the method is what a reader is judging. The work is not a check on somebody's sums. The work asks whether the rule that produced the number was applied in both directions, and only rebuilding the number answers that.

Accruals: what is the gap between profit and cash actually showing?

Accruals are the difference between the profit reported for a period and the cash that same period produced. The gap exists because a sale is recorded when it is earned rather than when the money arrives, and a cost is recorded when it is incurred rather than when it is paid. The mechanism is accrual accounting, settled in the accounting material and routed to the Institute of Chartered Accountants of India.

Reading that gap is a separate job. Profit is a measured quantity that depends on judgements about timing. Cash is closer to an observation. Set side by side over enough periods, the two show something neither one shows alone.

For Sarvani Coatings in year three, operating cash flowThe cash a business actually generated from running itself in a period, before what it spent on assets. How the statement is built was settled in the accounting material. was Rs 304 crore against a profit before taxThe rung of the ladder read after depreciation and financing costs but before the tax charge. Defined and built in the accounting material. of Rs 371 crore and a profit after tax of Rs 278 crore. Cash to profit after tax is 1.09 times, so cash ran Rs 26 crore ahead of profit. The published statement shows where it came from: EBITDA of Rs 446 crore, less the Rs 54 crore the working capitalThe money tied up in stock and in amounts owed by customers, less what the business owes its own suppliers. Built and defined in the accounting material. cycle absorbed, less Rs 88 crore of tax paid.

A gap that widens across several years is a question worth carrying; one year of a gap is barely even that. One observation of 1.09 times has established almost nothing. The ratio is not good, it is not bad, and it is not a signal. One point has no shape.

How Earnings Quality Affects Research Confidence: what exactly does it move?

Quality work pays for itself at this point, and the payoff is the part most often left out of a written note. Quality work changes two things and deliberately leaves a third alone.

The work moves the base a forecast starts from, usually by a small amount. On the case record the profit before tax base moves from a reported Rs 371 crore to Rs 364 crore, a shift of under two per cent. Every amount that could not be classified is an amount whose repeat behaviour is unknown, so the range around that forecast changes width, often by a great deal. And the direction of a view does not change at all, for the work produces no view.

The second effect is the valuable one, and it is the one that goes missing: a note that revises a number and says nothing about how confident that number now is has thrown away most of what the work produced.

Think about a household deciding what it can commit to on a monthly loan payment. Two households both take home about Rs 90,000/- a month. In the first, it is one salary credited on the same date every month. In the second, it is a shop taking anywhere between Rs 60,000/- and Rs 1,30,000/- depending on the season. The point estimate is the same for both. Nobody sensible would lend to them on the same terms, and the reason has nothing to do with the average and everything to do with the width around it.

How to analyse quality of earnings, and in what order does the work run?

Five steps, and the order is not decorative. Each one uses what the previous one produced, and running them out of order produces a figure that cannot be defended.

The last step is a deliverable and not an admission of failure, and a reader who leaves it out has published a narrower range than the evidence supports. Every real set of accounts contains amounts that cannot be placed. The choice is not between knowing and not knowing. The choice is between recording what is not known and quietly pretending otherwise. Recording it sizes the range. Pretending does not make the uncertainty disappear and only removes it from view.

What does the whole order produce, run once on Sarvani Coatings Limited's year three?

Every row below is worked from the published ladder and the notes alone. A reader who works each row lands where the table lands.

Step one yields three items, and the third is the one a hurried reader walks straight past. The insurance claim of Rs 9 crore sits inside other income of Rs 38 crore, so it is nearly a quarter of that line. The restructuring charge of Rs 6 crore sits inside other expenses of Rs 460 crore. And the provision write back of Rs 4 crore reduced other expenses, lifting the very same line the charge depressed. Without it the line would have read Rs 464 crore. The write back is easy to miss precisely because it helps.

Step two is where most of the work sits, and it has to be done at both levels of the ladder or the reader ends up running the one-sidedness they have just condemned. At the EBITDA line the three figures are already in hand: Rs 446 crore reported, Rs 452 crore presented, Rs 448 crore on the two-way test. The same test then carries down rather than changing between levels.

Read rows 2 and 3 together. The distinction is the finding. Only Rs 364 crore is built on the rule that produced Rs 448 crore one line above it, and Rs 368 crore is not a two-way figure however anyone labels it. Rs 368 crore is a perfectly defensible number, and it is what stopping after the two items the notes describe as non-recurring produces. Rs 368 crore simply is not the same test, and calling it symmetric mislabels the work.

Per share, the whole thing looks small. Rs 11.58/- reported, Rs 11.48/- on one test, Rs 11.36/- on the other. Twenty-two paise separates the reported figure from the base. Twenty-two paise is not a scandal and nobody should write it up as one. The adjustment is two per cent of a point estimate, and the reason the work was worth doing is not the two per cent. The amounts being relied on are now known and separated from those that are not, and that knowledge is what sets the range.

The work finishes here. A base of about Rs 364 crore before tax, on the two-way test run throughout. Rs 368 crore written down beside it as what stopping one item early would have given. One year of a cash ratio at 1.09 times, a starting point for a series rather than a finding. And a question carried forward: what the write back was, and whether anything like it appears in the two earlier years of the record.

What does this work hand back, and where exactly does it stop?

Three things, and they are worth naming precisely because a reader who does not know what to expect will keep waiting for a fourth.

A base figure, the number a forecast starts from. A range around it, whose width is set by what could not be classified. And a written list of questions, each one paired with the place it would be answered, so somebody else can pick up the work.

The work never returns a sentence saying an issuer has high quality earnings or low quality earnings, and a reader who wants that sentence is asking for the one thing it cannot produce. Not because it would be impolite. Because the work does not contain it. The work has separated amounts and sized an uncertainty. Neither of those operations produces a judgement about a company, and sliding one in at the end would be smuggling.

Which figures can be recomputed on paper, and which have to be read at their own source?

Sarvani Coatings Limited was built for teaching, its ladder was written so that the lines tie, and the three items the reading turns on were placed in its notes by hand. The arithmetic is what matters, and the amounts travel nowhere else. Where an accounting standard is named, the route is given with it: the mechanism by which a provision reaches a set of accounts is settled in the accounting material. Periods, thresholds and rates move, so each one comes from the regulator that sets it, on the site of that regulator and on the day it is needed.