Research Coverage: What Covering a Company Commits You To

Covering a company means holding a current published view on it, and keeping it current. Coverage is a standing obligation rather than a one time act of writing: the analyst publishes after every result whether or not anything moved, publishes a change when it happens, and publishes on the day the coverage stops. The written list of names carrying that obligation is the coverage universe, and capacity fixes its length.

The requirements of a checkable view were settled in the opening sequence of this subject area, and how a quarterly release is taken apart was settled in the earnings sequence. Underneath both of them sits a fact almost nobody states out loud. A view is not a document. A view is a position an analyst is standing in, in public, with a date attached, and the date keeps moving whether or not anything is done about it. The sentence comes apart against one list of four names, one quarter of published numbers, and a blunt count of what the arrangement costs in working days.

What does it actually mean to cover a company?

Something ordinary makes the point. Consider the person in a household who tracks the electricity bill. The person who tracks the bill is not the one who read it once in April and formed an opinion about it. The tracker is the one who, when somebody asks in November, can say what the bill is doing now and whether the April verdict still holds. The moment the bill stops being opened, the April opinion does not become wrong. Worse, the April opinion becomes undated. The opinion is still sitting there in everybody's memory, in that person's voice, sounding exactly as confident as it did when the bill had actually been looked at.

Research coverage is that arrangement, formalised. A view on a company has been written and published, and by publishing it the analyst has taken on a duty to keep it current rather than merely to have been right about it once. The commitment is to the publishing, not to the view. An analyst who has changed nothing at all still has something they are obliged to produce. Read that twice. The rule surprises people who come to this work from writing rather than from research. In writing, having nothing new to say is a reason for silence. In coverage, having nothing new to say is a finding, and it has to be filed like any other.

The word doing the heavy lifting is current. A view that was correct when written and has not been rechecked since is not a correct view any more; it is an artefact. Nobody can tell the difference from outside. A published view carries the date it was written rather than the date it was last examined, and those are wildly different facts about the same document.

What is a Coverage Universe, and what fixes its size?

The set of companies one analyst or one desk keeps a current view on is written down as a list, and that list is the coverage universe. The coverage universe is not a mood, a watchlist, or a set of things somebody finds interesting. The list is a published statement of the form: these are the names on which the view is current, and a reader picking one up may rely on that.

Which raises the only question that matters about it. Why is a particular company on the list? The tempting answer is that it is important, or growing, or in a field the desk understands. The honest answer is capacity: a name belongs on the list because there is time to keep it current, and an analyst carrying more names than they can read is publishing staleness under a current date. The capacity answer is a stronger statement than it looks. Publishing nothing on a company leaves a reader with no view and they know it. Publishing a view that has not been rechecked is worse. The reader is left with a view they believe is live, and the very uncertainty that would have made them go and check for themselves has been removed.

How does a name get on the list, and how does it come off?

Both are acts, and neither is a drift. A name arrives through an initiation. The initiation is the single occasion on which the entire view is constructed from nothing: the business, the way it makes money, the structure of its field, the numbers, the peer groupThe short list of other companies chosen so a comparison actually means something. Choosing that list honestly is worked separately, under peer group selection. it will be read against, and the assumption set the future work will test. An initiation is the longest single thing an analyst writes about a company, and it happens once.

A name leaves through a discontinuation. A discontinuation is a published statement that the view is no longer being maintained. Publishing an exit sounds like administrative housekeeping and it is not. The exit has to be published because a reader holding an old view has no way of learning that it stopped being updated, and from outside a document, silence and continued agreement look exactly alike. Picture a shop with a board outside listing today's prices. Taking the board down tells everybody something. Leaving it up and not changing it also tells everybody something, and it is a lie, though nobody wrote it.

What does coverage oblige an analyst to publish, and when?

Three occasions, and they are not equally comfortable. After every periodic result, whether or not anything moved. Whenever an assumption inside the published view changes, promptly rather than at whatever moment is convenient. And at the moment of stopping.

Notice that every result occasion ends in a publication, and there are only three shapes it can take: the view changed, the view did not change, or the view is being discontinued. There is no fourth branch labelled nothing this time. The obligation that gets quietly skipped is the middle one, and skipping it is precisely what turns a published view into a rumour about a published view. Nobody skips a change; a change is interesting and it makes the analyst look alert. The obligation quietly skipped is the note saying the quarter arrived, the assumptions were checked against it, and none of them moved. Writing that note feels like admitting there was nothing to add. The unchanged note is the most valuable document in the file and the one nobody wants to write.

How much work does one covered name actually create in a year?

Count it rather than feel it. Each name produces four periodic results and one annual report, five mandatory occasions in all. Four names is therefore sixteen result occasions plus four annual reports, twenty publications a year, and not one of them depends on anything interesting happening. An event driven note, a change of management, a large capital spend, a shift in the field, all of that sits on top.

Twenty across fifty two weeks is one every 2.6 weeks, and that sounds like a gentle rhythm. The real rhythm is nothing like it. Every issuer in one field reports within days of the others, so result occasions do not spread themselves through the year but arrive in bunches. Capacity is therefore set by the worst weeks rather than by the annual total. The bunching is the single most useful fact about a coverage load, and the multiplication actively hides it.

Now the number an analyst actually negotiates over. Suppose the quietest possible publication, the one that says the result arrived and nothing moved, costs one working day, and suppose a working year holds about 250 days. Neither figure is in the record and both are stated openly so that they can be argued with. On that footing four names cost twenty days, or 8.0 per cent of the year, and that is comfortable. But the year is not what binds. Inside one three week window there are fifteen working days, four names need four of them, and there is room to spare. A list of fifteen names fills that window exactly, with no day left for reading anything, for building a changed view, or for the company that reports late.

The load arithmetic is one multiplication. Its one genuinely interesting feature, that the occasions arrive in four dense bunches rather than spreading out, is a shape rather than a relationship, and a shape is easier to see across a whole year at once than to read off a total.

The worked occasion: one quarter, one name, one honest output

Meghna Iyer covers the four names above. Three are makers of paints and coatings and the fourth, Thottam Chemicals Limited, is a supplier of resins and additives sitting one step up the chain. Sarvani Coatings Limited publishes its fourth quarter, and this is the occasion worked in full.

Revenue for the quarter is Rs 580 crore against Rs 505 crore in the same quarter a year earlier, growth of 14.85 per cent over one year. The full year came in at Rs 2,415 crore against Rs 2,120 crore, growth of 13.92 per cent, also over one year. So the quarter ran just under one percentage point ahead of the year, and the obvious question is whether that means anything.

The gap does not mean anything, and the way to see that is to put the quarter back among its siblings rather than against the average of all four. Recomputed from the published quarterly revenue, the four quarters of the year grew 13.46, 12.37, 14.75 and 14.85 per cent against the corresponding quarters a year before. The spread from the slowest to the fastest is 2.48 points. A quarter sitting just under a point away from the annual rate is well inside a spread of 2.48 points. A gap that size is ordinary seasonalityThe habit of a business to earn unevenly across the year for reasons that repeat, such as a festive quarter or a monsoon one. It is a pattern rather than news. rather than a signal. The third quarter is the festive one and the second carries the monsoon, so quarters in this business are not interchangeable units and never were.

One measure agreeing with a view is not a check, so a second measure is checked before concluding. The quarter's margin on earnings before interest, tax, depreciation and amortisation (EBITDA) is 18.45 per cent, against a blended 18.47 per cent for the whole year, so the quarter came in 0.02 of a point below the year it belongs to. Against the same quarter a year earlier, at 16.44 per cent, the margin is up 2.01 points over one year. The expansion is the one the annual figures already published, not new information arriving late.

Now the part that a careless note would skip. Meghna Iyer's published view rests on a thesis with exactly three variables: whether gross marginRevenue less the cost of the materials that went into the product, expressed as a share of revenue. It arrives here from the accounting layer and is used rather than rebuilt. holds at the 46.0 per cent the year reached, whether volume growth stays above the sector rate, and whether the small share gain repeats. The quarterly release in this record carries revenue, EBITDA and profit after tax and carries no quarterly cost of materials at all, so the first of those three assumptions cannot be rechecked at a quarter, and the honest publication says so rather than pretending the quarter tested everything.

So the correct output is a short note that states the quarter arrived, states which assumptions were checked and against what, states which one the release could not touch and why, records that nothing in the view moved, and stops. The note changes no reader's mind, and it is a finished job. The same note is also what makes the whole record usable a year later, and the analyst who skips it has broken coverage without a single person noticing on the day.

What does reading the same name for years actually leave behind?

Here is the justification for the whole arrangement, and it is not the one people expect. A company read once produces an opinion. A company read every quarter for three years produces something else entirely: a dated trail of what the analyst believed, when they believed it, and what arrived immediately afterwards.

Count that trail for one name. Three years of coverage is twelve quarterly publications plus three annual ones, fifteen dated entries, each one a small sworn statement of the form on this date I believed the following and here is what I checked it against. Across four names it is sixty. The dated trail is the only thing that makes a later examination of a past call possible at all, and an analyst who covers a name without keeping it has thrown away the by-product that was worth more than the publications.

Why worth more? Because without it, examining an old call collapses into arguing about the outcome. Did the share go up or down, and was the analyst therefore clever or foolish. With the trail, a different question becomes answerable: what did this person assume, was that assumption reasonable on the evidence available then, and which specific one failed. Such a question is about the work rather than about the result, and it is the only kind of question whose answer improves anybody. How such an examination is actually run is worked separately, under the review of past calls.

Where does coverage push the analyst, and what holds the line?

The pressure is real, and pretending otherwise is how people walk into it. An analyst who must publish every quarter is under a standing demand to have something new, and the supply of genuinely new things about one company is smaller than the publication calendar. Twenty occasions a year on four names, and perhaps three of them carry a real change. So what fills the other seventeen?

Three things fill them, and all three are failure modes. A view drifts, a little each quarter. A moving view is more interesting to read than a stationary one. Small figures acquire large adjectives, and 0.02 of a point becomes a margin holding firm under pressure. And the analyst begins writing about the quarter rather than about the business. The third habit is the most seductive of the three and it always produces copy. There is always something to say about a quarter. There is often nothing new to say about a company.

The discipline that answers all three is the same sentence: the honest output most quarters is that nothing in the view changed, and saying so in one short note, with the checks named, is a finished job. An analyst who can write that note without embarrassment can carry a list. An analyst who cannot will pad, and the padding will eventually be indistinguishable from a view.

What is coverage not?

Three inferences get drawn from a coverage list that the list does not support, and it is worth closing each one off by name.

Coverage is not a ratingThe single word label a research house attaches to a name, such as buy or hold. A rating states what the view is, where coverage states only that a view is being maintained., and it does not imply one. Coverage says a view is being maintained, not what that view is. Coverage is not an endorsement of the issuer: an analyst may maintain a current view on a company they consider poorly run, and maintaining it is the job. And coverage is not a relationship with the company. Coverage creates no entitlement to management's time, no access, and no obligation on the company's part to speak to anybody. Readers routinely infer all three from the bare fact of a name appearing on a list, and every one of those inferences is supplied by the reader rather than claimed by the analyst.