A Rating Watchlist Entry: What a Review Under Way Says

A rating watchlist entry announces that a rating agency has begun reconsidering an assessment it already published, and that it has not finished. The statement is about the review, not about its result. On the morning the entry appears the assessment reads exactly what it read before, and the arithmetic behind a credit spread contains no input that an announcement could enter.

What does a rating watchlist entry actually announce?

The object itself comes first, before any account of what it means. A rating watchlistA published list of assessments that a rating agency has begun reconsidering, where no conclusion has yet been reached on any of them. is a published list. On it sit assessments that the agency which produced them has started looking at again. A published list is the whole of the object. The list is not a ranking, it is not a queue in any order of severity, and it carries no verdict on anything.

So what does an appearance on it say? A watchlist entryThe announcement that one particular published assessment has been placed under reconsideration by the agency that issued it. says two things and no third thing. The entry says a named agency has begun reconsidering a named assessment, and it says that reconsideration has not concluded. An entry is a statement about the status of a review, and it makes no claim of its own about the borrower. An entry is not a new assessment. An entry forecasts nothing about what the review will find, and it measures nothing at all.

The everyday version does the work of three paragraphs, so it is worth holding on to. A shop puts a card in the window reading prices under re-check. A passer-by who reads that card knows the prices are being re-checked. The card does not say that anything will cost more tomorrow, it does not say that anything will cost less, and it does not say that anything will change at all. The card has reported the shopkeeper's activity, and it has reported nothing about the price of rice.

Most readers accept that about the shop and then quietly refuse it about a rating agency, on the reasonable ground that agencies are careful and do not begin reviews idly. The ground is reasonable and the conclusion drawn from it is still wrong, and the distance between those two sentences is where the whole difficulty sits.

What has changed the morning the entry appears, and what has not?

Splitting the moment in two settles most of the question. Something did change when the entry appeared, and something conspicuously did not, and a reader who cannot say which is which will read every entry they ever meet in the wrong register.

Take what changed first. The agency has said publicly that it is looking again, and where it publishes such things it has usually also said what prompted the second look. The announcement is real information and it arrived today. Somebody who read the assessment yesterday and somebody who reads it today are not in the same position, and pretending otherwise would be silly.

Now what did not change. The assessment reads whatever it read the day before. Nobody has revised it, nobody has withdrawn it, and the agency has not put a different one in its place. If the whole point of an assessment is that it is a published judgement, then the published judgement standing this morning is the same published judgement that stood last night.

An entry is new information and it is not a new assessment, and treating the first as the second is the single failure worth guarding against. The two words look close enough that the distance between them collapses under any pressure of time. Information about a process and a judgement about a borrower are different objects, produced by different work, and capable of being wrong in different ways.

Think of a household waiting on a school admission. The school writes to say the file has been re-opened and is being looked at again. A parent who reads that letter as a rejection has invented a result. A parent who reads it as an acceptance has invented a different result. The letter carries neither. The letter carries the fact that somebody is reading, and the honest response to it is attention rather than conclusion.

What distinguishes a review under way from a change already made?

In most cases nothing about the borrower is known, so the test has to run without any knowledge of one. One such test is short enough to carry around: ask what the statement would have to get wrong in order to be false.

Run it on a change already made. A rating actionA change actually made to a published assessment, as distinct from an announcement that a change is being considered. asserts something about a borrower. If the agency has revised its judgement, that revised judgement can be mistaken about the borrower, and the mistake shows up eventually in what the borrower does. The statement has put itself at risk against the world.

Now run the same test on a review under wayA reconsideration that has started and not finished. Throughout it the assessment that was already published continues to stand.. Ask again what would have to be untrue for the announcement to be false. Only that the review is not actually happening. The announcement has not put itself at risk against the borrower at all. The announcement has put itself at risk against the agency's own conduct instead. Such a risk is much smaller and much duller, and the smallness of it is exactly the point.

An announcement and an assessment therefore belong in different categories rather than at two points on one scale. An entry is not a weak assessment or an early assessment. An entry is not an assessment at all. A weather forecast and a note saying the forecaster has come into the office are both about weather in some loose sense, and only one of them can be wrong about the rain.

Why does an agency publish that it is looking rather than waiting until it has looked?

The question looks like one about the borrower and it is not. The question is about the cost of silence, and the answer has nothing to do with any particular company.

Picture the alternative. An agency decides an assessment needs a second look, says nothing, and works quietly until it has an answer. For the whole length of that work a published assessment stands in front of readers with the agency's name attached. The agency itself is no longer fully behind it. Every person who reads that assessment during the review reads something its own author has already begun to doubt, and none of them can tell.

Publishing the review closes that gap. A reader who would otherwise be handed a figure is instead handed a figure plus the warning that it is being re-examined, and the second thing is strictly more honest than the first. The entry is not the agency saying it was wrong. The entry is the agency making public where it currently stands, and that place is the beginning of the work rather than the end of it.

When an assessment gets reviewed, how a review must be announced, and what an agency has to publish about its own method are all set by the rating agencies and by the Securities and Exchange Board of India (SEBI) at sebi.gov.in, and they are revised. Any of them written out from memory would not be merely out of date on the day of the revision; it would be wrong, and wrong in a way a reader could not detect. The rows where those items would sit therefore carry the route to the source instead.

A rating scale and the meaning attached to each step of it belong to the agencies and to SEBI at sebi.gov.in, so Palash Cements Limited, the only borrower named above, holds no credit rating. With no assessment there can be no entry against it. The worked artefact below is therefore an empty field rather than a filled one, and every sentence above holds without any rating being named.

What does an entry do to the arithmetic of a credit spread?

Nothing, and being told that is much weaker than watching it. So build the arithmetic in front of yourself and then look for the place an announcement could get in.

Palash Cements Limited pays 9.10 per cent a year on its invented five year bond, on Rs 1,000.00/- of face, annual compounding throughout. Over the same five years the government pays the five year government SPOT rateThe price of money handed over now and returned at one named date in the future, one rate for each date. of 6.90 per cent a year on the invented curve used throughout the working. Subtracting one from the other gives the difference.

The difference is the credit spreadA borrower's yield less the government SPOT rate for the same maturity, quoted as a rate a year and always named against what it sits over., covered separately rather than here. The spread is repeated because the next step needs it, and because a figure watched being made is harder to believe false things about.

The loss given defaultThe share of the amount owed that is not recovered after a borrower stops paying. It is one hundred per cent less the recovery rate, on the same base. of 0.60 is not a measured figure. The 0.60 follows from an assumption of 40 per cent recovery that no study supports, and the assumption is named again in the limits section below rather than left behind here.

Carry the four decimals inside that multiplication. The division of 2.20 by 0.60 is exactly three and two thirds, so the printed 3.67 per cent a year is a reading rather than the figure itself, and multiplying the reading back gives 2.2020 percentage points instead of 2.2000. The gap is 0.0020 percentage points, or 0.20 basis points. The gap is small and it is not nothing, and a reader who lands there deserves to have been told in advance which rounding produced which answer.

Now go looking for the announcement. Four quantities were used: a yield of 9.10 per cent a year, a five year government SPOT rate of 6.90 per cent a year, an assumed recovery of 40 per cent of the amount owed, and the arithmetic joining them. Not one of the four is a rating, an outlook or a review, so there is no slot anywhere in the calculation into which a watchlist entry could be placed. The impliedSolved backwards out of a price under a stated assumption, rather than counted from experience or forecast from a model. figure is computed from a price and an assumption. Until a price moves, the arithmetic has nothing new to read.

What may a reader honestly take from an entry?

Less than the announcement seems to offer, and the short list is worth memorising because it is the part people skip.

A reader may take that a named agency has begun reconsidering a named assessment and has not finished. A review is work with an end, so a reader may also take that a conclusion is expected rather than open ended. And where the agency publishes the direction of its own looking, that direction is a statement about where the agency is pointing its attention, and a reader may take it as exactly that.

Now the second column. A review that finds nothing is still a review that concluded, so a reader may not take a direction as an outcome. A probability is out of reach. No default study and no transition study underlies the arithmetic, and nobody has counted how often a review of any kind ended in any particular way. No series of past spreads underlies the arithmetic either, and nobody has recorded what any price did on any announcement, so the price of the bond is out of reach as well.

The third refusal is worth dwelling on. A reader most wants to walk around that one. A sentence saying spreads typically widen on such news would be easy to write, and no evidence supports it; a sentence that sounds like a finding while resting on nothing is worse than an admitted gap.

How does an entry come off the list?

A list that people join and never leave would carry no information, so the exit matters as much as the entrance. The exit is simple: the review concludes, and the entry comes off.

There are two ways for that to happen and they are not two grades of the same thing. The review may reach a different conclusion, in which case the assessment changes and the entry comes off. Or the review may reach the same conclusion it started with, in which case the assessment stays exactly where it was and the entry comes off anyway. Both are endings. Both are the review doing precisely what it said it would do.

An entry that ends with no change is not a false alarm and it is not the review failing. The review concluded, and a concluded review is the only thing the entry ever promised would happen. Read the promise again if that feels wrong. The entry said a reconsideration had started and had not finished. Both halves came true. Nothing in it ever said the answer would be different.

The everyday version once more. A doctor orders a test. The test comes back showing nothing wrong. The test was ordered to find out rather than to confirm, so nobody sensible calls it wasted or accuses the doctor of a false alarm. A review works the same way, and a reader who only counts the reviews that changed something has quietly redefined what a review is for.

What are the three limits that travel with the arithmetic?

All three are gathered here, so nothing depends on a reader having held each one in mind since the section it came from. All three attach to the 3.6667 per cent a year, and none of them is optional wherever that figure is printed.

First, the 40 per cent recovery is an assumption and no recovery study sits behind it. With the spread held perfectly still at 220 basis points and only the assumption moved, the answer moves with it.

One price, one spread, four answers, and the assumption is doing that much of the work. Notice too that a watchlist entry moved none of those rows either. The assumption is supplied by the reader and the spread is read off a price.

Second, the whole spread has been treated as compensation for credit. In a real market some part of what a borrower pays covers the difficulty of selling the bond again rather than the chance of not being paid, and every basis point of that counted as credit pushes the implied rate too high. Carve 0.40 percentage points out of the 2.20 and treat them as payment for something else, and 1.80 points of credit are left: 1.80 divided by 0.60 is 3.0000 per cent a year rather than 3.6667. Separating the two is beyond this working, and saying so is better than working around it.

Third, an implied default rate is what a price says. An implied rate is not a forecast and not a measured frequency of anything. Nobody counted a default to produce 3.6667 per cent a year; it was solved backwards out of one spread under one assumption. Reading it as the probability that Palash Cements Limited fails misreads the arithmetic that produced it, and an entry on a watchlist changes nothing about that either way.

So what is a watchlist entry actually worth to a reader?

Less than the drama around it suggests, and still enough to be worth reading. An entry hands the reader one true fact that arrived today: an agency that publishes judgements has said that it has begun re-examining one of them and has not finished. The fact is free, it is dated, and it points at a file worth opening.

An entry cannot say what the review will find, how likely any outcome is, or what any price will do. None of those things is in the announcement, and none of them can be got from the arithmetic either. The distance between what an entry says and what readers routinely take from it is the whole subject, and the honest measure of understanding is whether a reader can state the content of an entry in one short sentence without adding anything to it.