Investment Grade and High Yield: What a Divide Does

Investment grade and high yield are two groupings produced by a single mark drawn across a rating scale, and where that mark sits is set by the rating agencies and by the Securities and Exchange Board of India (SEBI). The mark converts a ranking into a permission that is only ever on or off, so a crossing obliges certain holders to act while every payment written into the bond stays exactly where it was.

What actually makes a bond investment grade or high yield?

Almost everybody meets these two names and hears them as descriptions of borrowers: one kind of company is sturdy and one kind is shaky, and somebody with more information has sorted them. The sorting story feels natural and it is the wrong shape. Neither name describes a borrower at all. Each one describes a position.

Here is the apparatus, and it has exactly two parts. The first part is an ordering: a rating scale with many steps, running from a placement that is stronger to a placement that is weaker, with every assessed borrower standing on one of those steps. The second part is a single horizontal line drawn across that ordering at one chosen height. An ordering and one line are the whole machine. Everything standing above the line is then called investment gradeThe grouping on one side of a line drawn across a rating scale. The grouping is defined by the position of the line, not by any property a borrower carries., and everything standing below it is called high yieldThe grouping on the other side of that same line. The name marks a position, not a promise about what any holder will earn.. The two groupings are made by the line, so the line, and not any borrower, is the thing to understand.

The arrangement is a familiar one, met several times over in ordinary life. A school prints a pass mark. Two students sit the same paper, one scores a single mark above the pass mark and the other a single mark below it, and from that morning onward the two of them are in different rooms: one holds a certificate and one does not, one may sit the next paper and one may not. Nobody looking at the two scores would say the students differ by much. Everybody looking at the two certificates would say they differ completely. Both statements are true, and the reason they are both true is that a mark was drawn.

The same shape runs through a height requirement on a fairground ride, a minimum income on a rental application, and a cut-off date for a school year. In every one of those cases the underlying quantity moves smoothly and the consequence does not move smoothly at all. A rating scale with a line across it is that arrangement applied to lending, and once it is seen as an arrangement rather than as a fact about companies, everything that follows falls out of it without much effort.

Why does the scale have to be read from the source rather than recalled?

Because the scale, its steps, the meaning attached to every step and the height at which the line is drawn are all set by the rating agencies and by SEBI at sebi.gov.in, and all of them are revised. A text that reproduces a scale from memory is not merely stale on the day it changes. A scale recalled rather than read is already wrong on the day the ink dries. So the divideA mark drawn across an ordering, which converts a continuous ranking into a permission that is either on or off. is drawn below as an unlabelled mark across an unlabelled ordering. What the mark does stays true whatever the steps are called. What any step is called has to be read at the source.

Palash Cements Limited, an invented issuer, holds no credit rating, and none is assigned to it below. Not one claim about the divide needs an assessment of Palash Cements Limited in order to be true. Where an assessment would sit, the row stays empty and carries the reason inside it.

How often a borrower on either side of the mark actually fails is a measured question, and answering it needs three things: a counted default frequency, a recovery study and a run of past spreads. The arithmetic below has none of the three. So the arithmetic shows a shape and puts figures on an invented case, and what a real price did after a real crossing stays a separate question with a separate answer.

Why draw a line across an ordering that already ranks everything?

The answer to that question is about writing rather than about credit. An ordering establishes that one borrower stands ahead of another, and that is genuinely useful information. But obedience needs a test, and a test needs an answer that two people reading it separately will agree on, so an ordering cannot be written into a sentence that somebody else has to obey.

Consider the difference between two instructions given to a person managing money for somebody else. The first reads: hold only reasonably strong borrowers. The second reads: hold only what sits above the mark. Reasonably strong is an opinion, and every person reading it will draw the boundary somewhere slightly different, so the first instruction is an invitation to an argument. The second is settled by looking. A line is drawn because a rule needs a switch, and an ordering on its own does not supply one.

The written instruction has a name in practice: the mandateThe written instruction a holder operates under, which sets out what it may hold and what it may not. A mandate may be worded against a credit assessment. the holder operates under. Some mandates are worded against a credit assessment and some are not, and how an assessment may be used inside a rule about who may hold what, along with which categories of holder may hold which categories of debt, is set by SEBI at sebi.gov.in and by the Reserve Bank of India at rbi.org.in. Both rule sets move, and both are named below with the regulator that sets them. The shape of the arrangement needs no rule set at all in order to be true.

Notice what the line has done to the ordering it was drawn across. Before the line, every step counted for the same thing: one place further along. After the line, one particular step between two adjacent positions is the only place where a sentence in a mandate begins to read differently, and that step counts for something no other step counts for. The scale did not change. The consequence of standing on one part of it did.

How can a very small movement change everything that is permitted?

Two objects are stacked on top of each other here and they behave in completely different ways. Underneath sits a quantity that flows: how strongly a borrower is placed relative to others. The placement can shift by a hair or by a mile, and it takes every value in between. On top sits a permission with exactly two positions and nothing between them. The permission is on, or the permission is off.

Put those two together and the consequence is immediate. A movement of almost nothing in the quantity underneath can flip the permission completely, provided it happens to straddle the mark. A movement of a great deal can flip nothing at all, provided it stays on one side. The size of the change in what is permitted therefore carries no information whatsoever about the size of the change underneath it. That sentence is the whole argument compressed, and it is worth reading twice, because almost every misreading further on comes from forgetting it.

The examination hall shows the same thing. The last mark that lifts a candidate past the pass mark is worth exactly one mark, and it is also worth the entire result. Neither of those sentences is an exaggeration and neither cancels the other. Both sentences are true because the certificate is a switch and the score is a flow, and no amount of staring at the certificate reveals whether the candidate scraped through or sailed through.

Of the symbols in the term table above, one carries the trouble. The trouble is not in the placement and not in the permission. The star on the mark carries it: that one figure is set elsewhere, it is revised, and every consequence traced above and below hangs from it.

Who has to act when something crosses, and what does having to act mean?

Some holders operate under a mandate worded against a credit assessment. When the thing they hold crosses the mark, those holders must act. The word must is doing real work here and is not a figure of speech, so read it carefully. Such a holder has no discretion about whether. There is usually a date by which the selling has to be done. The wording of such mandates is not a private matter, so the position such a holder stands in is no secret from the people on the other side of the transaction.

A holder in that position becomes a forced sellerA seller with no discretion about whether to sell, and usually a date by which the selling must be finished., and a forced seller is not a nervous version of an ordinary seller. A forced seller is a different object. A willing sellerA seller with discretion about whether to sell, about when to sell, and about the price at which it stops selling. holds three things the forced seller does not hold: discretion about whether to sell at all, discretion about when, and a price at which it simply stops and keeps the bond instead. Remove all three and what is left is a party that must complete a transaction inside a window, whatever the other side offers.

An everyday version, and it is uncomfortably close. A household that has decided to move house next year is one kind of seller: if the offers on the table are poor, it stays put and tries again later. A household whose posting begins on the first of next month is a different kind of seller entirely, and every buyer walking through the door can see the packed boxes. Same house, same street, same roof. The two households are not in the same negotiation.

A price produced by a seller who could not walk away is a different kind of evidence from a price produced by one who could, and nothing printed beside the price reveals which it was. That is not a criticism of anybody. A holder acting under a mandate is doing exactly what it undertook to do, and doing it on time. The point is narrower and stranger: the number that comes out of the transaction records the constraint as well as the borrower, and the two arrive fused together with no seam to pull apart.

What do the same unchanged payments fetch at a wider spread?

Enough shapes. A difference stated in adjectives can be nodded at, and a difference stated in rupees has to be reckoned with, so put rupees on the table. Palash Cements Limited pays Rs 91/- at the end of each of five years, and a further Rs 1,000.00/- of face amountThe amount a bond repays at maturity. The face amount is fixed by the contract and does not move for any price reason. at the end of the fifth. Six amounts in total. Not one of them is touched by anything in this guide, at any point, in any direction.

Every price below is struck on annual compounding, one discounting period a year, and that is stated inside the arithmetic rather than in a note beneath it. The convention is not housekeeping. The six amounts discount to Rs 1,000.000000/- exactly at 9.10 per cent a year, and that is what at par means. The identical coupon, maturity and yield on a semi-annual convention would produce a different price from figures that look the same in print. With the convention written beside the price, a reader can reproduce the sum. Without it, no amount of surrounding detail makes the sum reproducible.

The 320 basis points in that sentence is a suppositionA figure introduced on the face of the text in order to work an argument through, supported by nothing in the underlying record., carried across from the decomposition of a widened spread rather than invented afresh here, and labelled as a supposition every single time it appears. Nothing measured stands behind the 320 basis points, and no spread anywhere in this sequence ever moved. The supposition buys one thing: the arithmetic can be run, and the size of the answer seen in rupees.

Run it at the yield the bond was issued at and the six amounts come to Rs 1,000.000000/- exactly, par to the last digit, and par is where the whole sequence starts. Now run it again at the supposed yield of 10.10 per cent a year. The same Rs 91/-, the same Rs 91/-, the same Rs 91/-, the same Rs 91/-, the same Rs 91/-, the same Rs 1,000.00/-. The six amounts now come to Rs 962.188776/-, or Rs 962.19/- as it would be printed. The difference is Rs 37.811224/- on Rs 1,000.00/- of face amount.

Every rupee of that Rs 37.811224/- is a difference in the price somebody accepted, and none of it is a difference in what the bond has promised to pay. The contract is a sheet of paper with six dated amounts written on it, and no arithmetic performed on the far side of the market rubs out a single digit of it. The price is what somebody would hand over today to stand in line for those six amounts, and that is a separate object with a separate life.

What may honestly be read off a price reached like that?

Less than the price appears to offer, and the shortfall is worth being precise about. The number may be read as a price. Somebody transacted at that level, on that day, in that bond, and the number is a genuine record of that event. Nobody is disputing the transaction.

The number may not be read as a measurement of how likely the borrower is to fail. At least one of the parties who produced that number was acting under a constraint that has nothing to do with the borrower at all, and the number arrives as a single figure with no breakdown attached to it. There is no line in the record that separates the part contributed by an opinion about repayment from the part contributed by a date in a mandate. The two are welded together at the moment of the transaction, and no amount of subsequent staring will separate them.

The mechanism runs in both directions, and saying so is part of stating it honestly. Something can cross the mark the other way, and when it does the set of permitted holders enlarges rather than shrinks. A seller facing more permitted buyers stands in a different position from one facing fewer, and that is as far as the shape carries. Which way a price moves after a crossing is a separate question, and answering it would need a series of past spreads, a case in which some spread actually moved, and a count of the permitted holders before and after. No such record stands behind anything above, so no direction is claimed for any price.

One comparison a reader will reach for is simply unavailable here. No second issuer appears anywhere in this sequence, so no bond on one side of a divide can be set beside a bond on the other and the difference read off. The arithmetic instead discounts one unchanged set of payments at two different rates, a comparison of rates rather than a comparison of borrowers.

What are the three limits that travel with any of this?

Three limits ride along with every implied figure this sequence produces, and none of them is optional. Each of the three moves the implied figure, and none of the three is visible in the figure itself.

First, the 40 per cent recovery used to turn a spread into an implied default rate is an assumption, and no measurement stands behind it. Move it and the answer moves with it. Hold the 220 basis point spread perfectly still and the implied default rate walks from 3.1429 to 7.3333 per cent a year purely on the strength of a figure nobody measured.

Second, the whole spread has been treated as compensation for credit. In a real market some part of a spread pays for the difficulty of selling the bond when the holder wants to, and every basis point of that read as credit pushes the implied default rate too high. Carve 0.40 percentage points out of the 2.20 and call them payment for something else: 1.80 points of credit remain, and the implied rate falls from 3.6667 to 3.0000 per cent a year. Nothing in the price separates the two.

Third, an implied default rate is what a price says. An implied default rate is not a forecast and not a measured frequency of anything. Nobody counted a default to produce 3.6667 per cent a year; the figure was solved backwards out of one spread and one assumption. A reader who carries it away as the chance that Palash Cements Limited fails has misread the arithmetic that produced it, and the word implied was attached to every use of it precisely to stop that happening.

So what is a divide actually good for?

A divide is good for exactly one thing, and that one thing is real: it makes a written instruction testable. Everything else attributed to it is borrowed. A divide does not measure a borrower, does not rank two borrowers against each other, and does not say what will be paid. A divide converts a position on an ordering into an answer of yes or no, and it does that reliably enough that instructions can be written against it and checked afterwards by people who were not there.

From that one function everything else follows. Because the answer is yes or no, a tiny movement can flip it. Because the flip changes who may hold something, some holders must act. Because those holders must act, part of a price is produced by a constraint. And because a price arrives as one number, the constraint and the opinion cannot be told apart afterwards. Read a price as a price, ask who had to be present to make it, and keep the six unchanged amounts of the contract in view.